# Data Centre 254 (DC254) > Full-content edition of public/llms.txt: every DC254 article as clean markdown. An independent Kenyan educational and research platform explaining the physical infrastructure behind Kenya's digital economy - data centres, connectivity, power, AI, and the systems that make the digital world work. Published from Nairobi, Kenya. 118 articles follow, one "## " section each. The curated index of key pages lives at https://data-centers-254.vercel.app/llms.txt; per-article HTML at https://data-centers-254.vercel.app/articles/; RSS at https://data-centers-254.vercel.app/feed.xml. ## 5G Networks and Data Centre Demand in Kenya URL: https://data-centers-254.vercel.app/articles/5g-networks-data-centre-demand-kenya Cluster: Internet & Connectivity Published: 2026-08-28 Updated: 2026-08-28 Summary: How Kenya's 5G rollout by Safaricom and Airtel is driving demand for edge data centres, MEC nodes, and expanded colocation infrastructure. ## The 5G Revolution Arrives in Kenya Kenya's telecommunications landscape entered a new era when **Safaricom launched commercial 5G services** in October 2022, making the country one of the first in Sub-Saharan Africa to roll out the next-generation wireless technology. The initial deployment covered strategic locations in **Nairobi, Mombasa, and Kisumu**, with the telco investing heavily in new base station infrastructure to support the technology. Airtel Kenya has also signalled ambitious 5G plans, positioning itself as a competitor in the race to connect Kenya's growing digital economy. What many observers outside the infrastructure sector may not realise is that **5G is not just a network upgrade**, it is a fundamental shift in how computing resources are distributed. The ultra-low latency and massive bandwidth that 5G promises cannot be delivered by the network alone. Behind every 5G cell tower, there is an growing need for data processing capacity that sits as close to the end user as physically possible. This is creating a significant and largely unrecognised surge in demand for data centre infrastructure across Kenya. (New to 5G itself? Start with our plain-language guide: [what is 5G, from 1G to 6G, and why it matters in Kenya](/articles/what-is-5g-1g-to-6g-history).) ![Submarine cable coming ashore at a Kenyan beach landing](/images/mombasa-cable-landing.webp) ## The Communications Authority's 5G Spectrum Framework The **Communications Authority of Kenya (CA)** has been developing a spectrum allocation framework specifically designed to support 5G deployment across the country. The framework addresses the allocation of mmWave and mid-band spectrum, which are essential for 5G's high-speed, low-latency performance. According to the [Communications Authority of Kenya](https://www.ca.go.ke/5g-spectrum-allocation), the regulatory body is working to balance the interests of existing 4G operators with new market entrants, while ensuring that spectrum pricing does not become a barrier to deployment in underserved areas. The CA's approach has been compared favourably to other African regulators, with a phased allocation model that ties spectrum licences to specific rollout obligations. This regulatory clarity is crucial for the data centre industry because **spectrum allocation directly determines where 5G infrastructure will be built**, and consequently where edge computing capacity will be needed. Operators receiving spectrum in specific frequency bands will have technical requirements that dictate the density and placement of their base stations, and by extension, their MEC nodes. ## Why 5G Drives Data Centre Demand ### Multi-Access Edge Computing: The Core Driver The single biggest factor linking 5G to data centre demand is **Multi-Access Edge Computing (MEC)**. To understand why, consider the physics of 5G: the technology promises end-to-end latency as low as **1 millisecond**. However, light travels through fibre at approximately 200 kilometres per millisecond. If a user in Nairobi has to round-trip data to a centralised data centre 50 kilometres away, the network alone consumes 0.5ms, and that is before any processing time, protocol overhead, or queuing delays are factored in. **MEC solves this problem by placing compute resources within 10-20 km of users.** Each MEC node is functionally a small data centre, typically housing anywhere from **10 to 100 kW** of IT load. For a nationwide 5G deployment, this translates into hundreds of new facilities that need power, cooling, security, and connectivity, the core requirements of any data centre. This represents a paradigm shift for the [edge computing ecosystem in East Africa](/articles/edge-computing-east-africa). Traditionally, Kenya's data centre market has been concentrated in Nairobi and, to a lesser extent, Mombasa. 5G demands a distributed model where compute capacity follows population density and economic activity. ### Increased Bandwidth and Cloud Demand 5G networks deliver significantly more bandwidth than their 4G predecessors. **Safaricom's 5G network** has demonstrated speeds exceeding 700 Mbps in test environments, compared to typical 4G speeds of 15-30 Mbps. This dramatic increase in available bandwidth changes user behaviour: people and businesses consume more data-hungry services, including high-definition video streaming, cloud gaming, and real-time analytics. All of these services ultimately depend on backend infrastructure housed in data centres. As 5G drives higher data consumption, it creates proportional demand for cloud services, content delivery networks (CDNs), and the data centres that host them. The [GSMA's 5G in Sub-Saharan Africa report](https://www.gsma.com/r/sa-5g-report) projects that mobile data traffic in East Africa will grow by a factor of five between 2023 and 2030, with 5G being a primary catalyst. ### IoT Backend Processing Kenya's Internet of Things (IoT) ecosystem is expanding rapidly, driven by applications in agriculture, logistics, manufacturing, and utilities. 5G is designed to support **up to one million connected devices per square kilometre**, a massive increase over 4G's capabilities. Each of these devices generates data that needs to be collected, processed, stored, and analysed, all functions performed in data centres. In Kenya's agricultural sector alone, IoT sensors for soil monitoring, precision irrigation, and livestock tracking could generate terabytes of data daily at scale. This data needs local processing to be actionable in real time, further reinforcing the need for edge data centre infrastructure. ## Infrastructure Implications of 5G Edge Computing ### Technical Requirements for MEC Nodes ![Submarine cable landing supported by orange buoys in the surf](/images/mombasa-cable-landing-4.webp) Each MEC node in a 5G network has specific infrastructure requirements that distinguish it from both a traditional data centre and a simple cell tower: - **Power**: 10-100 kW per node, requiring reliable grid connections and battery or generator backup - **Cooling**: Precision cooling systems to maintain 18-27°C operating temperatures for IT equipment - **Fibre Backhaul**: Dedicated high-capacity [fibre optic networks](/articles/fibre-optic-networks-kenya-data-centres) connecting each node to core data centres and the wider internet - **Physical Space**: 50-500 square metres depending on capacity, typically housed in purpose-built enclosures or small buildings - **Proximity**: Located within 10-20 km of the users they serve to maintain sub-5ms latency - **Security**: Physical security systems including access control, CCTV, and environmental monitoring These requirements mean that **every MEC node is, by definition, a data centre**, albeit a small one. The capital expenditure for building a single 50 kW edge node in Kenya is estimated at KES 50-150 million, depending on location, existing infrastructure, and the level of redundancy required. ### The Economic Model: Who Builds Edge Nodes? The question of who finances and operates 5G edge data centres is one of the most significant strategic decisions facing Kenya's telecom and infrastructure sectors. Three models are emerging: **Telco-owned MEC**: Safaricom and Airtel Kenya can build and operate their own edge computing infrastructure at cell tower sites. This gives them full control over performance and service quality but requires massive capital investment and operational expertise in data centre management. **Colocation at the edge**: Established colocation providers like **Africa Data Centres**, **PAIX (Pan-African Internet Exchange)**, and local operators can extend their footprint to edge locations, offering shared infrastructure that multiple tenants (including telcos, cloud providers, and enterprises) can use. This model offers better economics through multi-tenancy. **Neutral host infrastructure**: Third-party companies build edge data centre facilities and lease capacity to multiple operators. This model, already common in mature 5G markets like the United States and South Korea, maximises infrastructure sharing and reduces redundant investment. In Kenya, the colocation and neutral host models are likely to gain traction because they align with the country's push for infrastructure efficiency and the **Communications Authority's open access** principles. ## 5G Use Cases Driving Edge Demand in Kenya ### Smart Agriculture ![Server racks in a Nairobi data centre hall](/images/dc-gpu-cluster-2.webp) Agriculture contributes approximately **22% of Kenya's GDP** and employs over 40% of the population. 5G-enabled IoT devices can transform this sector through precision agriculture: soil moisture sensors, drone-based crop monitoring, automated irrigation systems, and livestock health trackers. Each of these applications generates real-time data that requires local processing to deliver actionable insights to farmers. ### Remote Healthcare Kenya's healthcare system serves a population of over 50 million people, many in remote areas with limited medical infrastructure. 5G's low latency enables telemedicine applications including remote surgical consultations, real-time diagnostic imaging, and AI-assisted diagnosis. These applications require edge computing nodes in or near healthcare facilities to ensure reliable, low-latency connectivity. ### Smart City Applications in Nairobi **Nairobi County** has been pursuing smart city initiatives, including intelligent traffic management, public safety surveillance, waste management optimisation, and digital government services. 5G provides the connectivity layer, while edge data centres provide the compute layer. For example, real-time traffic analysis from hundreds of cameras across Nairobi requires significant processing power that must be located within the city to be effective. ### Autonomous Vehicles and Logistics While fully autonomous vehicles may still be years away in Kenya, the logistics sector is already adopting semi-autonomous systems including connected fleet management, automated warehouse operations, and drone delivery, particularly in areas with challenging road infrastructure. These systems depend on ultra-reliable, low-latency communication (URLLC), which is a core 5G capability that requires edge computing support. ## Projecting the Scale: How Many Edge Nodes Will Kenya Need? Estimating the number of MEC nodes required for Kenya's 5G rollout requires benchmarking against more mature markets and adjusting for local conditions: ![Labelled telecom tower components from antennas to BTS cabinet](/images/diagram-telecom-tower-components.webp) **South Korea**, one of the world's most advanced 5G markets with a population of 52 million, has deployed an estimated **1,500-2,000 edge computing nodes** to support its nationwide 5G network. The **United States**, with its vast geography, has seen major operators like Verizon and AT&T deploy edge nodes in over 100 cities, with thousands of nodes planned. For Kenya, a reasonable projection accounts for: - Population of approximately 55 million - High urbanisation rate (28% living in urban areas, but growing) - Initial 5G focus on major cities (Nairobi, Mombasa, Kisumu, Nakuru, Eldoret) - Rural 5G deployment expected from 2027 onwards Based on these factors, Kenya will likely need **between 500 and 2,000 MEC edge nodes** for meaningful nationwide 5G coverage. Even the lower end of this range represents a **massive expansion** of Kenya's data centre footprint, which currently consists of approximately 20-30 significant facilities. ## The Interplay Between 5G Edge and Centralised Data Centres It is important to understand that **edge data centres do not replace centralised facilities**, they complement them. The architecture that 5G demands is hierarchical: 1. **MEC edge nodes** (10-100 kW): Handle real-time, latency-sensitive processing within 10-20 km of users 2. **Regional data centres** (1-10 MW): Aggregate and process data from multiple edge nodes, host larger applications and databases 3. **Core data centres** (10+ MW): Provide the primary compute, storage, and cloud services that underpin the entire ecosystem Nairobi will remain the natural location for Kenya's core data centres, given the city's role as the country's economic and telecommunications hub. However, the growth of edge nodes creates a new tier of demand that extends far beyond the capital. This is why the [Kenya data centre market outlook](/articles/kenya-data-centre-market-outlook-2025-2030) projects significant growth in both centralised and edge facility capacity over the next five years. ## Revenue Opportunities for Data Centre Operators The 5G-driven expansion of Kenya's data centre infrastructure creates multiple revenue streams for operators: - **Edge colocation**: Leasing space, power, and cooling in MEC nodes to telcos, cloud providers, and enterprises - **Managed edge services**: Operating edge computing infrastructure on behalf of customers who lack the expertise - **Interconnection**: Providing fibre connectivity between edge nodes, regional facilities, and core data centres - **Hybrid cloud services**: Enabling seamless workloads distribution between edge and core facilities For existing data centre operators in Nairobi, the 5G opportunity represents both a threat and an opportunity. Those who extend their footprint to the edge will capture new revenue; those who remain centralised risk losing customers who require distributed infrastructure. ## Conclusion Kenya's 5G rollout is not merely a telecommunications upgrade, it is a catalyst for a fundamental transformation of the country's data centre landscape. The demands of Multi-Access Edge Computing, driven by ultra-low latency requirements, will necessitate the construction of hundreds of new edge data centre facilities across the country. Combined with increased bandwidth consumption and the explosion of IoT devices, 5G is poised to make Kenya one of Africa's most dynamic data centre markets. For infrastructure investors, colocation operators, and technology companies, the message is clear: **the 5G edge opportunity in Kenya is real, quantifiable, and approaching fast.** Those who position themselves now (whether through strategic partnerships, infrastructure investment, or regulatory engagement) will be best placed to capture the value that 5G and edge computing will create. ## Frequently Asked Questions ### How does 5G increase data centre demand in Kenya? 5G's ultra-low latency (as low as 1ms) and massive bandwidth require compute resources to be placed close to end users through Multi-Access Edge Computing (MEC) nodes. Each MEC node is essentially a small data centre consuming 10-100 kW, creating hundreds of new facilities needed nationwide. Additionally, 5G drives cloud adoption, IoT backend processing, and content delivery demand, all of which require expanded centralised data centre capacity. ### Where has Safaricom launched 5G in Kenya so far? Safaricom launched commercial 5G services in October 2022, initially covering parts of Nairobi (Westlands, CBD, Kilimani, and surrounding areas), Mombasa, and Kisumu. The rollout has since expanded to additional counties including Nakuru, Uasin Gishu, and Kakamega, with plans for nationwide coverage as spectrum allocation and infrastructure investment continue. ### What is Multi-Access Edge Computing (MEC) and why does it matter for data centres? MEC places compute, storage, and networking resources at the edge of the network, typically within 10-20 km of end users. For 5G to deliver its promised 1ms latency, processing cannot happen in distant centralised data centres. MEC nodes are essentially micro data centres co-located with or near 5G base stations, each requiring 10-100 kW of power, dedicated cooling, and fibre backhaul connectivity. ### Who will build and operate 5G edge data centres in Kenya? Three models are emerging: telcos like Safaricom and Airtel building their own MEC infrastructure at cell tower sites; colocation providers like Africa Data Centres and PAIX extending to edge facilities; and neutral host companies that build shared edge infrastructure leased to multiple operators. The colocation and neutral host models are likely to dominate as they offer economies of scale and multi-tenant flexibility. ### How many edge nodes will Kenya need for nationwide 5G coverage? Based on Safaricom's existing base station footprint and projected 5G densification, estimates suggest Kenya will need between 500 and 2,000 MEC edge nodes for meaningful nationwide coverage. This projection draws from benchmarks in South Korea, which deployed approximately 1,500 edge nodes for a comparable population density, adjusted for Kenya's urban-rural distribution and lower initial 5G penetration. ## Why Africa Only Hosts 5.5% of the World's Data Centres URL: https://data-centers-254.vercel.app/articles/africa-160-data-centres-imf-power-constraint Cluster: AI & Infrastructure Published: 2026-09-24 Updated: 2026-09-24 Summary: The IMF counts about 160 data centres in Africa: 5.5% of the global total. We read the report: electricity is the constraint and Kenya is the test case. The International Monetary Fund has published one of the most quotable infrastructure statistics of the year, and it is travelling fast through African policy circles: Africa hosts only about 160 data centres, or about 5.5 percent of global installations. The line comes from "Unlocking the Potential: AI in Sub-Saharan Africa", a 2026 IMF departmental paper (Departmental Paper 2026/013) that we have read in full. It lands with particular force in East Africa, where nearly every serious data centre conversation now starts with power prices and ends with power reliability. But the number doing the rounds on social media and in op-eds is doing less work than the report behind it. The paper's real contribution is not the count: it is the diagnosis of *why* the count is what it is, and a carefully argued thesis that the constraint is not demand, not capital in the abstract, and not African appetite for cloud services. It is electricity: its reliability, its price structure, and whether the region's grids can treat a data centre as a customer worth building for. That reframing matters for investors, operators and policymakers across Kenya, Uganda, Rwanda and Tanzania, because it moves the conversation from "when will Africa get its hyperscale moment" to "what has to be true about the grid first". ## Where the IMF's 160 comes from The full sentence, from the paper's section on AI infrastructure, reads: "compute capacity remains modest relative to global demand for AI infrastructure, with Africa hosting only about 160 data centers or about 5.5 percent of global installations (Kakindé 2025)". The figure is drawn from market research the IMF cites rather than an IMF census. The paper immediately adds the geography that matters: "nearly half of Sub-Saharan Africa's data centers located in South Africa, Nigeria, and Kenya", citing two additional research sources for that concentration. Two cautions belong next to any use of the number. First, installations are not capacity. African facilities are on the whole smaller than those in mature markets, which is why industry bodies have long put Africa's share of global multi-tenant capacity below 1% even while installation counts suggest 5.5%. Both statements are true, and the gap between them is itself a finding: the region's fleet is broad but shallow. Second, any count of "data centres" is sensitive to definition: does a single-room enterprise server facility count alongside a 30MW campus? This is the same measurement swamp we documented in [why Kenya's data centre market estimates disagree](/articles/why-kenya-data-centre-market-estimates-disagree): six research firms produced market values from $266 million to $606 million for the same country, largely on definitional differences. The IMF's 160 deserves respect as an order of magnitude, not a census. ## The constraint is electricity, not demand ![Labelled diagram of the power chain from grid to GPU rack](/images/diagram-power-chain-grid-to-gpu.webp) The demand side, the IMF concedes, is not the problem. Global technology firms are expanding cloud and AI infrastructure across the continent, and the paper catalogues the commitments: Microsoft and G42's announced $1 billion green data centre campus in Kenya, Cassava Technologies and NVIDIA's $700 million deal to deploy 12,000 GPUs across South Africa, Nigeria, Kenya, Egypt and Morocco, a $100 million IFC commitment to Raxio's expansion, plus over $300 million from Huawei and a $2 billion Tencent cloud investment. Capital is circling. But venture funding for African AI ventures still declined to about $2.2 billion in 2024, with 84% of flows going to just four markets (Egypt, Kenya, Nigeria and South Africa), which tells you how thin the margin of investability is. ![High-voltage power transformer bay at a substation](/images/hv-power-transformer-bay.webp) The supply side is where the paper is unsparing. Around half the region's population does not have reliable power; of the roughly 670 million people worldwide without electricity access, 85% live in Sub-Saharan Africa. Seventy-eight percent of firms in the region report routine outages, with average losses of 8.4% of annual sales against a 5.2% global average. Generator ownership (the market's own verdict on grid reliability) is widespread precisely where the data centres are: 86% of Nigerian firms own or share a generator, and 65% and 63% do so in Kenya and South Africa respectively. The arithmetic gets sharper as compute scales. Data centres already drew about 1.5% of global electricity in 2024, headed towards roughly 3% by 2030 on the IEA's base case. Even a modest scenario in which Sub-Saharan Africa holds just 0.5% of global AI data centre capacity by 2035 implies additional demand equal to about 10% of the region's entire 2023 installed generation capacity. The growth stakes of fixing this are the report's headline. Under current conditions, the IMF estimates AI adds just 0.2% to the region's productivity cumulatively over the next decade, about 0.4% of GDP once AI-related capital accumulation is counted. Under a high-adoption scenario, with power, connectivity, skills and agricultural-sector coverage improved, productivity gains rise to roughly 2.1% and the GDP effect to 4.0% over the decade. The paper is explicit that this upside "should not be read as an automatic upside case": it is policy-contingent, start to finish. ## Anchor tenants: the IMF's most useful idea for Kenya ![Olkaria geothermal power plant in Kenya's Rift Valley](/images/kenya-geothermal-plant-olkaria.webp) The paper's most constructive passage is easy to miss. Data centres, it argues, need not be a burden on weak grids: their "massive, predictable, and long-term energy requirements" can function as an industrial anchor tenant whose demand makes large-scale generation and grid expansion bankable for private investors, potentially extending electricity access to surrounding communities as a spillover. The catch is structure: whether anchor demand "catalyzes net additions to grid capacity or simply diverts existing supply" depends on how power purchase agreements and grid extension mandates are structured. Kenya is the natural laboratory, and the paper treats it as such. With over 90% of electricity generation from renewable sources, Kenya is positioning its geothermal endowment as a competitive advantage for green compute, the logic behind [the Olkaria geothermal corridor](/articles/olkaria-geothermal-data-centre-corridor) and the broader case for [geothermal energy in Kenyan data centres](/articles/geothermal-energy-kenya-data-centres). The IMF notes the announced Microsoft/G42 Naivasha campus was designed to draw power directly from geothermal generation. Here our own tracking sharpens the IMF's point. The IMF's evidence base cites that campus as a live $1 billion, 100MW commitment. As we documented, Kenya formally suspended the plan in May 2026 after concerns that the project would need about a third of the country's roughly 3,000MW installed capacity, and after Treasury never approved the funding concept note. [The full story of the Microsoft and G42 stall](/articles/microsoft-g42-kenya-data-centre) is a study in structure failing before physics does. The suspension is not a rebuttal of the anchor-tenant thesis; it is the thesis, stated negatively. The resource was never the binding constraint. The power purchase structure, the financing guarantees and the public-approval pathway were. The same lesson travels north: [Ethiopia's plan to export 400MW to Kenya](/articles/ethiopia-power-exports-kenya-data-centres) only pays off for compute if the trading and contracting structures move with the megawatts. Regionally, the G7-endorsed Africa Green Compute Coalition, linking UNDP, private sector and African governments, is the first institutional attempt to operationalise the anchor-tenant link between renewable developers and digital infrastructure investors. ## What it means for East Africa The IMF is blunt about distribution. Even under the high-adoption scenario, near-term gains accrue first to "early-mover economies and urban hubs with stronger electricity, connectivity, and compute capacity", specifically the areas where data centres and cloud infrastructure are already concentrated: South Africa, Nigeria and Kenya. Economies with weaker infrastructure risk becoming "increasingly dependent on externally hosted services", importing AI capability rather than hosting it. For Uganda, Rwanda and Tanzania, that is the strategic warning: the gap between hosting compute and renting it compounds. The region's counter-move is not a single headline project but the unglamorous accumulation of the things the IMF lists as complements: grid reliability, high-capacity connectivity, and rules that make long-term power contracts enforceable. Kenya's grid reliability record and its tax incentive framework for data centre investment are exactly the kind of policy levers the paper's scenarios turn on, and they are precisely the instruments we audit claim by claim in our [East Africa regulation comparison](/articles/africa-data-centre-regulation-compared) and in the [DC254 Policy Intelligence console](/policy/intelligence), where 52 of 59 policy claims across Kenya, Uganda, Rwanda and Tanzania are now verified against captured primary sources. The physical build-out is tracked the same way: our [data centre directory](/directory) follows 31 facilities and 18 operators across the four markets, so the region's response to the IMF's diagnosis can be measured, not guessed. ## The number to watch The IMF's framing deserves the last word, because it converts a discouraging statistic into a working agenda: "The lower estimate is not a verdict on AI's potential but a diagnostic of today's constraints." There is even historical precedent for what happens when the constraints fall: the submarine cable waves that connected Africa's coasts boosted employment between 4% and 10% in connected areas, a reminder that infrastructure and outcomes move together when the underlying economics are fixed. So watch the boring numbers, not the headline ones. Watch the megawatts that reach financial close with a data centre anchor attached, the power purchase agreements that fund new generation rather than re-cutting old supply, and the grid extension mandates that decide whether communities around a campus get power or just the campus does. The day an East African data centre is the reason a new geothermal unit was built, not the reason an existing one was fully booked, the 160 starts moving, and this region will be able to say it fixed the constraint rather than rented around it. ## Africa has 3% of EMEA capacity. Where does Kenya sit? URL: https://data-centers-254.vercel.app/articles/africa-3-percent-emea-data-centre-capacity Cluster: Market Analysis Published: 2026-10-07 Updated: 2026-10-07 Summary: Africa holds 3% of EMEA's 93 GW of data-centre capacity. Where does Kenya sit? The verified numbers behind Kenya's power, cables and 230 MW pipeline. A new market analysis has put Africa's data-centre opportunity into one uncomfortable number: 3 percent. DC Byte's latest EMEA analysis, reported by ITWeb, estimates the region's total data-centre capacity at 93 GW once live facilities, construction, committed projects and early-stage developments are counted together. Europe accounts for 88 percent of that total. The Middle East accounts for 9 percent. Africa accounts for just 3 percent. That is a small share of a large market. It is also the reason the continent keeps attracting attention. The question for Kenya is not whether Africa has room to grow. It clearly does. The more useful question is whether Kenya can turn its advantages, from renewable generation and subsea cable access to established interconnection and a growing digital economy, into capacity that is actually powered, connected and operational. > **What the 3 percent means.** The figure is Africa's share of EMEA data-centre capacity, not Africa's share of global capacity, and not a forecast that Africa automatically captures the next wave. It is a regional comparison across a dataset that spans live capacity and multiple development stages. That distinction is kept unavoidable throughout this article. ![Donut chart of EMEA's 93 GW data-centre capacity split: Europe 88 percent, Middle East 9 percent, Africa 3 percent](/images/diagram-emea-capacity-split.webp) ## The 3 percent is an opportunity, not a forecast The first point needs to be stated precisely, because the number is already travelling without its context. The 3 percent figure is Africa's share of EMEA data-centre capacity, measured across a broad DC Byte dataset that includes operational capacity and multiple stages of development. It should not be read as a verdict on the continent or as a promise about its future. The number does, however, show how unevenly digital infrastructure is distributed. Europe's established hubs still dominate. Frankfurt, London, Amsterdam, Paris and Dublin, often grouped as FLAP-D, account for almost a quarter of EMEA's total capacity. Those markets are now running into the constraints that decide where new capacity can actually be delivered: available land, grid access, planning timelines and the ability to serve large power-intensive campuses. That is why DC Byte's analysis points to South Africa as the African market most clearly positioned for the next wave. Land availability is one reason. Power delivery remains the condition. The analysis is explicit that a site is not attractive simply because land is available: operators still have to secure sufficient power, and assess how quickly that power can be delivered, before any site becomes a data centre. That distinction matters for every African market, including Kenya. ## Kenya's counter-pitch is not one number Kenya's current proposition rests on several assets that only work when they are counted separately. The country has a verified data-centre directory of 27 Kenyan facilities, of which 20 are operational. [The State of Kenyan Data Centres 2026](/research/state-of-kenyan-data-centres-2026) separates 10.5 MW of published live IT load from 42.9 MW of built capacity at operational sites. Behind that operating base sits a tracked pipeline of 230 MW. Those numbers must remain separated, because they measure different things. ![Kenya capacity staging card: 10.5 MW live IT load, 42.9 MW built capacity, 230 MW tracked pipeline, 93 percent renewable generation, 7 live subsea cables](/images/diagram-kenya-capacity-staging.webp) The 230 MW pipeline is not operating capacity. It combines 77 MW under construction, 53 MW committed and 100 MW at an early stage. Projects at different stages face different tests. Under-construction projects are physically underway. Committed projects have land and funding signals but are not yet built. Early-stage projects may still require approvals, financing, firm tenants or power-delivery commitments. This is exactly the discipline African markets need as the global conversation shifts toward AI and hyperscale demand. A pipeline is evidence of intent. It is not evidence that the grid connection, the building, the cooling plant and the network are already serving customers. ## Kenya's power advantage is real, but it has a ceiling Kenya's strongest counter-pitch is the quality of its generation mix. About 93 percent of the electricity generated on the national grid comes from geothermal, hydro, wind and solar, according to KenInvest commentary published in The Star. Geothermal alone supplies roughly 45 percent of generation and gives Kenya something solar-only propositions cannot provide without storage: renewable baseload power. Our [renewable energy coverage](/articles/kenya-renewables-industrial-power-data-centres) explains why that baseload quality matters more to data centres than the headline share. ![Steam pipelines and cooling towers at the Olkaria geothermal complex in Naivasha](/images/kenya-geothermal-plant-olkaria.webp) The strategic weight of that mix is growing as AI reshapes demand. The next facility is not always a conventional colocation building with incremental load. It may be a large campus that needs substantial power at one location, along with a credible path to expand. A workload hosted on a grid with Kenya's renewable share carries a different carbon story from one hosted in a coal-heavy system that must assemble its clean-power position through contracts and certificates. But renewable generation is not the same thing as firm, site-specific data-centre power. A grid can have enough capacity in aggregate while a particular site still faces connection, transmission, substation or delivery constraints. Kenya's own record makes the point: the 100 MW Microsoft and G42 project, the single largest early-stage record in the country's tracked pipeline, has stalled on grid power constraints. Delivering 100 MW or more to one site is a different engineering and commercial problem from serving a growing city. > **Kenya has a high-renewable generation base, but it now needs to prove that clean power can be delivered reliably, at scale and on the timelines large digital-infrastructure customers require.** ## Connectivity gives Kenya a second advantage Power is only one part of the location decision. Kenya also has a credible connectivity foundation, and unlike the pipeline, most of it already exists. Seven live subsea cable systems connect to the Kenyan coast: TEAMS, SEACOM, EASSy, LION2, DARE1, PEACE and 2Africa. Africa-1 has come ashore but remains pending a confirmed ready-for-service date. Daraja is announced, and LuLu is planned. The [cable tracker](/tracker/cables) keeps those categories separate rather than counting every announced route as live, which is the same staging discipline the capacity numbers follow. ![Mombasa port where Kenya's submarine cables come ashore](/images/mombasa-port-wide.webp) That coastline matters because Kenya's digital infrastructure plays out across two connected roles. Mombasa is the international gateway, where cable systems land and international routes enter the country. Nairobi is the infrastructure cluster, concentrating data centres, enterprise demand, operators, networks and interconnection. The national opportunity lies in the chain between the two: landing stations, terrestrial fibre, Nairobi facilities, internet exchanges and the businesses consuming digital services at both ends. The interconnection data shows where that depth is already accumulating. In the latest PeeringDB cross-check, [the market numbers](/articles/kenya-data-centre-market-numbers) record Africa Data Centres Nairobi 1 leading with 122 networks and four internet exchanges on site. iColo Mombasa One follows with 94 networks and three exchanges, a function of its proximity to the landing stations, while iColo Nairobi One carries 62 networks and four exchanges. These are public registration signals rather than a complete measure of facility quality, but they show a real interconnection layer forming, not an aspiration. ![Rows of server racks inside an operational Nairobi data centre hall](/images/atlancis-nairobi-datacentre-hall.webp) ## The real competition is deliverability Africa does not need more generic claims about being the next data-centre frontier. It needs to demonstrate delivery, and the analysts saying this loudest are the ones selling the market data. DC Byte's analysis identifies power availability and delivery timelines as the test that matters, and it warns that early-stage pipelines can exaggerate the capacity that ever becomes operational. Milan is the cautionary example: almost 70 percent of that market's total IT capacity remains at an early development stage, which is exactly how headline pipeline figures outrun reality. Announcements should not automatically be treated as completed facilities. Kenya's own figures make the same argument from the inside. The country has 230 MW of tracked announced pipeline, but only 10.5 MW of published live IT load and 42.9 MW of built operational capacity. That is not a weakness in the data. It is the point of staging the data honestly. For Kenya to move from a credible emerging market to a preferred destination for the next wave, four conversions matter: 1. **Pipeline to operational capacity:** projects must be completed and commissioned, not re-announced. 2. **Generation to firm delivery:** renewable electricity must reach large sites with the required reliability and redundancy. 3. **Connectivity to customer value:** cable access must translate into networks, cloud on-ramps, exchange fabric and low-latency services. 4. **Announcements to evidence:** every major claim needs a source, a status and a verification date. ## Where Kenya sits in the African race Kenya is not South Africa. It does not have the same depth of installed capacity, operator footprint or established hyperscale market, and it should not compete by pretending otherwise. In the same analysis, DC Byte identifies South Africa as the African market to watch and names Nigeria, Kenya, Egypt and Morocco among the continent's key data-centre markets, noting that Nigeria and Kenya are moving towards wholesale facilities capable of accommodating hyperscale demand. International operators are voting with deployments too: Continent 8 Technologies is entering Africa through Johannesburg, with its first African facility live in Q4 2026, close to one of the world's most intensive regulated betting markets. Kenya's distinctive combination is still real, and it looks like this: - A renewable-heavy generation mix with geothermal baseload near 45 percent of supply - Seven live subsea cable systems, with more landed or announced - A growing Nairobi interconnection cluster anchored by 122-network facilities - A mature mobile, fintech and enterprise demand base - 27 tracked facilities, with 20 operational - A 230 MW pipeline that is large enough to matter but still has to be converted - A location from which operators can serve East Africa and connect to wider regional markets The honest proposition is therefore not a slogan about becoming the next Johannesburg. > **Kenya can offer clean generation, international connectivity and a growing digital market, but its next competitive advantage will be the ability to deliver reliable power and commissioned capacity faster than its pipeline suggests.** ## The question for 2027 The next phase of Kenya's market will be decided less by the size of the announcement pipeline than by what becomes operational. The [current dataset](/infrastructure/map) puts specific names on the test. The watchlist includes the completion of iXAfrica NBOX1.2, Africa Data Centres Nairobi 2 and Nxtra by Airtel's Tatu City campus, with commissioning targeted for 2027, alongside iXAfrica NBOX2 at Tilisi moving from committed to built. It also includes the decision path around the Microsoft and G42 project and the interconnection ramp of iColo Nairobi Two, Digital Realty's freshly launched NBO2. Each of those projects answers one of the operational questions the analysis says will decide the next wave. Africa's 3 percent share of EMEA capacity is not a verdict. It is a starting position. The continent has room, demand and increasingly visible connectivity. But the next wave will go to the markets that can answer the operational questions: Is the power available? Can it be delivered to the site? How quickly can the facility be built? Is the route diversity real? Are the networks and customers already present? What portion of the pipeline is genuinely committed? Kenya has a credible answer to several of those questions already. The country's opportunity is to make the remaining answers measurable. That is where the race begins. ## Explore the evidence - [DC254 State of Kenyan Data Centres 2026](/research/state-of-kenyan-data-centres-2026) - [DC254 Infrastructure Map](/infrastructure/map) - [Kenya Data Centre Directory](/directory) - [DC254 Energy & Power coverage](/energy) - [Kenya Subsea Cable Tracker](/tracker/cables) ### Sources used in this article **Primary** - [ITWeb, Land availability gives SA edge in data centre race (5 October 2026): DC Byte EMEA analysis, 93 GW total, Europe 88 percent, Middle East 9 percent, Africa 3 percent, FLAP-D 24 percent, Milan early-stage caveat](https://www.itweb.co.za/article/land-availability-gives-sa-edge-in-data-centre-race/raYAyqorDmRMJ38N) - [DC254, State of Kenyan Data Centres 2026 (27 facilities, 20 operational, 10.5 MW published live IT load, 42.9 MW built, 230 MW announced pipeline)](/research/state-of-kenyan-data-centres-2026) - [DC254, Kenya Subsea Cable Tracker (7 in service, Africa-1 landed with RFS pending, Daraja announced, LuLu planned)](/tracker/cables) **Supporting** - [The Star, KenInvest commentary (5 October 2026): about 93 percent of grid-generated electricity comes from geothermal, hydro, wind and solar](https://www.the-star.co.ke/business/commentary/2026-10-05-mwendwa-why-investors-are-putting-their-money-in-kenya) - [DC254, Kenya's Renewable Energy Meets Data Centre Demand (geothermal near 45 percent of generation)](/articles/kenya-renewables-industrial-power-data-centres) - [Developing Telecoms, Continent 8 enters African market with Joburg data centre (5 October 2026)](https://developingtelecoms.com/telecom-technology/data-centres-networks/20920-continent-8-enters-african-market-with-joburg-data-centre.html) - [Continent 8, launch announcement for South Africa](https://www.continent8.com/continent-8-technologies-launches-in-south-africa-extending-its-private-global-network-into-africa/) - [Data Center Dynamics, Continent 8 Technologies launches data center in Johannesburg](https://www.datacenterdynamics.com/en/news/continent-8-technologies-launches-data-center-in-johannesburg-south-africa/) *Dataset checked: 7 October 2026.* ## Africa's Data Centre Boom: Managing the Disputes Risk URL: https://data-centers-254.vercel.app/articles/africa-data-centre-investment-disputes Cluster: Policy & Regulation Published: 2026-09-04 Updated: 2026-09-04 Summary: Less than 1% of global data centre capacity sits in Africa, and investors are racing in. HSF Kramer maps the dispute risks: power, land, water. Less than 1 percent of the world's data centre capacity sits in Africa. That statistic is unlikely to last, and, according to a September 2026 insight from global law firm HSF Kramer, neither will the relative absence of disputes that has accompanied the continent's early build-out. The firm's analysis, published as investors race toward African digital infrastructure, argues that the opportunity is enormous but the challenges are many: power constraints, land rights, supply chain pressures, water access, regulatory scrutiny, and all the usual execution risks each have the potential to turn promising projects into disputes. For investors, the firm concludes, understanding and managing these risks will be just as important as identifying the next growth market. That framing deserves a wider audience in East Africa, because the disputes conversation is the one regional stakeholders most often skip. Kenyan operators discuss megawatts, certifications, and submarine cables endlessly; they discuss arbitral seats, escalation clauses, and delay damages rarely, and usually only after something has gone wrong. This article unpacks HSF Kramer's risk map, translates it into the Kenyan and East African context, and sets out the practical structuring moves that separate projects that bend from projects that break. ## Why the Capacity Gap Is Closing, and Why Disputes Follow The tailwinds behind Africa's data centre build-out are no longer speculative. Demand for cloud services, AI infrastructure, and local data storage is rising fast across the continent, and governments are actively compounding it: data sovereignty and localisation frameworks emerging across Africa deliberately favour domestic and regional capacity, converting policy into anchor demand. Our own coverage of [Kenya's data sovereignty rules](/articles/kenya-data-sovereignty-localisation) and [how African regulation compares across markets](/articles/africa-data-centre-regulation-compared) traces the same dynamic in detail, regulated demand is the most durable demand there is. But the same forces that compress the capacity gap compress the timeline for legal stress-testing. Infrastructure sectors that grow slowly accumulate case law: developers learn from disputes, contracts standardise, and courts and tribunals develop expertise. Africa's data centre sector is growing too fast for that organic maturation. Contracts are drafted fresh for each project, counterparties range from sovereign utilities to first-time sponsors, and the underlying dependencies (grid connections, water allocations, equipment shipping windows) sit outside any single party's control. When growth outpaces institutional learning, the disputes that would have been absorbed quietly in a mature market instead arrive as novel, expensive, relationship-ending conflicts. ![Legal and regulatory documents beside data centre infrastructure](/images/dc-policy-regulation.webp) ## The Risk Map: Where Projects Turn Into Disputes HSF Kramer's insight identifies the pressure points that recur across African data centre investments, and each deserves a line of sight from anyone underwriting or operating a facility in the region. Power constraints lead the list, and for good reason: grid connection delays and generation shortfalls sit upstream of every other commitment a facility makes. When a colocation operator has promised a hyperscale tenant a go-live date, and the utility's feeder energisation slips by two quarters, the resulting claims cascade, tenant delay damages, contractor acceleration costs, financing standby charges. In Kenya, where [grid capacity and transmission timing](/articles/kenya-power-infrastructure-data-centres) shape every project plan, power-related risk allocation is the single most important set of clauses in any contract stack. Land rights come next, and they arrive in several disguises: contested title, overlapping customary or community interests, wayleave disputes for fibre and power routes, and change-of-use approvals that stall at county level. Supply chain pressures (long-lead transformers, switchgear, generators, and cooling equipment with multi-year manufacturing queues) create shared-risk exposures that contracts drafted in abundant-equipment eras never anticipated. Water access is emerging as its own flashpoint as cooling-hungry facilities meet municipal supply constraints and community expectations, a tension our [cooling systems explainer](/articles/data-centre-cooling-systems-explained) touches from the engineering side. Regulatory scrutiny adds the public-law dimension: licensing conditions, environmental approvals, and data protection compliance can each change the economics of an operating facility, and each generates its own category of dispute when expectations diverge. ![High-rise towers rising over the Nairobi skyline](/images/nairobi-expressway-wide.webp) ## Managing the Risk Before It Becomes a Dispute The consistent thread in HSF Kramer's analysis is that mitigation is a design activity, not a reaction. By the time a disagreement hardens into a dispute, the contractual architecture that determines who bears which risk was fixed years earlier, usually when both parties were optimistic and drafting attention was focused on price rather than process. The practical toolkit starts with contract architecture. Milestone payment structures tied to independently verified progress keep cash flow aligned with delivery reality. Liquidated damages regimes that price delay realistically (in both directions, including extension-of-time mechanisms that acknowledge African infrastructure dependencies) reduce the incentive to litigate every schedule slip. Force majeure clauses should be drafted for the continent's actual supply chain and utility conditions rather than imported wholesale from other markets, and change-of-law or tax-stability provisions matter in jurisdictions where the regulatory framework is still being written. None of this is exotic; almost all of it is skipped somewhere in East Africa every quarter. Dispute resolution design deserves equal intention. The choice of seat and rules (the Nairobi Centre for International Arbitration, the LCIA, the ICC, or another institution) should weigh enforceability, witness location, cost, and sector familiarity rather than defaulting to habit. Kenya's participation in the New York Convention means arbitral awards from properly constituted tribunals are enforceable locally, and the Kenyan Arbitration Act gives parties a modern statutory backbone; our [licensing framework overview](/articles/kenya-data-centre-licensing-framework) explains the regulatory context those tribunals will interpret. Investment treaty protection is the final layer: investors structuring through jurisdictions with bilateral investment treaties gain a public international law remedy that sits above the contract stack, protection worth understanding before, not after, a regulatory shock. ![Investment terms and financial analysis for data centre projects](/images/dc-business-investment.webp) ## The Kenya Lens Applying the risk map to Kenya produces a familiar picture with local texture. Power risk concentrates around [Kenya Power](/articles/kenya-power-tariffs-data-centres) interconnection timelines and the terms on which facilities secure dedicated feeders, making grid-connection agreements the most consequential documents most Kenyan operators sign. Land risk in Nairobi and Mombasa plays out through title diligence and county-level approvals, a theme our [location analysis](/articles/why-data-centres-cluster-nairobi-mombasa) examines from the siting side. Regulatory risk runs through the Communications Authority's [NFP-T1 and NFP-T2 licensing](/articles/kenya-data-centre-licensing-framework) and the compliance obligations of the [Data Protection Act](/articles/kenya-data-protection-act-data-centres), both of which create contractual promises operators must keep to enterprise customers. Kenya's disputes infrastructure, meanwhile, is a relative strength. The Nairobi Centre for International Arbitration gives the market a domestic seat with international ambitions, the judiciary has developed a reasonable track record on arbitration support, and the legal profession fields genuine infrastructure disputes expertise. For East African projects structured through Kenya, the pragmatic question is increasingly not whether Kenya can host a fair arbitration but whether the contracts were drafted to make the answer irrelevant. ## Underwrite the Exit, Not Just the Entry The deepest lesson in HSF Kramer's insight is one the broader African investment community is still absorbing: dispute readiness is a capital markets variable, not a legal afterthought. Lenders price enforcement risk. Equity investors discount governance ambiguity. Strategic acquirers, the same institutional buyers our [Keppel DC REIT analysis](/articles/keppel-dc-reit-tokyo-acquisition-signals) describes shopping for metrics, conduct dispute diligence on the way in, and a single unresolved construction claim can reprice an entire portfolio transaction. Projects that can evidence clean contract stacks, sensible escalation paths, and enforceable security are, mechanically, cheaper to finance. ![Map of African data centre markets](/images/africa-dc-map.webp) For Kenya's operators, developers, and investors, the moves are concrete. Audit the current contract stack against the risk map (power, land, supply chain, water, regulation) and find the gaps before a counterparty does. Standardise dispute resolution clauses across a portfolio so precedents compound. Build compliance narratives that match contractual promises, particularly under the Data Protection Act. And treat the HSF Kramer intervention as what it is: a signal from the disputes bar that it expects the work. Booms generate disputes everywhere; only well-drafted ones survive them with their valuations intact. ## Data Centre Regulation in Africa: 5 Countries Compared URL: https://data-centers-254.vercel.app/articles/africa-data-centre-regulation-compared Cluster: Policy Published: 2026-08-28 Updated: 2026-09-28 Summary: Compare how Kenya, South Africa, Nigeria, Egypt, and Rwanda regulate data centres: licensing, data protection, tax incentives, and lessons for Kenya. Africa's data centre market is growing faster than any other continent, but it is not growing uniformly. The pace and pattern of investment is shaped by regulatory frameworks that vary enormously from country to country, from South Africa's mature, complex regime to the near-absence of specific regulation in several smaller markets. For investors and operators comparing opportunities across the continent, understanding these regulatory differences is not an academic exercise, it directly affects project costs, timelines, risk profiles, and ultimately the return on investment. The short answer: South Africa has the most mature but most complex regime; Egypt's rules are the clearest but state-led; Nigeria pairs very large demand with very high power costs and an evolving framework; Rwanda runs the lightest-touch licensing with aggressive ICT incentives; and Kenya is converging on clarity, with a defined licence class, a GDPR-grade data protection act, and SEZ corporate tax rates as low as 10%. ![Nairobi Expressway stretching across the city](/images/nairobi-expressway-wide.webp) This analysis compares the regulatory environments for data centres in five key African markets: Kenya, South Africa, Nigeria, Egypt, and Rwanda. These five countries represent the most significant data centre markets or the most interesting emerging opportunities, and their regulatory approaches offer lessons for the entire continent. ## Kenya: Clarity Emerging Kenya's regulatory framework for data centres has become significantly clearer since the Communications Authority's decision to bring data centres under the [licensing framework](/articles/kenya-data-centre-licensing-framework). The requirement for data centre operators to hold an NFP-T2 licence (Non-Facility Based Type 2, costing KES 15 million for a 15-year term) established formal regulatory oversight for the first time. The [Policy Intelligence hub](/policy) tracks licence classes, tariffs, and compliance obligations as they are gazetted. The [Data Protection Act of 2019](/articles/kenya-data-protection-act-data-centres), modelled on the EU's GDPR, provides a comprehensive data protection framework that gives international customers confidence that their data will be handled according to globally recognised standards. The Office of the Data Protection Commissioner (ODPC) is still building its capacity, but the legislative foundation is solid. Kenya's renewable energy policy (over 90% grid power from geothermal, hydro, wind, and solar) provides a structural advantage that no regulatory incentive can match. The country's corporate tax rate of 30% applies to data centres, though the Special Economic Zone programme (applicable to Konza Technopolis and other SEZ-designated areas) offers a reduced rate of 10% for the first 10 years and 15% for the next 10 years. Import duties on data centre equipment (servers, UPS systems, networking gear) are a challenge. Kenya applies the East African Community Common External Tariff, which ranges from 0% to 25% depending on the equipment category. Some ICT equipment qualifies for VAT exemption under the EAC protocol, but the application is inconsistent and the customs classification process can be slow. Harmonising import duty treatment for data centre equipment across the EAC would significantly reduce deployment costs. ## South Africa: Mature but Complex South Africa has Africa's most developed data centre market, with over 200 megawatts of IT load capacity across dozens of facilities. Its regulatory framework is correspondingly the most comprehensive on the continent, though complexity is a defining characteristic. ### Licensing and Regulation The Independent Communications Authority of South Africa (ICASA) regulates electronic communications, including data centres. Colocation facilities that provide connectivity to multiple networks may require an Electronic Communications Service (ECS) licence. Facilities that primarily provide space, power, and cooling without offering network services may fall outside ICASA's direct jurisdiction, though the boundary is not always clear. This regulatory ambiguity has been a source of frustration for operators. ### Data Protection South Africa's Protection of Personal Information Act (POPIA), fully effective since July 2021, is closely modelled on the GDPR and is widely regarded as Africa's most comprehensive data protection law. The Information Regulator, established under POPIA, has been slower to enforce than some expected, but the law's requirements are clear and well-understood by the business community. International companies operating in South Africa typically apply POPIA compliance as a baseline. ### Power Challenges as a Regulatory Issue ![Map of African data centre hubs](/images/africa-dc-map.webp) South Africa's ongoing load shedding crisis (scheduled power outages implemented by Eskom to manage supply shortages) has become, in effect, a regulatory issue for data centres. While not a regulation per se, the unreliable grid forces data centres to depend heavily on diesel generators, increasing operating costs, emissions, and maintenance requirements. The South African government has introduced tax incentives for self-generation (including solar and battery storage), but these are partial remedies for a systemic problem. ### Tax Incentives South Africa offers several incentives relevant to data centre investment. The Investment Tax Allowance allows companies to deduct 100% of the cost of qualifying assets in the year of expenditure, significantly reducing the effective cost of data centre equipment. Special Economic Zones offer reduced corporate tax rates and customs duty benefits. The Section 12B tax incentive for renewable energy generation allows data centres to deduct the cost of solar installations from taxable income. ## Nigeria: High Potential, High Complexity Nigeria is Africa's largest economy and most populous country (220 million people), making it a natural data centre market. However, the regulatory and infrastructure challenges are significant. ### Licensing The Nigerian Communications Commission (NCC) requires data centre operators to hold appropriate licences. The specific licence category depends on the services offered, facilities providing connectivity require an ISP or carrier licence, while pure colocation may fall under a different category. The NCC has been modernising its licensing framework, moving toward a technology-neutral approach that is more flexible for data centre operators. ### Data Protection Nigeria's Data Protection Regulation (NDPR), issued by the National Information Technology Development Agency (NITDA) in 2019 and updated as the Nigeria Data Protection Act (NDPA) in 2023, establishes data protection requirements including restrictions on cross-border data transfers. The NDPA is less comprehensive than POPIA or Kenya's DPA, but it is evolving. NITDA has been active in issuing guidance and enforcement notices. ### The Power Problem Nigeria's unreliable power grid is the single biggest regulatory and operational challenge for data centres. Grid power is available for only a fraction of the day in many areas, forcing data centres to run on diesel generators for the majority of their operating hours. This dramatically increases operating costs, diesel-powered electricity can cost $0.25–0.40 per kWh, compared to $0.08–0.12 from Kenya's grid. There is no regulatory solution to this problem in the near term; it requires investment in power generation and grid infrastructure that will take years. ### Tax and Investment Climate Nigeria offers the Pioneer Status Incentive, which provides a tax holiday of 3–5 years for companies in qualifying industries. Data centres have been discussed as a potential qualifying industry, though the incentive has not been consistently applied to the sector. The Nigerian Investment Promotion Commission offers facilitation services for foreign investors, and the country's large market size provides a natural demand base. ## Egypt: Government-Led, Subsidised Egypt's data centre market is driven significantly by government investment and policy direction. The country's New Administrative Capital, east of Cairo, is being developed as a technology hub with data centre infrastructure as a core component. ### Licensing The National Telecommunications Regulatory Authority (NTRA) regulates data centres in Egypt. The licensing framework is relatively straightforward, with clear categories for facility-based and non-facility-based operators. The Egyptian government has been proactive in attracting data centre investment, offering regulatory support and streamlined licensing processes for qualifying projects. ### Data Protection Egypt's Personal Data Protection Law (PDPL), enacted in 2020 and progressively implemented, establishes data protection requirements including data localisation for certain categories of government data. The law restricts cross-border transfers and requires data controllers to obtain consent or demonstrate adequate protection in the recipient country. For international operators, Egypt's data localisation requirements for government data create a clear demand driver for domestic data centre capacity. ### Energy Subsidies ![National government building at dusk](/images/dc-policy-regulation.webp) Egypt's data centre industry benefits from subsidised natural gas for power generation, which keeps electricity costs among the lowest in Africa at approximately $0.05–0.07 per kWh for industrial consumers. This subsidy is a significant competitive advantage, though it is contingent on government policy that could change. The government has also invested heavily in renewable energy, particularly solar, as part of its energy strategy. ## Rwanda: Small but Strategic Rwanda is Africa's smallest data centre market by physical capacity, but its regulatory environment is among the most progressive on the continent. ### Licensing Rwanda has minimal specific data centre licensing requirements. The Rwanda Utilities Regulatory Authority (RURA) regulates telecommunications, but data centres that do not provide telecommunications services may not require a specific licence. This regulatory light-touch approach is intentional, Rwanda's government has made digital transformation a national priority and has deliberately removed regulatory barriers to digital infrastructure investment. ### Data Protection Rwanda enacted its Law No. 058/2021 Relating to the Protection of Personal Data and Privacy in 2021. The law establishes data protection principles similar to GDPR and creates an data protection authority. While enforcement capacity is limited by the country's size, the legislative framework is modern and investor-friendly. ### Investment Incentives Rwanda offers some of the most aggressive investment incentives in Africa for technology companies. The Kigali Innovation City project provides tax holidays, duty-free import of equipment, and streamlined business registration. The Rwanda Development Board offers fast-track business registration (often completed within 24 hours) and dedicated support for technology investors. Corporate tax for qualifying ICT investments can be reduced to as low as 0% for an initial period. Rwanda's challenge is market size. With a population of 14 million and a smaller economy than Kenya, Nigeria, or South Africa, the domestic demand for data centre services is limited. However, Rwanda's strategic positioning as a hub for East and Central Africa, its excellent fibre connectivity, and its government's commitment to digital transformation make it an interesting market for operators with a regional strategy. ## Comparative Summary | Factor | Kenya | South Africa | Nigeria | Egypt | Rwanda | |--------|-------|-------------|---------|-------|--------| | Licensing clarity | Improving | Complex | Evolving | Clear | Minimal | | Data protection law | Strong (DPA 2019) | Strong (POPIA) | Moderate (NDPA) | Moderate (PDPL) | Strong (2021) | | Grid power cost | Low | Medium | Very high | Very low | Medium | | Grid reliability | Good | Poor | Poor | Good | Good | | Tax incentives | SEZ available | IAA, SEZ | Pioneer status | Gas subsidies | Aggressive ICT incentives | | Market size | Large | Very large | Very large | Large | Small | | Data localisation | Restricted transfers | Restricted transfers | Partial requirements | Government data | Discussed, not enacted | ![Westlands commercial district in Nairobi](/images/nairobi-westlands.webp) ## What Kenya Can Learn Kenya's regulatory framework is competitive and improving, but there are lessons from other African markets. From South Africa, Kenya can learn the value of tax incentives that directly target data centre investment, the Investment Tax Allowance that allows 100% upfront deduction of equipment costs would significantly reduce the capital barrier for new facilities. From Rwanda, Kenya can learn the value of regulatory simplicity, a fast-track licensing process for data centre operators, with clear timelines and transparent criteria, would reduce the uncertainty that deters investment. From Egypt, Kenya can observe the strategic value of deliberate government direction, the New Administrative Capital project demonstrates how government commitment to a specific location can catalyse private investment. And from all markets, Kenya can observe that the trend toward stricter data protection and data transfer restrictions is accelerating, which structurally benefits domestic data centre capacity. For investors evaluating African data centre opportunities, the regulatory framework is one of several factors (alongside market size, connectivity, power, and skills) that determine where to invest. Kenya's combination of improving regulation, competitive energy costs, and strategic connectivity positions it well, but continued regulatory refinement is essential to maintain this advantage as other markets improve their own frameworks. ## Africa Data Centres: Operations in Kenya and Beyond URL: https://data-centers-254.vercel.app/articles/africa-data-centres-kenya-operations Cluster: Operators Published: 2026-08-29 Updated: 2026-08-29 Summary: Africa Data Centres, a Cassava Technologies company, operates colocation facilities in Nairobi and Mombasa within a 40+ site Pan-African network. Africa Data Centres (ADC) is the largest data centre platform on the African continent by geographic reach, operating more than 40 facilities across 13 countries. As a subsidiary of Cassava Technologies, the pan-African technology group chaired by Zimbabwean telecoms billionaire Strive Masiyiwa, ADC brings a scale and continental footprint that no other operator on the continent can match. In Kenya, the company operates three facilities, two in Nairobi and one in Mombasa, forming a critical part of the country's digital infrastructure backbone. ![Server racks in a modern data centre hall](/images/whats-inside-ai-data-center-wide.webp) This profile examines ADC's origins, its Kenyan operations, its carrier-neutral model, and the competitive dynamics shaping its position in one of East Africa's fastest-growing data centre markets. ## From Liquid Intelligent Technologies to Cassava Technologies Africa Data Centres traces its lineage through a complex corporate evolution that is important to understand because it explains both the company's strengths and its strategic positioning. The story begins with Liquid Telecom, the fibre and telecommunications company founded by Strive Masiyiwa's Econet Group. Over two decades, Liquid built Africa's largest privately owned fibre network, stretching across more than 100,000 kilometres and connecting over 15 countries. This network became the backbone of data centre strategy: if you owned the fibre connecting businesses and carriers across Africa, building data centres at key interconnection points was a natural vertical integration move. Liquid Telecom's data centre assets, along with those acquired through its purchase of Dimension Data's Middle East and Africa operations in 2016, formed the foundation of what would become Africa Data Centres. Dimension Data, a South Africa-headquartered IT services company owned by NTT (Nippon Telegraph and Telephone), operated data centre facilities in several African markets including Kenya. The Dimension Data acquisition gave Liquid not just physical facilities but also enterprise relationships and technical capabilities that accelerated its data centre ambitions. In 2022, the parent company rebranded from Liquid Intelligent Technologies to Cassava Technologies, signalling a shift from telecommunications to a broader pan-African technology platform spanning connectivity, cloud, cybersecurity, and digital infrastructure. Africa Data Centres was established as a distinct subsidiary brand within the Cassava portfolio, giving it operational independence and a focused identity. Cassava Technologies is majority-owned by Econet Global, with the Public Investment Corporation of South Africa holding a significant minority stake, and is listed on the London Stock Exchange through a secondary listing. ## Pan-African Footprint Africa Data Centres' defining characteristic is scale. The company operates facilities in South Africa, Kenya, Nigeria, Egypt, Morocco, Tunisia, Senegal, Cote d'Ivoire, Ghana, Zambia, Zimbabwe, and Uganda, with additional markets under development. South Africa represents the largest concentration of facilities, with multiple campuses in Johannesburg, Cape Town, and Durban, but Kenya is the company's most important market outside Southern Africa. This pan-African footprint is a significant competitive advantage because it enables multi-country deployments under a single contractual framework, with consistent service level agreements, security standards, and operational procedures. A bank headquartered in Nairobi with operations in Dar es Salaam, Lusaka, and Lagos can collocate across all four markets through one provider. This is a capability that national operators or single-market entrants cannot match. ## Kenya Operations: Nairobi 1, Nairobi 2, and Mombasa ![Engineer inspecting server rack cabling with a tablet](/images/dc-engineer-rack-inspection.webp) ADC's Kenyan operations consist of three facilities strategically positioned to serve both the Nairobi enterprise market and the international connectivity hub at the coast. ### Nairobi 1 (Sameer Park) ADC's original Kenyan facility, commonly referred to as Nairobi 1, is located within the Sameer Business Park along Mombasa Road in Nairobi's Industrial Area. Sameer Park has long been a hub for technology and telecommunications operations, housing multiple network operators and technology companies. Nairobi 1 provides enterprise colocation services with multiple carrier interconnections, serving financial institutions, government agencies, telecommunications operators, and technology companies. As ADC's first facility in Kenya, it carries the established customer relationships that form the company's foundation in the market. The facility offers carrier-neutral access to multiple network providers, allowing tenants to select and negotiate connectivity arrangements independently. ### Nairobi 2 (Mombasa Road) Nairobi 2 represents a more modern addition to the company's Kenyan portfolio, also located along the Mombasa Road corridor and developed to address growing demand with updated infrastructure specifications. The Mombasa Road location for both Nairobi facilities is strategically significant. It is the primary fibre corridor connecting Nairobi to the coastal submarine cable landing stations. All four major subsea cables serving Kenya, EASSy, SEACOM, TEAMS, and DARE1, land in Mombasa and their terrestrial extensions pass through this corridor. Facilities on Mombasa Road benefit from direct access to diverse fibre routes, low-latency connectivity to international gateways, and proximity to networks serving landlocked East African markets. ![Blue-lit server racks in a data hall](/images/hero-server-hall.webp) ### Mombasa Facility ADC's Mombasa data centre extends the company's reach directly to the coastal connectivity hub. Located near the submarine cable landing stations, it serves as a critical interconnection point where international cables meet terrestrial fibre networks carrying traffic inland to Nairobi and onward to Uganda, Rwanda, Burundi, South Sudan, and the eastern Democratic Republic of Congo. The facility is particularly important for content delivery networks, cloud providers, and wholesale carriers that need to cache content and exchange traffic close to the landing points. For a deeper understanding of the international connectivity landing at the coast, see our guide to [submarine cables in Mombasa](/articles/submarine-cables-landing-mombasa). ## Carrier-Neutral Model and Connectivity ADC's carrier-neutral model is central to its value proposition in Kenya and across the continent. Unlike facilities owned by telecommunications companies that may favour their own network services, ADC's facilities are designed to give tenants full freedom of choice in network provider selection. In Kenya, ADC facilities provide access to connectivity from Liquid Datapoint (the Cassava group's own network arm), SEACOM, WIOCC, and multiple other carriers. This diversity of connectivity options serves several purposes. Enterprise customers can maintain relationships with their preferred network providers while benefitting from the physical security, power reliability, and cooling infrastructure of a professional data centre. Network operators can establish points of presence to interconnect with each other and with enterprise customers without building their own facilities. Cloud providers can connect to multiple downstream carriers for optimal routing and redundancy. The carrier-neutral approach also supports the growth of internet exchange points and peering ecosystems. By providing neutral ground where networks can interconnect, ADC facilities contribute to the localisation of internet traffic, reducing the volume of Kenyan traffic that needs to transit through international links and improving both latency and cost for end users. ## Competition with iXAfrica The Kenyan data centre market has become increasingly competitive, and the most significant competitive dynamic is between ADC and iXAfrica, the Actis-backed operator that launched East Africa's first purpose-built hyperscale facility, NBOX1, on Mombasa Road in early 2025. The two operators represent fundamentally different approaches to the market. ADC brings the advantage of continental scale, an established customer base built over years of operation, and the backing of a parent company that also owns Africa's largest fibre network. Its facilities are proven and operational, with long track records of reliability serving Kenyan enterprises. For many customers, particularly those with multi-country African operations, ADC's pan-African platform is a decisive factor. iXAfrica, by contrast, offers a newer, purpose-built facility designed from the ground up for high-density workloads. NBOX1 delivers up to 50 kW per rack, a specification that exceeds what ADC's existing Nairobi facilities were originally designed to support. This makes iXAfrica the preferred option for AI training clusters, GPU-dense workloads, and hyperscale cloud deployments that require high rack densities and advanced cooling infrastructure. In practice, the two operators are likely to coexist and serve overlapping but distinct segments of the market. ADC will continue to dominate in traditional enterprise colocation, multi-country deployments, and customers who value the Cassava ecosystem integration. iXAfrica will capture the high-density and hyperscale segment. For customers evaluating between them, the choice depends primarily on workload requirements, rack density needs, and whether multi-country African coverage is a factor. For a full comparison of operators and facilities, visit our [data centre directory](/directory). If you are evaluating colocation for the first time, our guide to [colocation in Kenya](/articles/what-is-colocation-kenya) covers the fundamentals. ## Cloud On-Ramp Partnerships One of the most strategically important dimensions of ADC's business is its role as a cloud on-ramp provider. As more African enterprises migrate workloads to public cloud platforms, the data centre that hosts the direct, dedicated connection between the enterprise network and the cloud provider becomes a critical piece of infrastructure. ADC has established partnerships with several major cloud providers and technology platforms, positioning its facilities as the preferred points of presence for cloud services across Africa. These partnerships allow enterprise customers to connect to cloud platforms through dedicated, private connections rather than traversing the public internet, providing improved security, consistent performance, and predictable latency. In the Kenyan context, cloud on-ramp services are becoming increasingly important as the country's financial services sector, government agencies, and technology companies accelerate cloud adoption. The Central Bank of Kenya's cloud computing guidelines for financial institutions, combined with the broader push toward digital government services, are driving demand for secure, high-performance connections to cloud platforms. ADC's facilities in Nairobi are well-positioned to serve this demand. The cloud on-ramp strategy also positions ADC to benefit from the entry of hyperscale cloud providers into Kenya. Oracle has announced plans for a sovereign cloud region in the country. Microsoft has committed significant investment to establish a regional cloud zone. These deployments require local data centre infrastructure, and ADC's existing facilities, customer relationships, and continental platform make it a natural partner for such initiatives. ## Sustainability Efforts Sustainability has become a non-negotiable requirement for data centre operators globally, and ADC has made environmental responsibility a visible part of its strategy. The company has committed to achieving net-zero carbon emissions across its operations, a significant undertaking for a platform of its scale operating across diverse African energy markets. Kenya's energy profile gives ADC a structural advantage in this regard. Approximately 93% of Kenya's electricity grid is sourced from renewables, with geothermal energy from the Olkaria geothermal field in the Rift Valley being the single largest contributor. Kenya is the sixth-largest geothermal energy producer globally, a fact that means every kilowatt-hour consumed by ADC's Kenyan facilities has a substantially lower carbon footprint than equivalent consumption in markets dependent on coal, gas, or heavy fuel oil generation. ADC has invested in energy efficiency improvements across its facility portfolio, including modernising cooling systems, optimising power distribution architectures, and deploying monitoring systems that enable real-time tracking of Power Usage Effectiveness (PUE) and other efficiency metrics. The company reports performance metrics aligned with global sustainability reporting frameworks, reflecting the expectations of its international enterprise and cloud customers who require verifiable ESG credentials from their infrastructure partners. The broader sustainability conversation in Kenya's data centre sector also includes the anticipated enactment of power wheeling legislation, which would allow large consumers to purchase electricity directly from renewable generators. If implemented, this could enable ADC and other operators to contract directly with geothermal producers, potentially reducing both energy costs and further improving the carbon intensity of operations. ## Significance for East Africa ADC's Kenyan operations matter beyond the country's borders. Kenya serves as the digital gateway to a landlocked hinterland of over 200 million people across Uganda, Rwanda, Burundi, South Sudan, the eastern Democratic Republic of Congo, and parts of Tanzania and Ethiopia. Data centre infrastructure built in Nairobi inherently serves this broader regional market. ADC's position is strengthened by the fact that its parent company, Cassava Technologies, operates fibre networks that physically connect Nairobi to these landlocked markets. The synergy between Cassava's connectivity infrastructure and ADC's data centre facilities creates a vertically integrated platform that can offer end-to-end solutions: fibre connectivity from the enterprise location in Kampala or Kigali, through to colocation and cloud on-ramp services in Nairobi. No other operator in the region can offer this combination of data centre and connectivity at continental scale. The submarine cable landscape is also evolving in ways that favour Kenya and, by extension, ADC's Kenyan facilities. The 2Africa cable, one of the largest subsea projects globally, is expected to land in Kenya, adding massive international bandwidth capacity. Google's Africa-1 cable is under development with planned Kenyan landing. These next-generation cables will reduce the cost per megabit of international connectivity and enable new categories of data-intensive services. ADC's facilities on the Nairobi-Mombasa fibre corridor are architecturally positioned to serve as the primary on-ramp for this increased capacity. ## Expansion Plans and Outlook ADC has signalled continued investment in its Kenyan operations. The company's broader African expansion strategy involves both growing capacity at existing sites and developing new facilities in underserved markets. In Kenya, this is likely to mean additional capacity in Nairobi to meet growing demand, particularly as cloud adoption accelerates and as the competitive pressure from iXAfrica pushes all operators to upgrade their offerings. The company's expansion is supported by its access to capital through the Cassava Technologies group. In 2024 and 2025, Cassava announced significant capital expenditure commitments to grow its data centre platform, funded through a combination of equity from its shareholders and debt facilities from development finance institutions and commercial banks. This financial capacity gives ADC the resources to pursue expansion at a pace that most competitors would struggle to match. Looking ahead, several factors will shape ADC's trajectory in Kenya. The entry of hyperscale cloud providers will create both opportunity and competitive pressure. The growth of AI workloads will demand facility upgrades to support higher rack densities. The maturation of Kenya's regulatory environment, including data protection and data localisation requirements, will drive enterprise demand for in-country colocation. And the ongoing evolution of the submarine cable landscape will increase the strategic value of Kenya's coastal connectivity. ADC enters this period from a position of strength. It has the largest continental footprint, an established Kenyan customer base, the backing of a well-capitalised parent company, and the synergies that come from being part of Africa's largest integrated technology platform. The competitive challenge from iXAfrica and others will push the company to invest in facility modernisation and higher-density capabilities, but its scale and ecosystem advantages provide a durable foundation. For Kenya's digital economy, ADC's continued investment is unambiguously positive. More capacity, more connectivity options, and more competition ultimately serve the enterprises, governments, and citizens who depend on digital infrastructure. Whether you are evaluating colocation providers for your organisation or tracking the development of East Africa's digital ecosystem, Africa Data Centres is an operator that will remain central to the conversation for years to come. Explore all operators and facilities in our [data centre directory](/directory), learn the basics with our guide to [colocation in Kenya](/articles/what-is-colocation-kenya), or read about the international connectivity that makes Kenya's data centres strategically valuable in our overview of [submarine cables in Mombasa](/articles/submarine-cables-landing-mombasa). Weighing ADC against another pan-African operator? Our fact-checked [Raxio vs Africa Data Centres in Kenya](/articles/raxio-vs-africa-data-centres-kenya-comparison) comparison separates what is open from what is only announced. ## Africa's Fibre Is Built: Now Comes the Hard Part URL: https://data-centers-254.vercel.app/articles/africa-fibre-commercial-operating-model Cluster: Connectivity Published: 2026-09-05 Updated: 2026-09-17 Summary: WIOCC's $300m raise, Kenya's 2.1m fixed lines: the hard part of Africa's fibre boom is what comes after the build, from prepaid plans to wholesale billing. Africa's fibre build-out has entered a new phase, and the industry's own analysts have noticed. A 4 September 2026 open-access analysis by TelcoTitans (Neal Doran and Alex Birkhead, drawing on conversations with European BSS specialist ZIRA Group) argues that the continent's fibre boom is becoming "a test of commercial execution as much as network construction." Network reach, they contend, is only the first step; the harder task is turning capacity into services that households, businesses and service-provider partners can discover, afford, order and use. For anyone tracking East Africa's digital infrastructure (where every fibre route ultimately terminates in an exchange, a mobile backhaul link or a data centre hall) the argument deserves a close read, because it reframes where the next wave of returns (and failures) will come from. The thesis lands at a peculiar moment: the capital is arriving faster than the operating models. Within a single week at the end of August and early September 2026, Vivica Group's South African fibre units secured R14.4 billion in funding, and regional wholesaler WIOCC Group landed a $300 million investment from Africa Finance Corporation and Saudi Arabia's Vision Invest. Money, in other words, is not the constraint. The constraint is what happens after the trenching crews leave. ## The Money Is Arriving Faster Than the Business Models The deal flow tells the story on its own. On 1 September, WIOCC Group confirmed a $300 million facility from Africa Finance Corporation and Vision Invest earmarked for exactly the kind of system-level build this publication tracks: carrier-neutral data centres, open-access terrestrial fibre and subsea infrastructure. The investment is explicitly about accelerating data centre deployment and consolidation alongside fibre expansion, evidence that the continent's smartest wholesalers no longer see fibre, towers and data centres as separate businesses but as one connected digital system. As our coverage of [the submarine cables landing at Mombasa](/articles/submarine-cables-landing-mombasa) shows, capacity at the coast only matters when inland infrastructure can absorb it. ![Technician joining aerial fibre cables on a utility pole from a bucket truck](/images/aerial-fibre-technician-utility-pole.webp) WIOCC is not alone in thinking in systems. MTN has spent recent years extending its portfolio across fibre, towers, backbone, data centres, cloud and fintech, while new subsea systems, 2Africa most prominent among them, keep raising international capacity at the landing points. The Broadband Commission and ITU's State of Broadband in Africa 2025 projects that operators across sub-Saharan Africa will invest some $62 billion in connectivity infrastructure between 2023 and 2030, and current disclosures from Airtel Africa, MTN, Orange and Vodacom imply a combined annual capex run-rate of roughly $6 billion from the continent's largest groups. ![Map of submarine cables serving Africa](/images/submarine-cables-map.webp) The macro prize is real. GSMA's Mobile Economy Africa 2026 report (published in June 2026) calculates that mobile technologies and services contributed $240 billion to Africa's economy in 2025, about 7.8 percent of regional GDP, on a forecast path to $290 billion by 2030. But the same report identifies usage, rather than coverage, as the binding constraint in many countries. The Africa Finance Corporation makes a complementary argument: physical infrastructure needs complementary cloud capability, digital platforms and enterprise services if access is to generate real economic activity, pointing to Ghana, where it estimates digitally delivered-services exports already account for 6.3 percent of GDP, broadly comparable with India and the Philippines. Translation: pipes alone do not produce returns. Services running over pipes do. ## What the Numbers Say: Fastest-Growing, Least-Connected Fixed broadband is where the usage gap is widest, and where the growth is fastest. Broadband researcher Point Topic places Africa, and more recently the combined Middle East & Africa category, among the lowest-penetration but fastest-growing fixed-broadband regions in the world. Its Q2 2025 country growth ranking put South Africa first, with nine further African markets in the top twenty, and the firm forecasts 39.4 percent growth in fixed-broadband subscriptions across Middle East & Africa between 2023 and 2030, the fastest increase of any region. Algeria, Egypt and South Africa feature among the continent's strongest FTTH/B growth markets. Low penetration is not a weakness in this story; it is the headroom. When a region starts from a small base and grows fastest, the operators who win are not those with the most kilometres of fibre but those who convert coverage into paid subscriptions efficiently. That is a marketing question, a product question, a payments question, and, less glamorously, an operations-and-systems question. ## South Africa: Prepaid Fibre and the 120-ISP Market South Africa is the continent's live laboratory for fibre business-model innovation, and TelcoTitans' numbers are specific. About six million of the country's 22 million residential and business premises have been passed by fibre, with take-up estimated at around 40 percent. Those are good but not comfortable numbers: two of every five passed homes pay, which means the industry's growth now depends more on converting passed premises than on passing new ones. The most instructive deal is Vivica Group's. Its structurally separated fibreco Frogfoot Networks and ISP Vox Telecom (which includes the prepaid-fibre brand Hypa) secured R14.4 billion (about $900 million) in August 2026, structured as R8.4 billion in equity plus R6 billion in debt. The stated plan is to quadruple the annual build rate from roughly 80,000 to 360,000 homes passed per year, focused deliberately on underserved, lower-income townships. The strategic logic is double-edged: township expansion only pencils if the commercial model fits township cashflows. Monthly debit-order contracts assume salary regularity that many households cannot offer, so day-passes, weekly vouchers and prepaid top-ups become the growth engine, and those demand flexible product configuration, distribution through local agents, and payment systems that behave more like airtime than like European broadband. South Africa's market structure sharpens the lesson further. Frogfoot alone serves 120 ISP customers through wholesale FTTH and interconnect services, competing alongside giants like Vumatel/MAZIV (now in a joint venture with Vodacom), Telkom's Openserve and MetroFibre. In a market that crowded, coverage stops being the differentiator; pricing, partner onboarding speed and service delivery decide who wins each suburb. As ZIRA Group puts it, prepaid fibre will not become Africa's universal model, but product design, payment options, distribution and margin management must reflect local demand and cashflow conditions, and the systems underneath have to be flexible enough to support that innovation. ## Kenya's Lesson: Products That Match How People Pay Kenya, this publication's home market, offers the East African counterpoint, and it validates the same thesis from a different direction. The Communications Authority counted about 2.14 million fixed-internet subscriptions by June 2025, led by Safaricom with roughly 678,000 lines (about 36.5 percent market share), followed by Jamii Telecommunications and Wananchi Group's Zuku. TelcoTitans' analysis, drawing on Point Topic data, places the fibre-to-the-home subset at around 1.2 million lines by the same date, the remainder rides on fixed-wireless access, an important nuance in a market where mobile-led consumer behaviour shapes expectations for everything. Three Kenyan realities make the commercial-layer argument concrete. First, M-Pesa: digital payments are so embedded that fibre billing has never had the cash-collection friction seen elsewhere on the continent, but the flip side is that customers expect the same instant, self-service, top-up-as-needed flexibility they get from mobile money bundles. Second, fixed-wireless and [5G networks are themselves lifting data centre demand in Kenya](/articles/5g-networks-data-centre-demand-kenya), blurring the line between fixed and mobile products and forcing operators to design across both. Third, the market's growth engine is experience, not coverage: in Nairobi's fibre-rich neighbourhoods a household chooses between Safaricom, Faiba, Zuku and others on price, reliability and support, which is precisely the multi-ISP dynamic South Africa's wholesale market formalises. ![AccessKenya Fibre advertising billboard on a Nairobi street](/images/fibre-billboard-nairobi-street.webp) The timing of the TelcoTitans piece is not accidental either. ZIRA Group will be outlining its modular commercialisation platform at ITW Africa in Nairobi on 7–10 September, a signal that European BSS vendors now see East Africa's fibre operators as a priority market, and that the commercial-systems conversation has arrived in Kenya. Local operators arguably got here first: Kenya's ISPs have been improvising flexible commercial models since the M-Pesa era began. What the analysis adds is the warning that improvisation stops scaling, at 120 ISP partners or 2 million subscribers, spreadsheets and manual onboarding become the bottleneck. ## The Wholesale Turn: Open Access as a Systems Problem Underneath both markets sits the same structural shift: the separation of network builders from retail sellers. Wholesale and open-access models improve capital efficiency, reduce duplicate trenches, and widen consumer choice, the fibreco builds and operates, ISPs and enterprise providers distribute and serve. [How fibre optic networks feed Kenya's data centres](/articles/fibre-optic-networks-kenya-data-centres) is part of the same logic: infrastructure layered on infrastructure, each layer selling access to the next. ![Blue fibre optic cables coiled inside an underground handhole](/images/fibre-handhole-duct-cables.webp) But this division of labour changes what operators need from their systems. A retail provider serves one customer base; an open-access fibreco must serve dozens of partners simultaneously, each expecting reliable coverage and price data, fast ordering and fulfilment workflows, service assurance, and accurate billing and settlement. ZIRA Group's framework (set out in its paper The Fiber Revolution: Beyond Infrastructure) names four requirements: end-to-end fibre processes, wholesale-partner capability, open integration, and commercial scalability. Amir Turalić, the company's Chief Product Officer, puts the cultural point bluntly: "A fibre network is not automatically a wholesale business. The commercial foundations have to be built to support growth." The transferable proof comes from an unexpected market. When Liberty Global's Virgin Media Ireland moved from coaxial cable to FTTH and XGS-PON, it launched a wholesale-fibre business that required new workflows, product catalogue data, partner access and B2B2X lead-to-cash capability, integrating workforce management, fulfilment, GIS and incident systems, with self-service and TM Forum Open API access for partners. Technical use cases went live in about a year, followed by wholesale agreements with national retail providers. The lesson for African operators is not that an Irish model can be transplanted (it cannot) but that operators moving to multi-partner models need systems purpose-built for their chosen route to market, and that building those foundations is a one-year programme, not an afterthought bolted onto a network plan. ## Why This Matters for Data Centres Readers of this publication should care about fibre business models for a direct reason: fibre take-up is a leading indicator of data centre demand. Every household that buys fibre, every ISP that plugs into an open-access network, every business that moves its workloads from mobile hotspot to fixed broadband is adding traffic that must be exchanged, peered, cached, stored and computed somewhere. Kenya's exchange point story, [why KIXP matters](/articles/kixp-internet-exchange-point-kenya), and the growth of [data centre interconnection and peering in Kenya](/articles/data-centre-interconnection-peering-kenya) are downstream of exactly the fixed-broadband expansion this analysis describes. ![SC connectors and yellow patch cords on a fibre patch panel](/images/fibre-patch-panel-sc-connectors.webp) The WIOCC deal makes the connection explicit: the same $300 million funding data centre deployment and terrestrial open-access fibre. Carrier-neutral facilities and open-access networks share one commercial philosophy (build once, serve many partners) and they rise or fall on the same operational capability: coverage and capacity data partners can trust, ordering that does not require weeks of email, and settlement that does not require an audit to reconcile. Operators and investors evaluating Africa's digital infrastructure should therefore score fibre providers and data centre operators on the same axis: not just how much network they have built, but how efficiently they can turn that network into partner-ready products. The capital has kept arriving since that analysis. On 16 September 2026, the US International Development Finance Corporation committed up to $155 million in equity to WIOCC, the largest equity investment in the agency's history (Reuters, 16 September 2026), stacking on top of the $300 million above. We unpack what that money changes for Kenyan bandwidth prices and data centre buyers in [our DFC-WIOCC explainer](/articles/us-dfc-wiocc-155m-kenya-digital-infrastructure). ## The Takeaway for Operators and Investors The next phase of African fibre will be judged by more than construction progress. Local economics, payment patterns, regulation, geography and partner structures vary widely from Lagos to Nairobi to Cape Town, and across those differences, one principle holds. The operators who win will be those who can package capacity into offers customers and partners can actually buy: prepaid where cashflow demands it, monthly where it does not; wholesale where ISPs are strong, retail where they are not; open APIs everywhere. Emir Bukvić, ZIRA Group's CEO, frames the stakes in one line: "The ability to scale the commercial side of the business as quickly as the network will be central to turning infrastructure investment into sustainable growth." For fibre providers entering their next build phase, and for the data centre operators whose demand curves depend on theirs, the message is the same: making capacity easy to package, buy, deliver and manage may now matter as much as making it available. The boom has proved the continent will fund networks. The test ahead is whether it can operate them like businesses. ## AI Data Centres in East Africa: Kenya's Position URL: https://data-centers-254.vercel.app/articles/ai-data-centres-east-africa Cluster: AI & Cloud Published: 2026-08-27 Updated: 2026-09-28 Summary: AI needs compute close to users. Kenya's submarine cables, geothermal power, and growing digital economy make it East Africa's natural AI data centre hub. ![GPU servers racked in a data centre](/images/ai-gpu-servers.webp) ## Why AI Needs a New Class of Data Centre in Africa Generative AI has fundamentally changed what data centres must deliver. The shift from serving web pages and databases to running large language models, computer vision pipelines, and inference workloads has created demand for a completely different tier of infrastructure. Standard enterprise data centres built for colocation and cloud hosting cannot meet the power density, cooling requirements, or network fabric that AI workloads demand. The short answer: Kenya is today East Africa's most credible location for that new tier of infrastructure. It combines the region's densest submarine cable landings, a grid that is roughly 90% renewable and anchored by baseload geothermal, and the region's largest digital economy. The constraint is equally clear: power delivery at hyperscale, as the stalled Microsoft and G42 build shows. For how that compute demand is being served today, see our [GPU cloud infrastructure in Kenya](/articles/gpu-cloud-infrastructure-kenya) analysis. A typical GPU server housing eight NVIDIA H100 or equivalent accelerators draws between 10 and 15 kilowatts under load. Rack densities in AI facilities routinely exceed 40-80 kW per rack, compared to 5-10 kW in a conventional data centre. This is not an incremental upgrade. It is a generational shift in facility design, power delivery, and thermal management. For African markets, this creates both a challenge and an opportunity. The challenge is that very few facilities on the continent can currently support these densities. The opportunity is that demand for AI inference at the edge, close to the 1.4 billion users on the continent, will require purpose-built GPU infrastructure. Relying on European or American data centres for every AI inference call introduces 150-250 milliseconds of round-trip latency, which is unacceptable for real-time applications in healthcare diagnostics, financial services, agricultural analytics, and language translation. East Africa, and Kenya specifically, is structurally positioned to become the primary hub for this infrastructure on the continent. The reasons are not speculative. They are grounded in energy geology, submarine cable topology, and regional market economics. ## What Makes an AI Data Centre Different Before assessing East Africa's readiness, it is important to understand the specific technical requirements that distinguish an AI data centre from a traditional facility. **Power density.** AI training and inference clusters require 5-10x the power per square metre compared to standard enterprise colocation. A 1,000-GPU facility needs 10-15 MW for compute alone, with another 30-40% needed for cooling and supporting infrastructure. Total facility power often exceeds 20 MW for a mid-scale deployment. **Cooling.** GPU clusters generate extraordinary heat. Traditional air cooling becomes ineffective above 30 kW per rack. AI facilities increasingly rely on direct-to-chip liquid cooling or immersion cooling, which require specialised piping, coolant distribution units, and heat rejection systems. This adds capital cost but dramatically improves power usage effectiveness (PUE). **Network fabric.** AI training across multiple GPUs requires high-bandwidth, low-latency interconnects such as NVIDIA InfiniBand or RoCE over Ethernet. These networks operate at 400 Gbps and above, with microsecond-level switching. The internal network of an AI data centre is as critical as its external connectivity. **Resilience.** AI training jobs can run for weeks or months. An interruption mid-training wastes millions of dollars in compute time. AI data centres require N+1 or 2N redundancy on power and cooling, with robust on-site generation and energy storage. ![Grid technician working on overhead power distribution lines](/images/grid-technician-pole-work.webp) ## The Scale of AI Compute Demand in Africa Africa's AI market is projected to exceed USD 20 billion by 2030, driven by adoption in financial services, agriculture, healthcare, and government services. Mobile money platforms like M-Pesa process over 30 million transactions daily in Kenya alone, and fraud detection, credit scoring, and customer service automation are increasingly AI-driven. Nigerian fintechs process hundreds of millions of mobile money transactions monthly. Ethiopian and Rwandan governments are deploying AI for agricultural extension services and public health monitoring. South African mining companies use computer vision for safety and efficiency. Across the continent, the common constraint is not the application layer but the compute layer. Currently, the vast majority of AI inference for African users is served from data centres in Europe (Amsterdam, London, Frankfurt) or the Middle East (Dubai, Abu Dhabi). This architecture has three critical limitations. First, latency. Round-trip times of 150-200ms degrade real-time applications. Second, cost. Bandwidth from Africa to Europe is improving but remains more expensive than intra-continental traffic. Third, data sovereignty. Several African nations are enacting or considering data localisation requirements that will mandate on-shore processing of citizen data. These factors create a clear and growing demand for AI compute capacity located within Africa, serving African users, and compliant with African data regulations. ## Why East Africa Specifically East Africa has a combination of structural advantages that no other region on the continent can match for AI data centre development. **Geothermal energy.** Kenya sits on the East African Rift, one of the world's most geothermally active zones. The Olkaria geothermal complex alone generates over 800 MW, and Kenya's total installed geothermal capacity exceeds 1,000 MW, the largest in Africa and among the top 10 globally. Geothermal power is baseload, meaning it runs 24 hours a day regardless of weather, unlike solar or wind. It is also among the cheapest sources of electricity in Africa, with generation costs of USD 0.07-0.09 per kWh. ![Geothermal steam plant in the Rift Valley](/images/kenya-geothermal-plant-3.webp) For AI data centres, which consume power continuously and care deeply about cost predictability, geothermal is the ideal energy source. A GPU cluster running a three-month training job needs guaranteed, affordable power around the clock. Geothermal provides exactly that. **Submarine cable connectivity.** Mombasa is the landing point for seven live submarine cable systems, including EASSy, Seacom, TEAMS, DARE1 and 2Africa, with Africa-1 landed and Daraja in development. This gives Kenya more international bandwidth capacity than any other country in East or Central Africa. Combined with the [submarine cables landing in Mombasa](/articles/submarine-cables-landing-mombasa), Kenya has established itself as the region's internet gateway. For AI data centres, this connectivity serves two purposes. It provides low-latency links to global AI model repositories, cloud platforms, and research networks. It also enables Kenya-based facilities to serve as regional hubs, connecting to landlocked neighbours via terrestrial fibre through Uganda, Rwanda, Burundi, eastern DRC, South Sudan, and Ethiopia. **Market position.** Kenya's tech ecosystem, often called "Silicon Savannah," is the most mature in East Africa. Nairobi hosts the African headquarters of Google, Microsoft, Amazon Web Services, and numerous global tech companies. The country has over 600 tech startups, a robust mobile money ecosystem, and a growing AI research community at institutions like the University of Nairobi, Strathmore University, and the Kenya School of Government. ![Server racks in a Nairobi data centre hall](/images/dc-gpu-cluster-2.webp) ## Kenya vs. Other African Markets for AI Infrastructure To understand Kenya's positioning, it is useful to compare it against the other two markets most frequently cited for data centre investment in Africa. | Factor | Kenya | South Africa | Nigeria | |--------|-------|-------------|---------| | **Primary energy source** | Geothermal (baseload, clean) | Coal (baseload, high emissions) | Natural gas / hydro (mixed reliability) | | **Grid reliability** | Moderate; improving with geothermal expansion | Moderate; load shedding historically common | Low; chronic generation deficit | | **Electricity cost (USD/kWh)** | 0.07-0.09 (geothermal) | 0.08-0.12 | 0.10-0.15 | | **Submarine cables** | 8+ systems in Mombasa | 6+ systems (multiple landing points) | 5+ systems (Lagos) | | **Regional market reach** | 300M+ (East & Horn of Africa) | 200M+ (Southern Africa, limited by distance) | 400M+ (West Africa, but fragmented) | | **Existing data centre capacity** | ~30 MW (growing rapidly) | ~200 MW (most mature in Africa) | ~50 MW (growing) | | **Data sovereignty momentum** | Moderate; East African Community framework developing | Advanced; POPIA enacted | Emerging; NDPR in effect | | **AI ecosystem maturity** | Growing; Microsoft, Google presence | Established; most AI startups on continent | Emerging; large developer community | | **Key risk** | Grid capacity for very large facilities | Power cost and carbon intensity | Power reliability and security | South Africa has the most developed data centre market in Africa, with established operators like Africa Data Centres, Teraco, and DCX. However, its reliance on coal-fired power creates both cost and sustainability challenges for energy-intensive AI workloads. Nigeria has the largest population and a massive developer community, but its power grid remains unreliable, forcing data centre operators to depend heavily on diesel generation, which is expensive and environmentally problematic. Kenya occupies a middle ground that is increasingly favourable for AI-specific infrastructure. It has less existing data centre capacity than South Africa, which means less legacy infrastructure to work around. Its geothermal resource is genuinely world-class and directly aligned with the 24/7 high-power demands of AI compute. ![Steam plumes over a geothermal field](/images/kenya-geothermal-plant-olkaria.webp) ## The Microsoft-G42 Lesson: Power Is the Constraint In May 2024, Microsoft and UAE-based AI company G42 announced a USD 1 billion investment to build a state-of-the-art AI data centre in Kenya, as part of a broader partnership that included [Microsoft and G42's Kenya AI initiative](https://www.microsoft.com/en-us/ai). The announcement was widely celebrated as a landmark moment for African technology infrastructure. We track the partnership's Kenya data centre plans, and what stalled them, in a separate [Microsoft and G42 analysis](/articles/microsoft-g42-kenya-data-centre). By May 2026, the project had stalled. Reporting from multiple sources indicated that Kenya's national grid, operated by Kenya Power and Lighting Company (KPLC), could not reliably deliver the several hundred megawatts the facility required. The [Kenya power and data centres](/articles/kenya-power-infrastructure-data-centres) dynamic is the single most important factor determining whether large-scale AI infrastructure can be deployed in the country. This outcome is not a failure of vision or of Kenya's potential. It is a specific, solvable infrastructure gap. The national grid's total installed capacity is approximately 3,500 MW, with peak demand around 2,200 MW. Allocating 200-500 MW to a single data centre customer, while maintaining supply to existing industrial, commercial, and residential users, requires deliberate grid expansion and potentially dedicated transmission infrastructure. The lesson is clear: AI data centres in Kenya will not be built at hyperscale until power delivery is guaranteed. The geothermal resource exists. The demand exists. The connectivity exists. What must catch up is the grid's ability to deliver that power to a specific facility with the reliability and redundancy that a USD 1 billion investment requires. ## What Needs to Happen Next For Kenya to actualise its position as East Africa's AI data centre hub, four things must happen in parallel. **Dedicated power infrastructure.** Large AI data centre operators will not rely solely on the national grid. The model that works is a hybrid approach: dedicated geothermal generation capacity, either owned by the data centre operator or through a power purchase agreement (PPA) with a geothermal producer like KenGen or an independent power producer (IPP), supplemented by grid connection for redundancy. The [Google AI in Africa initiative](https://www.google.com/africa/) has signalled interest in supporting renewable energy infrastructure on the continent, and similar corporate PPAs could underwrite new geothermal development. **Regulatory streamlining.** The [Kenya data centre licensing](/articles/kenya-data-centre-licensing-framework) framework must evolve to accommodate the unique characteristics of AI facilities, which may not fit neatly into existing categories for telecommunications facilities or commercial buildings. Fast-tracking approvals for facilities that meet specific power, connectivity, and security standards would reduce the time from investment decision to operational deployment. **Talent development.** Kenya's universities and technical institutions need to develop specialised programmes in high-performance computing infrastructure, GPU cluster management, AI operations (MLOps), and data centre facility engineering. Partnerships between global operators and local institutions, modelled on the Microsoft ADC programme or Google's Career Certificates, could build the workforce needed to operate these facilities. **Regional fibre expansion.** Kenya's terrestrial fibre backbone to Uganda, Rwanda, and beyond needs continued investment to ensure that AI inference served from Nairobi reaches users across the East African Community with sub-30ms latency. The Eastern Africa Backbone System (EABS) and other regional initiatives are progressing but require accelerated deployment to match the pace of AI adoption. ## The Bottom Line AI is not a trend that will pass. It is a fundamental restructuring of how software is built, deployed, and consumed. Every industry in Africa, from mobile banking to agriculture to government services, will increasingly depend on AI capabilities that require local compute infrastructure. The question is not whether AI data centres will be built in East Africa, but where and when. Kenya has the strongest structural case: geothermal baseload power, unmatched submarine cable connectivity in the region, a growing tech ecosystem, and a strategic geographic position. The Microsoft-G42 experience shows that power delivery, not power availability, is the binding constraint. Solving that problem, through dedicated generation capacity and grid investment, will unlock billions of dollars in AI infrastructure investment. The window is open. Kenya's competitors are not standing still. South Africa is leveraging its first-mover advantage, and Nigeria is aggressively courting data centre investment. But neither has Kenya's combination of clean baseload energy and strategic connectivity. If Kenya's power infrastructure can close the gap in the next three to five years, the country will not just participate in Africa's AI infrastructure buildout. It will lead it. ## AI Ethics and Governance for Data Centres in Kenya URL: https://data-centers-254.vercel.app/articles/ai-ethics-governance-kenya Cluster: AI & Machine Learning Published: 2026-08-28 Updated: 2026-08-28 Summary: Kenya's AI governance landscape is evolving fast. What the regulatory developments and sector-specific concerns mean for data centre operators. ![Wind turbines on a green Kenyan hill](/images/kenya-wind-turbines.webp) ## Why AI Ethics and Governance Matter Now Artificial intelligence is no longer a future technology in Kenya, it is an operational reality. Banks are using AI for credit scoring, hospitals are deploying AI-assisted diagnostics, telecoms operators are optimising networks with machine learning, and government agencies are exploring AI for citizen service delivery. With this rapid deployment comes a corresponding need for **ethical governance frameworks** that ensure AI systems are fair, transparent, accountable, and aligned with Kenyan values and legal requirements. The urgency is not abstract. In 2024 and 2025, several high-profile AI governance failures made global headlines, biased lending algorithms, AI-generated misinformation, and privacy breaches through AI systems processing personal data without adequate safeguards. Kenya is not immune to these risks. In fact, the application of AI systems trained on non-African data to Kenyan populations introduces unique bias risks that require locally informed governance approaches. This article examines the global AI ethics landscape as it applies to Kenya, the country's evolving regulatory framework, sector-specific concerns, international standards being adopted locally, and what all of this means for data centre operators and technology companies. ## The Global AI Ethics Landscape and Kenya's Position ### Bias in AI Systems: The African Data Problem The most fundamental ethical challenge for AI in Kenya is **data representation**. The large language models and machine learning systems that organisations deploy are overwhelmingly trained on datasets from North America, Europe, and parts of Asia. African data, cultures, languages, and socioeconomic contexts are significantly underrepresented. This gap has concrete, harmful consequences: - **Facial recognition systems** trained primarily on lighter-skinned faces show significantly higher error rates for darker-skinned individuals, a well-documented problem that directly affects Kenyans. - **Natural language processing models** perform poorly on Kiswahili, Sheng, and Kenya's many vernacular languages, leading to misinterpretation of user intent. - **Credit scoring algorithms** trained on Western financial behaviour patterns may penalise Kenyans for perfectly normal local financial practices, such as relying on mobile money, informal savings groups (chamas), or seasonal income. - **Healthcare AI** trained on European or American patient populations may produce inaccurate risk assessments for Kenyan patients with different genetic profiles, disease prevalence, and healthcare access patterns. Addressing these biases requires not just technical solutions (better training data, model fine-tuning) but **governance structures** that identify, assess, and mitigate bias before AI systems reach production. ### Privacy and Personal Data Processing AI systems are voracious consumers of data. Large language models, recommendation engines, and predictive analytics tools process vast quantities of personal information. In Kenya, this intersects directly with the [Kenya Data Protection Act and data centres](/articles/kenya-data-protection-act-data-centres), which establishes principles of data minimisation, purpose limitation, and lawful processing. The tension between AI's appetite for data and data protection principles is one of the central governance challenges of our time. An AI model that delivers better results with more data is simultaneously a model that may violate data protection law by collecting more personal information than necessary for its stated purpose. ### Transparency and Explainability When an AI system makes a decision that affects a person's life (denying a loan, flagging a transaction as fraudulent, or recommending a medical treatment) that person has a legitimate interest in understanding *why*. The principle of **algorithmic transparency** requires that AI systems be explainable in terms that affected individuals and oversight bodies can understand. This is technically challenging. Deep neural networks, the architecture behind most modern AI systems, are inherently opaque. The field of explainable AI (XAI) is developing tools to address this, but Kenya's regulatory framework does not yet specify what level of explainability is required for different types of AI-assisted decisions. ![Biometric access control at a data centre](/images/dc-biometric-access.webp) ## Kenya's Evolving Regulatory Framework ### The Kenya Data Protection Act 2019 The DPA remains Kenya's most significant piece of legislation relevant to AI governance, even though it was enacted before the current generative AI wave. Several provisions have direct AI implications: - **Section 25 (Automated Decision-Making)**: The DPA states that a data subject has the right not to be subject to a decision based solely on automated processing that produces legal effects or similarly significant effects. This provision directly applies to AI-driven credit scoring, insurance underwriting, and employment screening systems. Organisations must provide meaningful information about the logic involved and the significance and envisaged consequences of such processing. - **Section 48 (Cross-Border Data Transfers)**: Sending data to foreign AI APIs triggers cross-border transfer requirements, including adequacy assessments and appropriate safeguards. The Office of the Data Protection Commissioner (ODPC) has been increasingly active in enforcing these provisions. - **Data Protection Impact Assessments (DPIAs)**: The ODPC requires DPIAs for high-risk processing activities. AI systems processing personal data at scale (particularly in sensitive sectors) will typically trigger DPIA requirements. ### The Proposed National AI Strategy Kenya has been developing a national AI strategy, drawing on input from the ICT Authority, the Ministry of Information, Communications and the Digital Economy, academia, and the private sector. While the final strategy has not been formally enacted as of mid-2026, key themes from consultation drafts include: - Establishing a **national AI governance body** or designating an existing institution to coordinate AI policy - Developing **sector-specific AI guidelines** for high-risk applications in banking, healthcare, and government services - Creating **AI ethics requirements** for public sector AI procurement - Investing in **AI governance capacity building** through training and institutional development - Promoting **Kenyan representation in global AI governance forums** The proposed strategy draws heavily on international frameworks, particularly the [UNESCO Recommendation on the Ethics of Artificial Intelligence](https://www.unesco.org/en/artificial-intelligence/recommendation-ethics), which Kenya endorsed. ### The ICT Authority's Role The ICT Authority, which falls under the Ministry of Information, Communications and the Digital Economy, is the government body most directly involved in shaping Kenya's AI governance approach. Its responsibilities include developing ICT standards, advising government on technology policy, and overseeing e-government initiatives. As AI becomes central to government service delivery, the ICT Authority's role in AI governance is expected to expand significantly. ### The Communications Authority of Kenya (CA) The CA regulates Kenya's telecommunications sector, including the deployment of AI in telecommunications services. As Safaricom, Airtel Kenya, and other operators increasingly use AI for network optimisation, customer service, and fraud detection, the CA's oversight of AI-enabled services is growing. The CA has signalled that it will develop specific guidelines for AI use in regulated telecoms services, focusing on consumer protection, transparency, and service quality. ## Sector-Specific AI Governance Concerns ### Banking and Financial Services Kenya's banking sector, regulated by the Central Bank of Kenya (CBK), is one of the most active adopters of AI. The CBK has issued guidance on the use of technology in financial services, including expectations around algorithmic lending: - **Fair lending**: Banks using AI for credit decisions must ensure that algorithms do not discriminate against protected groups. The CBK has indicated that algorithmic fairness audits may become a regulatory requirement. - **Explainability**: Borrowers have a right to understand why a loan application was approved or denied. AI systems used for credit scoring must provide explanations that meet CBK's standards. - **Model risk management**: CBK's prudential guidelines require banks to manage model risk, including AI and machine learning models. This includes model validation, ongoing monitoring, and governance over model changes. The Central Bank's approach is informed by international standards from the Basel Committee on Banking Supervision and the Financial Stability Board, which have both issued guidance on AI in financial services. ### Healthcare AI in Kenyan healthcare (from diagnostic imaging AI to clinical decision support systems) raises particularly acute ethical concerns because the stakes involve human health and life. Key governance issues include: - **Clinical validation**: AI diagnostic tools must be validated on Kenyan patient populations before deployment. A model trained on European radiology data may not perform accurately on Kenyan patients. - **Patient data protection**: Medical records are classified as sensitive personal data under the DPA, requiring the highest level of protection. AI systems processing this data must implement strict access controls, encryption, and audit logging. - **Professional accountability**: When an AI system assists or influences a clinical decision, the accountability framework must be clear. The Kenya Medical Practitioners and Dentists Council is developing guidance on the use of AI in clinical practice. ### Telecommunications Telecoms operators using AI for network optimisation, customer analytics, and fraud detection must navigate both the CA's regulatory framework and the DPA. Specific concerns include: - **Subscriber data**: AI systems analysing call patterns, location data, and usage behaviour for network optimisation must comply with purpose limitation principles. Data collected for network management cannot be repurposed for marketing or profiling without additional consent. - **AI-driven pricing and offers**: When AI determines which mobile plans or promotions are offered to specific subscribers, fairness and transparency requirements apply. - **Real-time surveillance concerns**: AI-powered network monitoring systems must be designed with safeguards against misuse for surveillance of individuals or communities. ![Nairobi skyline lit up at night](/images/nairobi-skyline-night.webp) ## International Frameworks Shaping Kenyan AI Governance ### UNESCO's Recommendation on the Ethics of AI Adopted in November 2021, the [UNESCO Recommendation on the Ethics of Artificial Intelligence](https://www.unesco.org/en/artificial-intelligence/recommendation-ethics) is the first global standard on AI ethics. Kenya endorsed the recommendation and has referenced it in developing its national approach. The UNESCO framework emphasises proportionality, do no harm, safety and security, right to privacy, transparency, explainability, responsibility, accountability, and awareness. ### The African Union's Continental AI Strategy The [African Union Continental AI Strategy](https://au.int/en/documents/20250212/continental-artificial-intelligence-strategy-africa), adopted in 2025, provides a pan-African framework for AI development and governance. It emphasises: - **Africa-owned AI development** that serves African priorities rather than importing foreign AI systems without adaptation - **Data sovereignty** as a continental priority - **Capacity building** for AI governance across AU member states - **Inclusive AI** that addresses rather than exacerbates inequality - **Regional harmonisation** of AI regulations to create a unified African approach For Kenya, the AU strategy reinforces the importance of developing local AI governance capacity rather than relying solely on imported regulatory models. ### The EU AI Act and Its Extraterritorial Reach The EU AI Act, which entered into force in August 2024 and is being phased in through 2027, is the world's most comprehensive AI regulation. Its **extraterritorial provisions** mean that it applies to providers and deployers of AI systems outside the EU when the systems' outputs are used within the EU. This has direct implications for Kenyan companies: - **Kenyan BPO companies** providing AI-enabled services to European clients must comply with the EU AI Act's requirements for the AI systems they deploy. - **Kenyan fintech companies** serving European customers or processing European residents' data must classify their AI systems according to the EU risk framework. - **Kenyan data centre operators** hosting AI systems for EU-based clients may need to demonstrate that their infrastructure supports the transparency, audit, and data governance requirements of the Act. The CBK has also signalled awareness of the EU AI Act, noting that Kenyan banks with European correspondents or subsidiaries will need to align their AI governance practices with European standards. ## What Data Centre Operators Need to Know AI governance is not only a concern for the organisations building and deploying AI models. **Data centre operators providing infrastructure for AI workloads have governance responsibilities too**, both legal and commercial. ### Infrastructure Responsibility Data centres hosting AI workloads must ensure their facilities can support the governance requirements of their customers. This is increasingly part of the [data centre security explained](/articles/data-centre-security-explained) conversation, extending beyond physical security to include: - **Audit logging and chain of custody**: The ability to demonstrate where data has been, who has accessed it, and how it has been processed. - **Data residency guarantees**: Contractual and technical assurances that data processed in a specific facility remains within Kenyan jurisdiction. - **Secure data destruction**: The ability to definitively delete data, including backup copies, when required by data subjects exercising their rights under the DPA. - **Incident response**: Clear procedures for responding to data breaches involving AI systems, including notification to affected customers and regulators. ### Supporting Customer AI Governance As AI regulation tightens, data centre customers will increasingly require their infrastructure providers to support their governance obligations. This includes providing: - **Dedicated GPU infrastructure** with appropriate security controls for AI inference workloads - **Network isolation** to ensure AI workloads are segmented from other customers' data - **Compliance certifications** that demonstrate the facility meets relevant security and governance standards - **Contractual terms** that allocate responsibility for AI governance appropriately between the data centre operator and the AI deployer ### The Environmental Dimension The environmental impact of AI compute is an emerging governance concern. Training a single large language model can emit as much carbon as five automobiles over their entire lifetimes. For Kenyan data centre operators, this creates both an ethical obligation and a commercial opportunity. Kenya's geothermal energy (which provides over 40% of the country's electricity) offers a genuinely green power source for AI compute. Data centres that can demonstrate low-carbon AI infrastructure will increasingly attract environmentally conscious customers and support their own ESG reporting obligations. ![Cutaway infographic of an AI data centre's systems](/images/whats-inside-ai-data-center.webp) ## AI Ethics Committees and Impact Assessments ### Establishing AI Ethics Oversight Leading Kenyan organisations are establishing internal AI governance structures. These range from formal **AI ethics committees** with board-level representation to smaller AI review panels within technology teams. Key functions include: - Reviewing proposed AI systems for ethical risks before deployment - Establishing guidelines for acceptable AI use within the organisation - Monitoring deployed AI systems for emerging bias, performance degradation, or unintended consequences - Advising leadership on AI governance compliance requirements - Managing stakeholder concerns about AI's impact on employees and customers ### Algorithmic Impact Assessments Algorithmic impact assessments (AIAs) are structured evaluations of an AI system's potential effects on individuals, groups, and society. Similar to environmental impact assessments, AIAs evaluate risks before a system is deployed and recommend mitigations. While not yet mandated by Kenyan law, AIAs are being adopted voluntarily by leading organisations and are expected to become a regulatory requirement as the national AI strategy is formalised. An effective AIA for a Kenyan organisation should evaluate: - **Bias and fairness**: Has the system been tested for disparate impact across Kenya's ethnic, linguistic, and socioeconomic groups? - **Privacy implications**: What personal data does the system process, and are appropriate safeguards in place? - **Transparency**: Can the system's decisions be explained to affected individuals? - **Accountability**: Is there a clear human accountability chain for the system's outputs? - **Kenyan context**: Has the system been validated for the specific cultural, linguistic, and socioeconomic conditions of its intended Kenyan users? ![Aerial view of a data centre campus with rooftop solar](/images/dc-environment-sustainability.webp) ## The Growing Demand for AI Governance Professionals in Kenya As AI governance matures from a niche concern to a mainstream business requirement, a new professional category is emerging in Kenya: the **AI governance specialist**. These professionals sit at the intersection of technology, law, ethics, and policy, and their skills are in growing demand. Typical roles include: - **AI Ethics Officers** within large corporations and government agencies - **AI Governance Consultants** advising organisations on compliance and best practices - **Algorithmic Auditors** who independently assess AI systems for bias, fairness, and compliance - **AI Policy Analysts** working with regulators, industry bodies, and international organisations - **Data Protection Officers** (already required under the DPA) expanding their remit to cover AI-specific governance Kenyan universities and professional bodies are beginning to develop programmes to prepare professionals for these roles. Strathmore University, the University of Nairobi, and the Kenya School of Law have all introduced or are developing modules covering AI governance, technology law, and digital ethics. ## Conclusion: Governance as Competitive Advantage AI ethics and governance in Kenya are not obstacles to innovation, they are **enablers of sustainable, trustworthy AI adoption**. Organisations that invest in robust AI governance today will be better positioned to deploy AI systems that Kenyan citizens trust, that regulators approve, and that deliver genuine value rather than harm. For data centre operators, the message is clear: providing infrastructure for AI workloads comes with responsibility. The facilities that can demonstrate security, compliance, data sovereignty, and support for their customers' AI governance obligations will win in a market where trust is the ultimate currency. As Kenya's national AI strategy takes shape and international frameworks like the EU AI Act extend their reach into East Africa, the organisations that have proactively built AI governance capabilities will be the ones that lead. The time to invest in AI ethics and governance is now. ## Frequently Asked Questions **Does Kenya have a specific AI law or regulation?** Kenya does not yet have a dedicated AI Act. AI governance currently operates through existing legislation, primarily the Kenya Data Protection Act 2019, which governs how AI systems process personal data. The government has been developing a national AI strategy, and the ICT Authority is expected to play a central role in coordinating AI governance. The Communications Authority of Kenya also has jurisdiction over AI-enabled telecommunications services. **How does the EU AI Act affect Kenyan companies?** The EU AI Act has extraterritorial reach, it applies to any organisation whose AI systems are placed on the EU market or whose outputs are used within the EU. Kenyan BPO companies, fintech firms serving European clients, and data centre operators hosting AI systems for EU customers may all fall within its scope. This means Kenyan tech companies need to understand EU AI risk classifications even without domestic AI legislation. **What is an AI ethics committee, and do Kenyan organisations need one?** An AI ethics committee is a governance body that reviews AI systems for ethical risks before and during deployment. While not yet legally mandated in Kenya, leading organisations including Safaricom, commercial banks, and the Microsoft ADC have established internal AI review processes. The proposed national AI strategy is expected to recommend AI ethics oversight for high-risk sectors like banking and healthcare. **Why is AI bias a particular concern in Kenya?** Most large AI models are trained predominantly on data from North America and Europe. When these models are applied to Kenyan users, they can produce biased outcomes, misidentifying African faces, misunderstanding Kenyan English or Kiswahili, or making incorrect financial risk assessments based on patterns that do not apply to the local context. This bias can lead to discriminatory lending decisions, inaccurate medical diagnoses, and exclusionary service delivery. **What role do data centres play in AI governance?** Data centres provide the infrastructure on which AI systems run, which comes with governance responsibilities. Data centre operators hosting AI workloads must ensure appropriate security controls, data residency compliance, audit logging capabilities, and the ability to support data deletion requests. As AI regulation tightens globally, data centre customers will increasingly require contractual guarantees that their infrastructure provider supports their AI governance obligations. ## AI Model Training in Kenya: Use Cases, Infrastructure URL: https://data-centers-254.vercel.app/articles/ai-model-training-kenya-use-cases Cluster: AI & Machine Learning Published: 2026-08-28 Updated: 2026-08-28 Summary: How Kenyan organisations are pursuing AI model training locally, the infrastructure gaps they face, and what GPU-ready data centres will take. ## Why AI Model Training Is Coming to Kenya Kenya's digital economy has matured rapidly, and with that maturity comes an inevitable question: can the country support not just AI *inference* but full-scale AI *model training* on its own soil? For years, Kenyan organisations have relied on cloud providers in distant regions to handle computationally intensive workloads. That dynamic is beginning to shift as data sovereignty concerns grow, bandwidth costs remain stubbornly high, and the strategic value of indigenous AI capability becomes clear. ![GPU servers racked in a data centre](/images/ai-gpu-servers.webp) Model training (the process of feeding large datasets through neural networks to adjust weights and produce a useful AI model) is fundamentally different from running inference. It demands orders of magnitude more compute, memory, storage throughput, and networking bandwidth. Understanding what this requires, and where Kenya stands in meeting those requirements, is essential for anyone tracking the [East African data centre market](/articles/ai-data-centres-east-africa). ## What Model Training Actually Requires ### GPU Clusters and Compute Density Training modern AI models (whether large language models, computer vision systems, or recommender engines) requires clusters of specialised GPUs. A single NVIDIA H100 GPU delivers roughly 4 times the training throughput of the previous A100 generation. However, no single GPU is sufficient for serious training workloads. Production training happens across clusters of 64 to 512 GPUs working in parallel, organised into configurations such as NVIDIA DGX SuperPOD. Each H100 GPU consumes approximately 700 watts under load. A modest 64-GPU cluster draws around 45 kW just for the GPUs alone, before factoring in CPUs, memory, storage, networking, and cooling. A serious training facility (the kind that would serve multiple Kenyan enterprises) would need **5 to 10 MW of dedicated power capacity**. ### High-Speed Interconnects Distributed training across multiple GPUs requires extremely low-latency, high-bandwidth interconnects. InfiniBand, specifically NVIDIA's Quantum-2 InfiniBand at 400 Gbps, is the gold standard. Ethernet alternatives like RoCE (RDMA over Converged Ethernet) can work but introduce latency that slows training. For a deeper look at why GPU computing matters for Kenya, [our dedicated analysis](/articles/gpu-computing-kenya-ai-needs-data-centres) covers the full picture. ### Storage and Data Pipelines Training a large language model might require ingesting terabytes of text data per hour. Storage systems must deliver sustained throughput of hundreds of gigabytes per second, far beyond what conventional enterprise storage provides. Parallel file systems such as Lustre or WekaIO are standard in training environments. ### Power and Cooling At H100 density levels, traditional air cooling becomes impractical. **Liquid cooling** (either direct-to-chip or immersion cooling) is essential to manage the 40-50 kW per rack thermal output that GPU training clusters generate. Kenya's relatively mild climate in Nairobi (average temperatures of 18-25°C) offers some advantage for free cooling, but liquid systems remain necessary for GPU-dense deployments. ![Technician terminating fibre optic cable](/images/fibre-splicing-termination.webp) ## Current AI Use Cases Driving Demand in Kenya ### Banking and Financial Services Kenyan banks are among the most aggressive AI adopters on the continent. **NCBA Bank** has deployed machine learning models for credit scoring that analyse mobile money transaction histories, M-Pesa statements, and alternative data sources to serve customers without formal credit histories. **KCB Group** uses similar approaches for loan default prediction and fraud detection, processing millions of transactions daily through models that must be regularly retrained on fresh data. These workloads don't yet require training foundation models from scratch, but they do involve frequent retraining of supervised learning models on large, sensitive financial datasets. The [intersection of ML and Kenyan banking](/articles/machine-learning-kenyan-banking-fintech) reveals just how much compute this sector alone could consume. ### Telecommunications **Safaricom**, Kenya's largest telecom with over 45 million subscribers, applies AI across its operations. Network optimisation models predict traffic patterns and dynamically allocate bandwidth. Customer service chatbots handle millions of queries annually. Predictive maintenance models analyse tower sensor data to anticipate equipment failures. Training and retraining these models on Safaricom's proprietary network data represents a significant and growing compute demand. ### Agricultural Supply Chains **Twiga Foods**, one of Kenya's most prominent agritech companies, uses machine learning for demand forecasting across its supply chain connecting thousands of farmers to urban markets. Their models optimise routing, pricing, and inventory management. The agricultural sector generates enormous volumes of data that could fuel much more sophisticated AI models if adequate compute infrastructure existed locally. ### Academic and Research Institutions The **University of Nairobi's School of Computing and Informatics**, **Strathmore University's @iLabAfrica**, and **Kenyatta University** all conduct AI research, but face severe GPU resource constraints. Most academic AI research in Kenya relies on limited on-premises GPUs (often a handful of NVIDIA RTX 3090 or 4090 cards) or cloud credits donated by providers like Google and AWS. This limits the scale of research and slows the development of local AI talent. The **iHub** ecosystem in Nairobi continues to incubate AI-focused startups, and initiatives like **iCompute** have experimented with shared GPU resources. However, these remain small-scale compared to what is needed. ## The Infrastructure Gap: Why Kenya Can't Train at Scale, Yet ### No Commercial GPU Cloud in Kenya As of mid-2026, **there is no commercially available GPU cloud in Kenya**. Organisations that need GPU compute for training must either: - **Use AWS Cape Town** (the closest African AWS region, offering P4d and P5 instances with NVIDIA A100 and H100 GPUs, but with cross-continent latency from Nairobi) - **Use European or US cloud regions** (higher latency, higher egress costs, and potential data residency complications) - **Deploy on-premises NVIDIA setups** (capital-intensive, requiring in-house expertise) The lack of local GPU cloud is not just an inconvenience, it is a structural impediment to Kenya's AI ambitions. According to [Microsoft's Azure infrastructure documentation](https://azure.microsoft.com/en-us/global-infrastructure/regions/), the closest Azure regions with GPU instances are in South Africa and the UAE, neither of which provides ideal latency for Kenyan users. ### Data Transfer Costs Moving a terabyte of data from Nairobi to AWS Cape Town costs approximately $80-120 in egress fees alone. Training a serious model might require moving 10-50 TB of data, translating to thousands of dollars purely in transfer costs, before any compute charges. For organisations with continuously growing datasets, these recurring costs become a significant operational burden. ### The Data Sovereignty Imperative The **Kenya Data Protection Act 2019**, overseen by the Office of the Data Protection Commissioner (ODPC), requires that personal data processing be conducted in accordance with Kenyan law. While the Act does not explicitly prohibit cross-border data transfers, it imposes conditions (including adequacy assessments and appropriate safeguards) that create compliance complexity for organisations sending Kenyan citizens' data to foreign GPU clusters. Banks and telcos, in particular, face scrutiny from the Central Bank of Kenya and the Communications Authority regarding where customer data is processed. ## Emerging Players and Investments ### Microsoft Africa Development Centre Microsoft's **Africa Development Centre (ADC)** in Nairobi, established in 2019 with a multi-million dollar investment, employs hundreds of engineers working on global products including Azure, Teams, and AI services. While the ADC does not currently operate its own GPU cluster in Nairobi, its presence signals Microsoft's commitment to the Kenyan tech ecosystem and raises expectations that Azure's regional infrastructure will eventually expand to include GPU capacity closer to East Africa. ### Google AI Lab in Accra [Google's AI research lab in Accra, Ghana](https://research.google/outreach/africa/), opened in 2018, serves the broader African continent and has produced research on applications relevant to East Africa, including agricultural AI and healthcare models. While not based in Nairobi, the lab's work demonstrates that major tech companies see Africa as a viable location for AI research investment, and Kenya's more mature tech ecosystem arguably makes it a stronger candidate for the next such investment. ### Local GPU Hosting Experiments Several Kenyan tech companies and academic consortia have begun experimenting with local GPU hosting. These range from small GPU-as-a-service offerings (renting access to shared NVIDIA servers) to discussions within the **Kenya ICT Action Network** about establishing a national AI compute resource. None has yet achieved commercial scale. ## What a Kenyan AI Data Centre Would Need Building a facility capable of supporting serious AI model training in Kenya would require: **Compute:** A minimum of 64-128 NVIDIA H100 GPUs, expandable to 256+, organised in a DGX SuperPOD or equivalent architecture. Estimated capital cost: KES 2-5 billion for the GPU hardware alone. **Networking:** NVIDIA Quantum-2 InfiniBand at 400 Gbps, with spine-leaf topology to minimise bisection bandwidth bottlenecks. **Storage:** 500 TB to 2 PB of high-performance parallel storage, delivering 400+ GB/s sustained throughput. **Power:** 5-10 MW of dedicated, redundant power with at least N+1 generator backup and UPS systems. Kenya Power's grid reliability in Nairobi has improved, but industrial-scale GPU training demands utility-grade redundancy. **Cooling:** Direct-to-chip liquid cooling or rear-door heat exchangers as a minimum, with immersion cooling as the preferred long-term solution for H100 and future GPU generations. **Connectivity:** Multiple diverse fibre paths, direct peering at KIXP, and ideally submarine cable diversity through both the EASSy and SEACOM systems landing in Mombasa. ![UPS and switchgear line-up in a data centre power room](/images/ups-switchgear-power-room.webp) ## Projecting Future Demand Kenya's digital economy is projected to contribute over 10% of GDP by 2030, up from approximately 7.5% in 2023. The fintech sector alone processes transactions worth trillions of shillings annually through M-Pesa and banking platforms. Each of these transactions generates data that could improve AI models, if the compute existed to train on it. Conservative estimates suggest Kenya will need **50-100 MW of data centre capacity by 2030**, up from approximately 30 MW today. A significant portion of that new capacity will need to be GPU-ready. If even 10% of the projected capacity is allocated to AI training, that represents 5-10 MW of GPU-dense infrastructure, comparable to a mid-sized AI cluster in a European or Asian market. The question is no longer *whether* Kenya needs AI training infrastructure, but *when* and *who* will build it. The convergence of data sovereignty requirements, growing enterprise AI adoption, and the strategic importance of indigenous AI capability makes a compelling case that the first commercial GPU cloud in East Africa will likely launch in Kenya within the next three to five years. ## Frequently Asked Questions ### Can AI model training actually happen in Kenya today? Limited model training is possible using on-premises NVIDIA GPU setups at universities and some enterprises, but there is no commercially available GPU cloud in Kenya. Organisations currently rely on AWS Cape Town (the closest African AWS region) or cloud providers in Europe and the US for serious training workloads. Smaller models and fine-tuning can run on local multi-GPU servers, but large-scale foundation model training remains offshore. ### Why would a Kenyan company train AI models locally instead of using overseas cloud? Several factors drive the push for local training. The Kenya Data Protection Act 2019 imposes obligations on personal data processing, making data sovereignty a real concern for banks and telcos handling sensitive customer data. Sending large training datasets abroad incurs significant egress costs, transferring terabytes of financial or health data to US-based GPUs can cost thousands of dollars. Additionally, latency matters for iterative ML development workflows, and local training enables faster experimentation cycles. There are also strategic arguments around building indigenous AI capabilities rather than being dependent on foreign infrastructure. ### What hardware does a Kenyan AI data centre need for serious model training? A production-grade AI training facility requires NVIDIA A100 or H100 GPUs organised in clusters of 64 to 256+ GPUs, connected via InfiniBand or NVIDIA Spectrum networking for low-latency inter-GPU communication. Storage must be high-throughput parallel file systems (such as NVIDIA DGX SuperPOD architecture) capable of feeding data at hundreds of GB/s. Power draw ranges from 5 to 10 MW for a meaningful cluster, and liquid cooling becomes necessary at H100 density levels. The facility also needs redundant fibre connectivity, ideally with direct access to the Kenya Internet Exchange Point (KIXP) for efficient data transfer. ### Which Kenyan organisations are actively using or developing AI models? NCBA Bank and KCB Group are using machine learning for credit scoring and fraud detection. Safaricom leverages AI for network optimisation and customer service chatbots. Twiga Foods applies ML models for supply chain optimisation and demand forecasting. Academic institutions (the University of Nairobi, Strathmore University, and Kenyatta University) conduct AI research, often with limited GPU resources. The iHub ecosystem and iCompute have incubated AI-focused startups, and the Microsoft Africa Development Centre in Nairobi contributes to global AI product development. ### How does Konza Technopolis fit into Kenya's AI infrastructure plans? Konza Technopolis, a key Vision 2030 project, has designated land and planning for data centre development. While its primary data centre component has not yet materialised at scale, Konza represents the government's long-term vision for a technology city that could host the kind of high-power, high-connectivity facilities that AI training demands. If a GPU cloud provider were to establish a presence in Kenya, Konza would be a logical location given its planned power infrastructure and fibre connectivity. ## Amaco's Mombasa AI Data Centre: A Floating Power Plant URL: https://data-centers-254.vercel.app/articles/amaco-hercules-mombasa-ai-data-centre Cluster: AI & Infrastructure Published: 2026-09-15 Updated: 2026-09-17 Summary: A Greek energy firm plans a $1.5 billion AI data centre in Mombasa with its own floating power plant. What is confirmed, what is not, and why it matters. One of the biggest numbers ever attached to Kenya's technology sector arrived from an unexpected direction: Greece. In August 2026, Amaco Energy Group, a Greek multinational energy and infrastructure firm, proposed a $1.5 billion (about Sh194.2 billion) artificial intelligence data centre for Mombasa (Business Daily, 17 August 2026). No data centre proposal in Kenya has carried a bigger price tag, and if it is ever built, it would do something no Kenyan facility has done before: make its own electricity from a floating power plant instead of buying it from the national grid. The story has moved quickly from idea to paperwork. In September 2026, Amaco announced a cooperation agreement with GE Vernova, the world's largest manufacturer of large gas turbines, to supply equipment for the project's power barge (Business Daily, 10 September 2026). The company's founder and chief executive, Theodore Theodoropoulos, has met officials in Nairobi, including ICT Cabinet Secretary William Kabogo. But almost everything that decides whether this project becomes real, the financing, the customers, the final size, the approvals, has not been settled. So this article does what DC254 always does: it collects what is confirmed, explains the technology in plain language, and separates the parts of the story that are solid from the parts that are still a pitch. ![Container yard and blue gantry cranes at the Port of Mombasa](/images/amaco-mombasa-port-terminal.webp) ## What is actually proposed Amaco calls the project HERCULES, short for Molecules-to-Electrons Offshore AI Smart Power and Data Center (Construction Review Online, 10 September 2026). The name is a mouthful, but the idea is simple. A normal data centre buys electricity from the grid, then spends a large share of it on cooling its servers. HERCULES bundles the three pieces, power generation, cooling, and computing, into one design that sits near the Port of Mombasa and runs independently of Kenya's grid. The computing part would be an AI data centre: a facility packed with servers arranged to train and run AI models. AI facilities are power-hungry in a way ordinary data centres are not. A conventional facility might draw a few megawatts, while a large AI campus can draw 100 MW or more, roughly the electricity used by 100,000 households (TechCabal, 20 August 2026). Globally, data centres consumed about 415 terawatt-hours of electricity in 2024, a figure the IEA expects to approach 945 terawatt-hours by 2030, with AI the main driver. At that scale, where the power comes from stops being a detail and becomes the design decision, which is exactly the problem Amaco is proposing to solve. **The power barge, explained in plain terms.** The heart of the proposal is a power barge, an unmotorised floating platform that carries a power plant on its hull, the way a cargo barge carries goods (Business Daily, 10 September 2026). Power barges are not science fiction; they are used around the world where building a plant on land is slow or complicated. The plant is assembled in a shipyard, floated to its destination, connected, and switched on. ![LNG carrier ship with spherical storage tanks at sea](/images/amaco-lng-carrier.webp) In Amaco's plan, the barge burns liquefied natural gas (LNG, which is natural gas cooled until it turns into a liquid that ships can carry) to generate electricity. That electricity feeds the data centre directly, and the same integrated system handles cooling, with heat and cold recovered and reused inside the platform rather than vented (w.media, 21 August 2026). Because the package makes its own power, it does not depend on the grid and does not compete with Kenyan homes and hospitals for electricity. Amaco also says the design can later add renewable energy and hydrogen, though it starts on gas (TechCabal, 20 August 2026). ## Who is Amaco Energy Group? Amaco Energy Group describes itself as a Greek multinational energy and infrastructure firm, led by founder and chief executive Theodore Theodoropoulos, who has personally led the Kenya negotiations (w.media, 21 August 2026). The company maintains offices in Greece and the United States, and its Kenyan project vehicle is registered as AMACO Ltd (Kenya), according to its own project pages (fetched 15 September 2026). The Mombasa proposal did not appear from nowhere. The company registered an LNG-to-power project in Kenya in 2024, with the reference code ES00101/AMACO/2024, covering LNG delivery, storage and regasification (turning liquid gas back into gas) at Kwale County near the Mombasa port. The company lists a regasification capacity of 10.5 billion cubic metres (about 6.2 million tonnes) per year, a target start year of 2027, and a current status of permitting phase (AMACO project page). The data centre announcement is effectively step two of that energy plan: the gas infrastructure now has a computing customer attached. ## What the GE Vernova agreement changes In September 2026 the project gained its most credible ingredient so far. Amaco announced a cooperation agreement under which GE Vernova, the US energy equipment company descended from General Electric, will supply gas turbines for the Hercules power barge, with electrification and digital technologies possibly following as the platform develops (Business Daily, 10 September 2026). In the company's own words: "Amaco-Hercules has entered into a cooperation agreement with General Electric that will see it integrate GE Vernova gas turbines into the Hercules power barge facility." Why does this matter? Because turbine supply is one of the hardest bottlenecks in the global power business right now, and GE Vernova is the industry's biggest player, with more than 7,000 large turbines totalling over 800 GW of capacity across more than 120 countries (Business Daily, 10 September 2026). A vendor of that size engaging publicly signals the project has moved beyond brochure stage. To keep it honest: a cooperation agreement is still not a hardware order with delivery dates, both sides describe integration "as the platform develops". But it is more than most concept proposals ever secure. ![Landing crew guiding a submarine cable ashore on a Kenyan beach](/images/amaco-cable-landing-beach.webp) ## Why Mombasa, and why off-grid Kenya's data centres cluster in Nairobi because the customers are there; our analysis of [why Kenya's data centres cluster in Nairobi and Mombasa](/articles/why-data-centres-cluster-nairobi-mombasa) explains the three forces behind that map. Mombasa has always offered a different set of inputs. Seven live submarine cable systems make landfall on its coast (SEACOM, TEAMS, EASSy, LION2, DARE1, PEACE and 2Africa), with Africa-1 landed and Meta's Daraja in development, and [the landing stations themselves](/articles/submarine-cables-landing-mombasa) already anchor facilities like [iColo MBA1](/directory/icolo-mba1), the coast's busiest interconnection point. The port matters twice for this project: it is where LNG fuel can be delivered, and it is where the cables come ashore. The candidate sites are Dongo Kundu and Kilindini, next to the port and the 3,000-acre Dongo Kundu Special Economic Zone, where industrial park developer ARISE operates (Construction Review Online, 10 September 2026). Amaco points to a demand pipeline of roughly 75 MW of potential base-load demand from heavy industry, plus another 75 to 100 MW potentially from initial data centre customers (Business Daily, 10 September 2026). In other words, the project bets on the port economy and the SEZ rather than on Nairobi's banks and telcos. The off-grid choice addresses the constraint that has capped every big Kenyan data centre plan so far: electricity. Kenya's record peak demand was 2,439 MW, reached in December 2025, so a single 100 MW AI campus would add demand equal to about 4 percent of the national peak (TechCabal, 20 August 2026). The cautionary tale is the Microsoft-G42 project near Olkaria in Nakuru County, announced in May 2024 at about $1 billion, which has stalled partly over a power request that grew from 60 MW to 1,000 MW, roughly a third of Kenya's total installed generation (Kenyans.co.ke, 22 August 2026; Business Daily, 10 September 2026). Amaco's answer is to sidestep that fight entirely: bring your own power plant. We unpack the generation-versus-reliability logic behind moves like this, and the questions buyers should ask any facility about power, in [our power reliability explainer](/articles/kenya-power-reliability-data-centres). There is even a plan for the surplus. Amaco has said the system could contribute additional generation to Kenya's broader energy needs, and Construction Review Online reports a potential export path to the local grid near the Mariakani substation (Kenyans.co.ke, 22 August 2026; Construction Review Online, 10 September 2026). That would make the project unusual in a second way: a data centre that adds electricity to the grid instead of straining it. ![Floating power barge with gas turbine exhaust stacks moored near a coastline](/images/amaco-power-barge.webp) ## What is confirmed, and what is not **Confirmed.** The $1.5 billion (Sh194.2 billion) proposal for Mombasa, first reported on 17 August 2026 (Business Daily). The GE Vernova cooperation agreement for the power barge, announced 10 September 2026 (Business Daily). Dongo Kundu and Kilindini as candidate sites, citing port proximity and Special Economic Zone demand (Business Daily, 10 September 2026). Government engagement, including CEO Theodore Theodoropoulos's meetings in Nairobi and his August meeting with ICT Cabinet Secretary William Kabogo (Business Daily, 10 September 2026). The underlying 2024 LNG-to-power registration at Kwale County, in permitting phase with a 2027 target start year (AMACO project page). **Reported, not confirmed by Amaco.** The 28-month construction estimate after approval (Construction Review Online, 10 September 2026) and the surplus-export plan via Mariakani (Construction Review Online; Kenyans.co.ke, 22 August 2026). **Not disclosed at all.** The data centre's final capacity in megawatts, the financing behind the $1.5 billion figure, the customers, the fuel supply contracts, the water arrangements, and the environmental assessment for siting an LNG-fired plant next to a working port. Amaco has not published a construction timeline for the data centre itself (Business Daily, 10 September 2026). One housekeeping note: as of 15 September 2026, DC254 has not added this project to its directory count of 27 tracked facilities. A proposal without approval, financing, or a confirmed site does not yet meet the directory's verification bar, whatever the size of its headline number. [Kenya's data centre market in numbers](/articles/kenya-data-centre-market-numbers) tracks what has actually been built and disclosed, including the roughly 10.5 MW of operational IT load published by the eight facilities that disclose figures. ## The open questions **The money.** A $1.5 billion project needs financing, and no lenders or equity partners have been named. TechCabal's caution from 20 August 2026 is worth keeping in view: Africa has no shortage of ambitious infrastructure proposals that never progressed beyond the announcement. **The size.** Amaco's spokesperson described "one of the world's largest AI data-centre facilities" without a megawatt figure (w.media, 21 August 2026). Every comparison in this article, and every capacity planning decision by the grid operator or the county, waits on that number. **The fuel.** LNG must be imported, and gas in a country whose electricity is mostly renewable (Kenya runs about 1 GW of geothermal, one of the world's greenest grids) invites a cost and carbon debate. The IEA estimates gas supplies about 26 percent of data centre electricity globally, but in Kenya the LNG system has to prove an advantage on cost and reliability over the alternatives (TechCabal, 20 August 2026). **The water.** Mombasa has long struggled with water shortages and relies on supplies from outside the county. Amaco says the integrated system avoids pressuring local water resources, a consequential claim in that city, and one that should be established clearly before any approval (TechCabal, 20 August 2026). **The approvals.** The project is in the permitting phase, and the timing is notable: Kenya is mid-transition in how it regulates data centres, with the Communications Authority having placed them under the telecommunications licensing regime and now consulting on a [standalone data centre licence](/articles/kenya-ca-standalone-data-centre-licence) (Business Daily, 10 September 2026). ![Server racks inside an AI data centre hall](/images/whats-inside-ai-data-center-wide.webp) ## What to watch next Big proposals announce themselves loudly and then go quiet, so the useful discipline is to watch for observable milestones. Each of these would move the project from pitch to plan: 1. **A formal approval or environmental licence** from the relevant Kenyan authorities, which is the gate everything else sits behind. 2. **A firm turbine order**, meaning a hardware contract with delivery dates rather than a cooperation agreement. 3. **A berthing or anchoring arrangement** with the Kenya Ports Authority for the barge within the Port of Mombasa ecosystem. 4. **A disclosed capacity figure** and ideally a named anchor customer for the computing capacity. 5. **Financing**, in the form of named lenders, equity partners, or a final investment decision. 6. **Offtake interest** from the Dongo Kundu SEZ's industrial tenants, the demand pool Amaco says justifies the site. ![Artist's impression of the Dongo Kundu Special Economic Zone near Mombasa](/images/amaco-dongo-kundu-sez.webp) If those milestones start landing, Mombasa's long-standing role as a cable and transit stop could grow into something new: the first credible attempt at a third force in Kenya's data centre map, built on self-supplied power and port demand rather than the grid. That is exactly why we flagged the proposal when it first appeared, and why this article will be updated as the facts change. Until then, the honest label is the one we gave it in August: a serious test of whether the coast can host compute at scale, and no more than that yet. ## Frequently asked questions ### What is the Amaco Mombasa data centre project? It is a proposed $1.5 billion (Sh194.2 billion) artificial intelligence data centre in Mombasa, announced by Greek energy firm Amaco Energy Group in August 2026 and developed under the name Project HERCULES. Its defining feature is that it would generate its own electricity from an LNG-fired floating power barge instead of buying power from Kenya's national grid. The project is at proposal stage: approvals, financing, customers and a final capacity figure have not been announced. ### Who is Amaco Energy Group? Amaco Energy Group is a Greek multinational energy and infrastructure firm led by founder and chief executive Theodore Theodoropoulos, with offices in Greece and the United States. Its Kenyan project vehicle is AMACO Ltd (Kenya). The company registered an LNG-to-power project in Kenya in 2024 (reference ES00101/AMACO/2024), covering LNG delivery, storage and regasification at Kwale County near the Mombasa port, and the data centre proposal builds on that energy plan. ### What is a power barge? A power barge is an unmotorised floating platform that carries a power plant on its hull, the way a cargo barge carries goods. It can be built in a shipyard, floated to its destination and connected to customers there, which makes it useful where building a plant on land is slow or complicated. In the Amaco plan, the barge burns liquefied natural gas to generate electricity for the data centre and its cooling systems. ### Why Mombasa and not Nairobi? Mombasa offers inputs Nairobi does not: seven live submarine cable systems make landfall on its coast (with Africa-1 landed and awaiting service), its port can deliver LNG fuel by ship, and the nearby 3,000-acre Dongo Kundu Special Economic Zone supplies industrial demand. Amaco cites a pipeline of about 75 MW of potential industrial base-load demand plus another 75 to 100 MW from initial data centre customers (Business Daily, 10 September 2026). Nairobi still holds most of Kenya's data centre demand, so the project bets on the port economy rather than the capital's enterprise base. ### How big will the Mombasa data centre be? No capacity figure has been disclosed. Amaco's spokesperson described it as "one of the world's largest AI data-centre facilities", and reported demand at the site runs 75 to 100 MW for initial data centre customers. For scale, a 100 MW AI campus would add demand equal to roughly 4 percent of Kenya's record peak electricity demand of 2,439 MW, reached in December 2025 (TechCabal, 20 August 2026). Treat any capacity number you see as unofficial until Amaco publishes one. ### When will the Amaco data centre be built? Construction has not started and no official timeline has been published. Construction Review Online reported on 10 September 2026 that, if approved, the project is expected to take around 28 months to build, and Amaco's own LNG project page lists a start year of 2027 with the project in the permitting phase. Both figures are plans, not commitments, and the project still needs regulatory approval. ### Will it use Kenya's national grid? No, the design is intentionally off-grid: the Hercules barge generates electricity for the data centre and its cooling directly. Amaco has said the system could contribute surplus power to Kenya's wider energy needs, and Construction Review Online reports a potential export path to the local grid near the Mariakani substation. That would be a bonus, not the core design. ## AmCham 2026: $600M US Pledges, Oracle Cloud, iColo $80M URL: https://data-centers-254.vercel.app/articles/amcham-2026-us-pledges-oracle-icolou-80m Cluster: Market Analysis Published: 2026-09-14 Updated: 2026-09-14 Summary: AmCham 2026 delivered $600 million (Sh77.5bn) in one-day US pledges for Kenya, headlined by Oracle's first African cloud region and iColo's $80M project. The AmCham Business Summit 2026 in Nairobi produced the largest one-day investment number Kenya has collected this year: more than $600 million (about Sh77.5 billion) in pledges and commitments from US companies, announced in a single day. AmCham Kenya board president Angela Ng'ang'a's own accounting was precise: "We did see just in one day over $600 million of pledges and commitments from US-American firms," she said, as quoted by The Star on 14 September 2026. For the general reader, the summit story is about manufacturing and trade. For this site's readers, two of the announced projects are the story: Oracle selecting Kenya for its first public cloud region in Africa, in partnership with African Data Centres, and iColo (now Digital Realty) announcing an $80 million data centre project in Nairobi. Both land directly on the infrastructure this site tracks, and both deserve the same treatment we give every pledge: context, attribution, and a verification plan. ## The Headline Number, in Context The $600 million figure needs its denominator. Kenya announced $2.9 billion across 20 deals earlier this year at the Kenya International Investment Conference, and previous AmCham summits have generated more than $2 billion in tracked commitments over their runs, per The Star. One day of $600 million is therefore a strong showing rather than a national record, and its composition matters more than its size: pledges spread across manufacturing (Coca-Cola's fresh $175 million; Mars Wrigley's $103 million production line), healthcare (SC Johnson's planned factory for up to two million malaria-related units daily; Pfizer's 15 new products), technology and data centres. ![AmCham Kenya quote card citing board president Angela Ng'ang'a on summit investments](/images/amcham-2026-quote-angela-nganga.webp) The macro backdrop is friendlier than it has been in years. Kenya-US trade in goods and services reached $3.4 billion (Sh439.3 billion) in 2025 with the balance of trade near equilibrium, Trade and Investment Cabinet Secretary Lee Kinyanjui told the summit, and AGOA's duty-free access has been extended to 31 December 2028, which removes the deadline anxiety that has hung over Kenyan exporters. President William Ruto's framing to investors was blunt in the right way: "Capital goes where there is opportunity, but it stays where there is confidence." ## Oracle's First African Cloud Region The Oracle announcement is the strategically larger of the two data-centre items. Per The Star, Oracle selected Kenya for its first public cloud region in Africa, in partnership with African Data Centres (AADC), with the move expected to expand local cloud infrastructure and strengthen Kenya's position as a regional digital hub. What a cloud region actually means, in this site's vocabulary: Oracle-operated compute and storage capacity installed inside local data centre halls, consuming local megawatts, and letting Kenyan enterprises and government keep workloads in-country instead of paying latency and data-residency penalties to Frankfurt or Mumbai. If confirmed and built, it is demand-side news for the whole market: a cloud region anchors multi-megawatt, long-tenor capacity commitments in a way that colocation deals rarely do. The verification status is honest and simple: this comes from summit coverage, and DC254 has not yet seen the announcement in Oracle's or AADC's own channels. Cloud region announcements also have a history of long gaps between selection and go-live, so the number to watch is not the press release but the commissioning date and the MW it draws. We will update this article and the directory when either appears. ## iColo's $80 Million Pledge The second announcement extends a story our readers already know. The Star reported that iColo, now operated under parent company Digital Realty, announced an $80 million data centre project in Nairobi at the summit. Three days before that report, on 7 September 2026, Digital Realty formally launched NBO2, the second facility of the iColo campus in Karen, inside a master plan of three data centres and a captive substation consuming over 20 MW at full build. ![Officials cutting the ribbon at the NBO2 data centre launch in Nairobi](/images/nbo2-launch-ribbon-cutting.webp) Read together, the sequence is coherent: the campus that just commissioned its second building is advertising its third act. But the discipline matters as much as the momentum. The $80 million is a project-level pledge quoted in summit coverage, not a published building-level cost, and it has not yet been tied by Digital Realty to a specific building, groundbreak or timeline. Our NBO2 analysis already treats campus master plans as intent rather than inventory; this pledge joins that category until a shovel moves. ## Why a Summit Matters for Data Centres It is fair to ask what a chamber-of-commerce summit has to do with MW. The connection is the demand base underneath every facility in our directory. Cloud regions pull enterprise workloads onshore. Manufacturing expansions consume local hosting and connectivity. Multinationals with Kenyan footprints (the beverage, healthcare and consumer goods companies in this batch) become cloud and colocation customers, and their data-residency and resilience requirements convert directly into rack contracts. The $600 million is not a data centre number; it is the economy that data centres serve getting bigger, which is the demand side of every capacity decision our directory tracks. The verification posture is unchanged: pledges are announced in summits, capacity is built in projects, and only commissioned MW count as supply. The directory now tracks 27 facilities in Kenya with 20 operational, and both announced items here (Oracle's region and iColo's project) enter our watchlist as pipeline, not inventory, until they publish figures that survive contact with a meter. Sources: The Star, "US-Kenya summit opens door for Sh77.5bn in potential investments" by Martin Mwita (14 September 2026), including quotes from AmCham Kenya president Angela Ng'ang'a, Cabinet Secretary Lee Kinyanjui and President William Ruto. Pledge and trade figures as reported therein. NBO2 and campus details from DC254's NBO2 launch analysis (7 September 2026) and directory verification (September 2026). ## AWS Commits $1.5bn to Africa: The Compute Gap Behind It URL: https://data-centers-254.vercel.app/articles/aws-1-5bn-africa-compute-gap-kenya Cluster: Cloud Computing Published: 2026-09-14 Updated: 2026-09-14 Summary: AWS has committed $1.5bn to African infrastructure through 2029, on top of $819m since 2018. The leapfrog analogy is appealing, and incomplete. Amazon Web Services has committed a further $1.5 billion to African infrastructure through 2029, on top of more than $819 million invested since 2018, and says it has trained over one million Africans in cloud and AI skills. Jyoti Ball, AWS's general manager for sub-Saharan Africa, delivered the numbers at the AWS Summit in Johannesburg, and iAfrica.com's report on 12 September 2026 did something trade coverage rarely does: it printed the commitment and stress-tested the argument behind it in the same article. That argument deserves the attention of anyone tracking African data centre capacity, because it is the same argument Kenyan policymakers use. This article walks through what AWS committed, why the leapfrog analogy that frames these commitments is appealing and incomplete, and what the numbers imply for a market like Kenya. ![AWS logo on the wall of a conference venue](/images/aws-logo-conference-venue.webp) ## What AWS Actually Committed The headline is the $1.5 billion through 2029. The useful details sit around it. The Cape Town cloud region now carries 154 services, which Ball used to argue that African developers are no longer working with a stripped-down regional feature set. Her framing of strategy was about direction of trade: enabling local developers to build for regional markets rather than consuming technology developed elsewhere. The comparison set is Microsoft, not Google or Oracle (for now). Microsoft has spent R20.5 billion on data centres in Johannesburg, Cape Town and Durban, with a further R5.5 billion Centurion facility planned, roughly $1.4 billion combined at current rates, per the iAfrica report. Two hyperscalers at $1.4 to $1.5 billion each is a real pattern, and it sets the scale at which serious African cloud infrastructure is being financed: billions, spread over years, concentrated in South Africa first. ## The Leapfrog Case, and Its Gap Ball's central case is the one African ICT ministers have made for two decades: the current AI cycle offers Africa another chance to bypass stages that wealthier economies passed through. Her precedent is mobile money: "Africa did not wait for conventional banking infrastructure to become widespread before developing large-scale mobile payment systems," pointing to South Africa's jump from limited mobile connectivity in the early 1990s to more than 100 million connections. iAfrica's critique of that analogy is the best part of the report, and it is analytically correct. Mobile money leapfrogged branch banking because the infrastructure it actually depended on (mobile networks) was already built. Africa did not skip telecoms; it skipped bank branches by running services over telecoms. AI's equivalent dependency is compute, and that has not been built. The report's numbers make the point stark: Africa holds roughly 1% of the world's data centres against about 15% of global population, and McKinsey projects that African capacity must rise from around 0.4 gigawatts to as much as 2.2 gigawatts by 2030, requiring $10 billion to $20 billion of investment. Read those two numbers against the commitments and the gap quantifies itself. AWS's $1.5 billion and Microsoft's $1.4 billion equivalent together cover a fraction of the McKinsey range. The leapfrog story is therefore not wrong; it is a description of where the demand is heading, funded so far at a tenth of the level its own analogy requires. Leapfrogs do not skip infrastructure. They skip incumbent business models by building the infrastructure early. That is what M-Pesa rode, and it is what any African AI story must eventually stand on: powered halls with racks in them. ![Logistics workers move a server rack through a fulfilment facility](/images/aws-server-rack-logistics-workers.webp) ## Why the Cloud Region Detail Matters for East Africa There is a geographic asymmetry inside the numbers that matters for Kenya. The region Ball cites is Cape Town; the $1.4 billion Microsoft figure is South African. Southern Africa is where hyperscaler capital has concentrated, because it has the grid, the fibre and the enterprise base today. East Africa's share of that investment wave is so far measured in partnerships and positioning rather than regions: Kenya's pitch leans on iXAfrica's hyperscale AI-ready campus, Digital Realty's freshly launched NBO2, and a power story (geothermal-backed, low-carbon, competitively priced) that is genuinely differentiated on the input costs that matter for compute. The implication is not that Kenya loses the wave. It is that Kenya wins it by making the leapfrog's dependency investable: grid capacity into industrial areas, a licensing regime that treats data centres as infrastructure (the Communications Authority's standalone licence consultation is live), and a directory-grade evidence base that lets financiers verify what exists. Hyperscalers build regions where demand, power and permitting line up; Kenya's job is the lining up. ## How to Track Commitments Like These The verification habit this site applies to every hyperscaler announcement applies here. Commitments are dated intentions; capacity is commissioned MW. The numbers to watch from this announcement are none of the headline figures: they are the eventual facility list (which halls, which countries), the MW those halls draw, and the service count trajectory of any East African region if one is announced. Against the McKinsey 2.2 GW target for 2030, four years is short, and power projects move at the speed of transmission lines. That arithmetic, not enthusiasm, is what will decide whether 2029 reads as the start of Africa's compute decade or the preamble to it. ![Labelled diagram of the power chain from grid to GPU rack](/images/diagram-power-chain-grid-to-gpu.webp) Sources: iAfrica.com, "AWS Commits $1.5bn to Africa Through 2029 and Makes the Leapfrog Case. The Analogy Has a Gap" (12 September 2026), reporting remarks by Jyoti Ball (AWS general manager, sub-Saharan Africa) at the AWS Summit in Johannesburg; AWS investment and training figures, Cape Town service count, Microsoft R20.5 billion / R5.5 billion spend, and McKinsey 0.4 to 2.2 GW / $10-20 billion projections as reported therein. Kenya facility figures from the DC254 directory (September 2026 verification). ## Bypassing the Red Sea: Berbera's Backup Route for Kenya URL: https://data-centers-254.vercel.app/articles/berbera-northern-route-kenya-fibre-redundancy Cluster: Connectivity Published: 2026-09-16 Updated: 2026-09-16 Summary: Somcable lands 2Africa and PEACE at Berbera and is building terrestrial routes into Ethiopia and South Sudan. What a northern alternative means for Kenya's cable risk. For two days in September 2026, the most important room in African connectivity was the Kenyatta International Convention Centre in Nairobi, where more than 1,800 delegates gathered for ITW and Datacloud Africa 2026 to decide where the continent's next internet capacity gets bought and built. The most interesting pitch in the building came from a company most Kenyan buyers have never had a reason to call. Somaliland's Somcable stood up as the builder of a "Northern Route": dual submarine landings at Berbera, wired overland to Ethiopia and beyond, offering the region a second gateway at the exact moment Red Sea disruptions have exposed how concentrated East Africa's international bandwidth really is. This article unpacks what Somcable actually announced, how it fits with the eight cable systems already landing at Mombasa, and what it changes for Kenyan businesses that depend on international capacity. ## What Somcable Announced The company behind the pitch is not a newcomer. Somcable was founded in 2009 and is headquartered in Hargeisa, and it operates two of the most significant submarine cable landings on the Horn: the 2Africa and PEACE systems, both terminating at the Port of Berbera. Business Daily reported on 15 September 2026 that Somcable is linking those landings to terrestrial fibre networks extending into Ethiopia, South Sudan and neighbouring countries, and that it has opened talks with Ethiopia's state-owned Ethio Telecom to expand international capacity lease agreements. The technical argument the company made is worth restating plainly. "Our dual landings at Berbera give us a concrete technical advantage, two physically diverse subsea routes into East Africa directly integrated with a terrestrial backbone built to reach landlocked markets," said Abdurahman Yusuf, Somcable's chief technology officer. Chief executive Mohammed Ibrahim Ahmed framed the goal as resilience rather than raw bandwidth: "East Africa's digital economy demands more than bandwidth, it demands resilience," with the company aiming to provide diverse, high-capacity routes for coastal and landlocked countries. In other words, Berbera's pitch is not to carry more traffic than Mombasa, it is to carry traffic that Mombasa physically cannot reach if something breaks. ## Kenya's Cable Arithmetic Start with what Kenya actually has. Seven major submarine systems are live on the Kenyan coast as of September 2026, per the [DC254 verified register](/articles/submarine-cables-landing-mombasa): SEACOM, TEAMS, EASSy, LION2, DARE1, PEACE and 2Africa, with Africa-1 landed at Mombasa in 2024 pending full service and Daraja (Salalah to Mombasa, hosted by Safaricom) announced in October 2025 and in development. On paper this is exceptional redundancy, and it is the foundation of [Kenya's data centre market](/directory). The caveat sits in the geography: most of these systems reach Europe through the Red Sea and the Suez region, one of the world's great maritime chokepoints, and several share sea corridors with each other. That caveat has bitten. The Star's ITW coverage (14 September 2026) noted that Kenya learned during recent Red Sea cable disruptions how quickly traffic from Mombasa slows when key systems fail, a reminder that capacity without diverse international routes is a risk. When a cable is cut, carrier traffic re-routes automatically over surviving systems (border gateway protocol does this by design), but the maths are unforgiving: rerouted traffic crowds onto fewer cables, congestion rises, and effective speeds drop until repair ships do their work. Redundancy on paper only protects you if the backup path is physically independent of the failure. ## The Northern Route, Mapped Berbera's geography is the story. The port city sits on the Gulf of Aden, directly opposite the Arabian peninsula, and the two systems Somcable lands there (2Africa and PEACE) approach the Horn from route corridors that differ from the ones carrying most Mombasa-bound traffic. On top of the landings, Somcable is building the terrestrial leg: fibre from Berbera inland through Hargeisa into Ethiopia, with the company describing extensions toward South Sudan, and lease negotiations with Ethio Telecom that would turn that backbone into sellable international capacity. Kenya is not a stranger to northern alternatives. DARE1, live since 2015, already runs a direct express route from Mombasa to Djibouti, giving Kenyan operators a second coastal exit toward the north. What Berbera adds is a different kind of option: a gateway outside Kenya's borders that can carry landlocked traffic (Ethiopia, South Sudan) without touching Mombasa at all, and that strengthens the overall mesh the region's networks can fail over onto. For Principal Secretary for Broadcasting and Telecommunications Stephen Isaboke, who opened the ITW gathering, the framing was confident rather than defensive: Kenya's strategic location, regulatory environment and skilled workforce keep it positioned as the regional gateway for connectivity, data centres and cloud services. A stronger northern neighbour is consistent with that pitch, Kenya's hub status has always been about aggregation, and aggregation is easier when the surrounding mesh is dense. ## What It Means for Kenyan Buyers For enterprises that depend on international bandwidth, the practical lessons are procurement lessons. Buy capacity across multiple cable systems rather than multiple products on the same system. Ask carriers to demonstrate physical route diversity (which cables, which landing points, which inland paths) rather than accepting a resilience slide. And when contracts allow, configure failover so that critical workloads can ride a genuinely independent path when a system degrades, the [dual-entry discipline](/articles/fibre-optic-networks-kenya-data-centres) good facilities apply to metro fibre applies to international routes too. For data centre siting, the calculus barely moves but the options sharpen. Nairobi remains where the customers, exchanges and skills are, and Mombasa's landing stations (with facilities like [iColo MBA2](/directory/icolo-mba2) minutes from the beach manholes) remain the natural home for latency-critical international transit. What Berbera adds is insurance: a second gateway that lowers the odds of a regional outage isolating Kenyan infrastructure, and a future negotiating lever on international capacity prices. Watch two things through 2027: whether the Ethio Telecom lease agreements convert into lit capacity, and whether Daraja lands on schedule. Both would measurably deepen the region's route diversity, which is the metric that actually matters. Sources: Business Daily (15 September 2026, Somcable landings, terrestrial build, Ethio Telecom talks, executive quotes); The Star (14 September 2026, ITW and Datacloud Africa 2026, Red Sea disruption impact, PS Isaboke remarks). Cable register (seven live systems, Africa-1 landed pending service, Daraja in development, DARE1 route detail, Daraja announcement October 2025) from the DC254 verified map register, corrected 16 September 2026 after verification showed EIG never landed in Kenya. Facts verified 16 September 2026. ## Beyond PUE: AI Data Centre Efficiency in Kenya URL: https://data-centers-254.vercel.app/articles/beyond-pue-ai-data-centre-efficiency-kenya Cluster: Energy Efficiency Published: 2026-10-20 Updated: 2026-10-20 Summary: PUE alone no longer describes AI-era data centre efficiency. What LG's CEO argument, Uptime's rack numbers and IEA forecasts mean for Kenyan operators. Every Kenyan data centre pitch deck quotes the same number first. Power usage effectiveness, PUE, the ratio of total facility energy to the energy that actually reaches the IT equipment, has been the industry's headline efficiency figure for so long that buyers stop asking what it does and does not measure. This week the argument against stopping there got a high-profile restatement: writing in Forbes on 19 October 2026, Jae-cheol Lyu, president and CEO of LG Electronics, made the case that PUE alone can no longer describe efficiency in the AI era, and that the industry needs to evaluate three things it has historically kept separate from the metric: hardware flexibility, heat movement, and site-specific resilience. He is writing about a global market. But the argument lands with particular force on a market like ours, where the climate is generous, the grid is unusually clean, and the next generation of AI racks is about to stress every assumption PUE was built on. ## What PUE still does well Nothing in this piece is an argument to retire the metric. As we set out in [our PUE explainer](/articles/data-centre-pue-power-usage-effectiveness), PUE is genuinely good at one thing: tracking a facility's own trajectory over time. A Nairobi operator who moves from 1.8 to 1.5 across a retrofit has a real, comparable story to tell, and PUE is the right way to tell it. Lyu's position, and ours, is that the trouble starts when PUE is read as a cross-site verdict rather than a self-history. Two facilities can post identical PUEs while one runs mild highland air through its coils and the other battles coastal humidity with chillers running flat out. The identical number hides different engineering, different water bills and different exposure when something breaks. That is the gap the AI era blows open, because AI workloads stress exactly the variables PUE was never designed to see. ## The three questions that come after PUE Lyu's framework is worth reading on its own terms, so let us restate it faithfully and then translate each part into Kenyan specifics. **First: how flexibly does the facility keep pace with the hardware?** The numbers tell the story. Uptime Institute's latest survey, cited in the piece, found the share of operators running racks of 30 kW or more rose from 19% in 2025 to 24% in 2026, and roughly one in seven operators now runs racks of 50 kW or more. A facility designed around 6 kW cabinets cannot flex its way to an NVIDIA-class GPU hall; the power train, the floor loading, the airflow and the breakers all pre-date the demand. Kenyan operators are already meeting this: IXAfrica's Nairobi campus has built GPU-ready capacity at the density end of the local market, and every serious retrofit conversation in the city now starts with the same question, not "what is your PUE" but "what rack can you actually take tomorrow?" Flexibility is an efficiency property. A hall that cannot host next year's chips is inefficient by definition, no matter what its meter says, because its capital sits idle while the demand goes to another market. ![Power chain diagram](/images/diagram-power-chain-grid-to-gpu.webp) **Second: how efficiently does heat move from the chip to beyond the facility?** This is the axis where Kenya's advantage shows and also where its next capital bill hides. Nairobi's highland climate, mild days and cool nights, means free cooling still does real work for conventional racks, which is part of why local facilities post respectable PUEs without exotic mechanical plant. But the AI curve does not care about our weather. Once racks pass roughly 30 kW, air alone stops being a viable heat carrier: the heat has to move into liquid at the chip, through coolant distribution units, and out of the building, and the facility that cannot move heat at that rate simply cannot host the workload. Lyu's point is subtler than "buy liquid cooling", though. He argues IT, power and cooling should be designed as one coordinated system centred on computing performance, rather than three procurement lines optimised separately. In Kenyan terms: the chiller decision, the transformer sizing and the GPU order are the same decision, and the operator who lets a supplier optimise one line item in isolation will pay for it at the rack. **Third: how reliably does the site operate under local power, water and climate conditions?** Here the piece quotes numbers that should make any operator pause. S&P Global, as cited by Lyu, reports that 43% of 9,055 surveyed data centres sit in regions of high water stress. Kenya knows this variable intimately: the drought cycles of the last decade are national memory, and a facility that quietly depends on county water or boreholes for its cooling tower make-up carries a climate risk that never appears in its PUE. The Kenyan counterweights are real, the binary units at Olkaria are air-cooled, the grid is roughly nine-tenths renewable, so the carbon arithmetic that dominates European siting debates matters less here, but the water arithmetic is moving the other way as Nairobi grows. Reliability also means the grid dimension: PUE says nothing about ride-through when the feeder dips, and Kenya's operators, running on a grid with a thin reserve margin, size UPS and diesel for exactly the hours PUE cannot see. Two facilities with identical PUE can have very different odds of keeping a load alive through a bad week. ## The numbers behind the squeeze The urgency in Lyu's piece comes from arithmetic. The International Energy Agency projects global data centre electricity consumption will more than double, from 415 terawatt-hours in 2024 to roughly 945 terawatt-hours by 2030. Efficiency, in other words, has stopped being a procurement nicety and become a strategic variable: utilities, regulators and neighbours now all have a view on how much power these buildings deserve. The public mood data cited in the piece is striking, a June 2026 Reuters/Ipsos report found 57% of Americans oppose a data centre being built in their area, and 77% worry AI growth could push electricity rates up. Kenya runs the opposite experiment. We are not a market resisting data centres; we are a market courting them, with a national compute-hub ambition and tax incentives for the sector. That is precisely why the beyond-PUE argument matters here: in a country that wants the investment, the licence to operate will be kept by transparency, not by restraint. The operators who publish measured numbers, water drawn, diesel hours, real PUE, will be the ones who keep communities and regulators on side as the sector scales. We have argued this before in [our disclosure piece](/articles/kenya-data-centres-what-operators-should-disclose), and the global mood shift Lyu documents is the best evidence yet that disclosure is a competitive asset rather than a compliance chore. ![Olkaria geothermal plant](/images/kenya-geothermal-plant-olkaria.webp) ## Five questions Kenyan buyers should ask instead of "what is your PUE" The practical output of this argument is a better question list. When you are evaluating a Kenyan colocation provider, a landlord pitching you a shell, or a partner for a GPU deployment, ask: 1. **What rack density can you host today, and what is the roadmap?** Ask for kW per cabinet actually deployed, not designed, and what changes (power, cooling, floor) to reach 30 kW and beyond. 2. **Is the quoted PUE measured or designed?** Design PUE is a spreadsheet number. Ask for twelve months of metered data at the utility entry and UPS output. 3. **Where does cooling water come from, and what happens in a drought year?** If the answer is county water, ask for the borehole and storage contingency, and for any water usage figures the operator tracks. 4. **What is the ride-through architecture?** UPS minutes, generator hours, fuel contracts, and how many grid dips the site absorbed last year. 5. **What happens to the heat?** For new AI halls, direct-to-chip loops are coming whether we plan for them or not; the operators thinking about heat reuse now are the ones worth shortlisting. None of these questions has a PUE answer, which is the point. ## The DC254 take PUE will keep being quoted, and it should be, it is a clean, auditable measure of one real thing. The error to avoid is letting a single metric stand in for a conversation about flexibility, heat and local resilience that the AI era has made unavoidable. Kenya starts this conversation from an unusually strong position: a near-renewable grid, a climate that flatters conventional cooling, and a competitive market where the leading operators are already building for density. The facilities that will win the next three years of Kenyan demand are the ones that can answer the five questions above with numbers they are willing to publish, because in this market, trust is not a side effect of good engineering, it is the product. ## Building Codes for Data Centres in Kenya URL: https://data-centers-254.vercel.app/articles/building-codes-data-centres-kenya Cluster: Policy Published: 2026-08-28 Updated: 2026-08-28 Summary: The building codes and approvals that govern data centre construction in Kenya: county permits, NCA registration, NEMA EIAs, and fire safety. ## The Regulatory Landscape for Data Centre Construction in Kenya ![National government building at dusk](/images/dc-policy-regulation.webp) Building a data centre in Kenya is not simply a construction project, it is a multi-regulatory undertaking that involves **national government agencies, county governments, statutory bodies, and international standards**. Unlike commercial office buildings or residential developments, data centres have unique physical requirements: high power density, specialised cooling systems, raised floors, stringent fire suppression, and reinforced structural loads. Kenya's regulatory framework was not designed with data centres specifically in mind, which means operators must navigate a patchwork of general construction, environmental, electrical, and safety regulations while also meeting international data centre design standards. This article provides a comprehensive breakdown of every major regulatory layer that governs the physical construction of data centres in Kenya, from county-level building approvals to national environmental assessments, electrical codes, fire safety requirements, and occupational health obligations. It also compares Kenya's framework to international benchmarks and offers practical guidance for data centre developers. ## County Government Approvals and Building Plans ### Nairobi City County The **Nairobi City County Government** is the primary approval authority for data centre construction projects within Nairobi, which remains the dominant market for data centre development in Kenya. The county's approval process includes: - **Change of User**: If the land is not already zoned for industrial or commercial use, the developer must apply for a change of user to the appropriate zoning classification. Data centres are typically classified under **industrial or institutional use**, depending on the ownership and operational model. - **Building Plan Approval**: Detailed architectural and structural plans must be submitted to the Nairobi City County's Department of Urban Planning for review and approval. The plans must be prepared by a registered architect and structural engineer. - **County Rates and Land Levies**: Data centre operators must pay annual county land rates and may be subject to additional levies for large-scale developments. - **Signage and Wayfinding Approvals**: External signage for the data centre facility requires separate county approval. ### Mombasa County **Mombasa County** governs data centre development in Kenya's second-largest data centre market, which is strategically significant due to its proximity to submarine cable landing stations at the **Mombasa Marine Cable Landing Station**. The Mombasa County approval process is structurally similar to Nairobi's but may involve additional coastal zone management considerations under the **Kenya Maritime Authority** and **National Environment Management Authority (NEMA)** coastal zone regulations. ### Other Counties As data centre development extends beyond Nairobi and Mombasa (to locations such as **Kisumu**, **Nakuru**, and **Machakos**) developers must engage with the respective county governments. Each county may have variations in processing timelines, fee structures, and specific local requirements, though the broad framework is consistent under national legislation. ## National Construction Authority (NCA) Registration The **National Construction Authority (NCA)**, established under the **National Construction Authority Act, 2011**, is the primary regulator of Kenya's construction industry. Any construction project exceeding a prescribed value threshold must be registered with the NCA. - **Project Registration**: Data centre construction projects must be registered with the NCA, with the registration category determined by the project's estimated value. Large-scale data centres (particularly those built by hyperscale operators or major colocation providers) typically fall into **Category 1** (projects above KES 500 million) or **Category 2** (KES 100 million to KES 500 million). - **Contractor Registration**: The contractors engaged on the project, including the main contractor, mechanical and electrical (M&E) contractors, and specialist subcontractors, must be registered with the NCA in the appropriate category and trade. - **Site Supervision**: The NCA requires that all registered projects have adequate site supervision by registered professionals, including a project manager, resident engineer, and quantity surveyor as applicable. - **NCA Inspections**: The Authority conducts periodic site inspections to verify compliance with approved plans, safety standards, and construction quality. Non-compliance can result in stop-work orders. For developers conducting due diligence on a data centre project, understanding the NCA registration status of the land, the contractor, and the project itself is essential. Our [data centre due diligence guide for Kenya](/articles/data-centre-due-diligence-kenya) covers this in greater detail. ## KEBS Standards Applicable to Data Centres The **Kenya Bureau of Standards (KEBS)** develops and enforces national standards across a wide range of products and processes. For data centre construction, several KEBS standards are directly relevant: - **KS ISO/IEC 27001**: Information security management systems, adopted from the international standard and applicable to data centre operations. - **KS 2468**: Electrical installations for buildings, based on IEC 60364, governing the design, installation, and testing of electrical systems in buildings, including data centres. - **KS 530**: Code of practice for electrical wiring in buildings, providing specific requirements for wiring, earthing, and circuit protection. - **KS EN 15232**: Energy performance of buildings, relevant to data centre energy efficiency assessments. - **KS ISO 50001**: Energy management systems, applicable to data centres seeking to formalise their energy management practices. While KEBS does not have a dedicated data centre construction standard, data centre developers typically adopt **international standards** such as **IEC 60364-5-52** (wiring systems for IT equipment), **IEC 61850** (communication networks and systems in substations), and **TIA-942** (the telecommunications infrastructure standard for data centres), and then ensure compliance with the applicable KEBS standards that overlap with those requirements. ![Technician working on aerial utility infrastructure](/images/aerial-fibre-technician-utility-pole.webp) ## NEMA Environmental Impact Assessment Requirements The **National Environment Management Authority (NEMA)**, established under the **Environmental Management and Coordination Act (EMCA), 1999**, requires that certain categories of development projects undergo an **Environmental Impact Assessment (EIA)** before construction can commence. ### When is an EIA Required for a Data Centre? Under Kenya's **EIA and Audit Regulations, 2003**, data centre projects fall under the category of **industrial and infrastructure developments**. The specific triggers include: - Projects with a **built-up area exceeding a prescribed threshold** (typically above 5,000 square metres for industrial developments). - Projects involving **significant power infrastructure**, including backup generators, transformers, and UPS systems. - Projects located in or near **environmentally sensitive areas**, including wetlands, forests, water catchment areas, or coastal zones. - Projects that may generate **noise pollution** from cooling systems and backup generators. - Projects involving the **use or storage of hazardous materials**, such as diesel fuel for generators and fire suppression chemicals. ### The EIA Process 1. **Appointment of an EIA Expert**: The developer engages a registered Environmental Impact Assessor licensed by NEMA's **National Environment Management Authority, Practitioners Institute** (NEMAPI-KE). 2. **Scoping Report**: The EIA expert prepares a scoping report identifying the key environmental and social issues associated with the project. 3. **Full EIA Study**: A comprehensive study covering air quality, noise, water, soil, biodiversity, socio-economic impacts, and waste management. 4. **Public Participation**: NEMA requires public consultation with affected communities, including neighbours, local administration, and relevant government agencies. 5. **EIA Report Submission**: The completed report is submitted to [NEMA](https://www.nema.go.ke) for review. 6. **NEMA Review and Decision**: NEMA reviews the report, may request additional information, and issues either an **EIA Licence** (approval) or a rejection. 7. **Environmental Audit**: After construction, an annual environmental audit is required to verify ongoing compliance. For data centres, the EIA process typically addresses **noise from cooling plants and generators**, **emissions from diesel backup systems**, **stormwater management**, **thermal discharge**, and **electromagnetic field (EMF) emissions** from electrical equipment. ## Fire Safety Codes and Requirements Fire safety is one of the most critical regulatory areas for data centres, given the high density of electrical equipment, the presence of combustible materials (cabling, insulation), and the catastrophic consequences of a fire in a facility housing critical digital infrastructure. ### Kenya Fire Brigade Act (Cap 56) The **Kenya Fire Brigade Act** provides the national framework for fire prevention and response. While it is a relatively old piece of legislation, it establishes the principle that building owners are responsible for fire safety in their premises and empowers fire authorities to inspect premises and issue compliance notices. ### County Fire Safety By-Laws Each county government has enacted **fire safety by-laws** that provide more specific requirements. In Nairobi, the **Nairobi City County Fire Safety By-Laws** require: - Installation of **fire detection and alarm systems** throughout the facility. - Provision of **automatic fire suppression systems**, for data centres, this typically means **clean agent suppression systems** (such as FM-200, Novec 1230, or inert gas systems like IG-541) in the server room and electrical rooms, as water-based sprinklers can cause secondary damage to electronic equipment. - **Fire-rated doors and walls** separating the data hall from other areas of the building. - **Emergency evacuation routes**, illuminated exit signs, and fire assembly points. - **Fire hydrant connections** and standpipe systems for fire brigade access. - Regular **fire safety audits** conducted by licensed fire safety consultants. - Provision of **fire extinguishers** of appropriate types (CO2 for electrical fires, dry chemical for general areas) at specified intervals. ### NFPA 75 and NFPA 76, International Benchmarks While not legally binding in Kenya, **NFPA 75** (Standard for the Fire Protection of Information Technology Equipment) and **NFPA 76** (Standard for the Fire Protection of Telecommunications Facilities) are widely adopted by Kenyan data centre operators as supplementary standards. These US-based standards provide detailed requirements for: - **Early warning smoke detection systems** (such as VESDA, Very Early Smoke Detection Apparatus) in data halls. - **Compartmentation** to contain fire within a single zone. - **Shutdown procedures** for HVAC and electrical systems in the event of a fire. - **Cable management practices** that reduce flame spread. ## Electrical Installation Standards Data centres are extraordinarily power-intensive facilities, and Kenya's electrical installation standards are directly relevant to their design and construction. ### KEBS and IEC Standards Electrical installations in Kenyan data centres must comply with: - **KS 530 / IEC 60364**: The foundational standards for electrical wiring in buildings, covering circuit design, cable sizing, earthing, and protection against electric shock and overcurrent. - **IEC 60364-5-52**: Specific requirements for wiring systems in IT and data processing environments. - **KS 2468**: Electrical installations for buildings, the Kenyan adoption of general electrical installation standards. ### Licensed Electrical Contractors and Engineers Under the **Energy Act, 2019**, electrical installation work must be carried out by a **licensed electrical contractor** and supervised by a **registered electrical engineer**. Upon completion, the installation must be tested and certified by the contractor, and an **electrical installation certificate** issued. The **Energy and Petroleum Regulatory Authority (EPRA)** oversees the licensing of electrical contractors and the enforcement of electrical safety standards. ### Power Infrastructure Specific to Data Centres Data centres in Kenya require: - **Dual utility power feeds** from **Kenya Power (KPLC)**, ideally from separate substations for redundancy. - **Backup diesel generators** (typically N+1 or 2N redundancy configurations) with adequate **fuel storage** (subject to NEMA and county fire safety regulations for fuel storage). - **Uninterruptible Power Supply (UPS) systems**, battery-based systems that bridge the gap between utility power loss and generator start-up. - **Power Distribution Units (PDUs)** and **Automatic Transfer Switches (ATS)** designed to data centre specifications. - **Earthing and grounding systems** designed to handle the high fault currents typical of data centre electrical installations. ## Occupational Health and Safety (OSHA, Kenya) ![Construction workers in hard hats reviewing a building site](/images/construction-workers-site-review.webp) The **Occupational Safety and Health Act, 2007** (OSHA, Kenya) and its accompanying regulations establish requirements for workplace safety during both the construction and operational phases of a data centre. ### Construction Phase - **Site safety plans** must be developed and maintained by the main contractor. - **Personal Protective Equipment (PPE)** must be provided to all workers. - **Scaffolding, excavation, and working-at-height** safety procedures must be documented and enforced. - **First aid facilities** and trained first aiders must be available on site. - **Accident reporting** to the **Directorate of Occupational Safety and Health Services (DOSHS)** is mandatory for any workplace injury or fatality. ### Operational Phase - Data centre operators must conduct **regular workplace risk assessments** covering electrical hazards, noise exposure (from cooling and generator systems), chemical handling (diesel, fire suppression agents), and ergonomic risks. - **Safety training** must be provided to all data centre staff, including emergency response procedures. - **Safety committees** are required for workplaces with more than 20 employees. - **Incident reporting** to DOSHS is mandatory for workplace injuries, illnesses, and dangerous occurrences. ## Kenya Building Code and Zoning Considerations The **Kenya Building Code**, based on the British Standard BS 8110 (structural) and other international codes, provides the baseline requirements for building construction in Kenya. For data centres, several elements of the Building Code are particularly relevant: ### Zoning Data centres must be located in zones where industrial or commercial activities are permitted. The key zoning considerations include: - **Industrial zones** (typically preferred for data centres) allow for high-power installations, generator operations, and 24-hour activity. - **Commercial zones** may also permit data centres but may have restrictions on noise levels, hours of operation, and the size of backup generator installations. - **Residential zones** generally do not permit data centre development due to noise, traffic, and power infrastructure concerns. ### Setback Requirements The Kenya Building Code and county by-laws specify **setback distances** from property boundaries, roads, and neighbouring buildings. For data centres, which often require large footprints, setback requirements can significantly affect the usable land area. Typical setback requirements in Nairobi include: - **Front setback**: 6 to 9 metres from the road. - **Side setbacks**: 3 to 6 metres from side boundaries. - **Rear setback**: 3 to 6 metres from the rear boundary. ### Structural Requirements Data centres impose **exceptional floor loads** due to the weight of server racks, UPS systems, and transformers. The structural design must accommodate: - **Raised floor loads** of 10–15 kN/m² or higher (compared to 2–3 kN/m² for typical office buildings). - **Equipment room floor loads** for heavy transformers and generators. - **Seismic considerations**, particularly for facilities in Nairobi, which lies in a moderate seismic zone. ## Approval Timeline: What to Expect ![Tower cranes over a high-rise construction site](/images/highrise-construction-cranes-kenya.webp) The total timeline for obtaining all necessary approvals to build a data centre in Kenya typically ranges from **6 to 18 months**, depending on: | Approval Stage | Typical Duration | Key Authority | |---|---|---| | Land search and due diligence | 2–4 weeks | Ministry of Lands | | Change of user (if required) | 2–6 months | County Government | | Building plan approval | 2–4 months | County Government | | NCA project registration | 2–4 weeks | NCA | | NEMA EIA process | 3–6 months | NEMA | | Fire safety clearance | 1–2 months | County Fire Brigade | | Electrical installation approval | Ongoing (during construction) | EPRA | | Occupancy certificate | 2–4 weeks | County Government | Developers should plan for **parallel processing** where possible, for example, initiating the NEMA EIA process concurrently with building plan submissions. However, the NEMA EIA licence and the county building plan approval are typically both required before construction can commence. Our [Kenya data centre licensing framework](/articles/kenya-data-centre-licensing-framework) article provides additional detail on the broader licensing landscape beyond physical construction approvals. ## Comparison with International Standards ### Uptime Institute Tier Requirements The **Uptime Institute's Tier Classification System** (Tier I through Tier IV) is the global benchmark for data centre reliability and redundancy. Kenya's building codes do not directly reference Tier classifications, but most commercial data centre operators in Kenya design to at least **Tier III** standards (concurrently maintainable infrastructure). Achieving Uptime Institute certification requires meeting specific requirements for: - **Redundant power and cooling distribution paths**. - **Concurrent maintainability**, the ability to remove any single component for maintenance without disrupting operations. - **Fault tolerance** (Tier IV), the ability to withstand any single failure without impact to IT operations. Kenyan building codes ensure structural safety and basic fire protection, but they do not address the specific redundancy, cooling, and electrical distribution requirements that define Tier classifications. Data centre operators must therefore **self-impose** these international standards and ensure their designs are compatible with local code requirements. Our [data centre tier ratings explained](/articles/data-centre-tier-ratings-explained) article provides a full breakdown of each tier level. ### NFPA 75/76 (US Fire Codes) As discussed in the fire safety section, **NFPA 75** and **NFPA 76** provide detailed fire protection requirements specific to IT and telecommunications facilities. Kenya's fire safety codes are less specific and do not address data centre fire scenarios in detail. Kenyan data centre operators that adopt NFPA standards must ensure that their clean agent suppression systems, VESDA detection, and fire-rated compartmentation also satisfy the requirements of the Kenya Fire Brigade Act and county by-laws, which may include additional provisions not covered by NFPA. ### EU Code of Conduct on Data Centre Energy Efficiency The **EU Code of Conduct on Data Centre Energy Efficiency** is a voluntary framework that sets best-practice guidelines for energy-efficient data centre design and operation. While developed for the European market, its principles, including Power Usage Effectiveness (PUE) targets, airflow management, free cooling utilisation, and waste heat recovery, are increasingly relevant to Kenyan data centres. Kenya's building codes do not prescribe specific energy efficiency targets for data centres, but the **Kenya Bureau of Standards (KEBS)** has adopted energy management standards (KS ISO 50001) that data centres can leverage to formalise their energy efficiency programmes. ![Aerial view of a data centre campus with rooftop solar](/images/dc-environment-sustainability.webp) ## Practical Recommendations for Data Centre Developers For organisations planning to build or expand data centre capacity in Kenya, the following practical recommendations can streamline the regulatory process: 1. **Engage early with county authorities**: Initiate pre-application consultations with the relevant county planning department before submitting formal applications. This can identify potential issues before they become costly delays. 2. **Hire locally registered professionals**: Ensure your architect, structural engineer, electrical engineer, and EIA expert are all registered with their respective Kenyan professional bodies (Architectural Association of Kenya, Engineers Board of Kenya, NEMAPI-KE). 3. **Adopt international standards proactively**: Designing to Uptime Institute Tier, TIA-942, and NFPA standards from the outset (and documenting compliance) will make the local approval process smoother and build credibility with clients. 4. **Parallel the NEMA EIA process**: Start the EIA early, even before building plans are finalised, to avoid it becoming a bottleneck. 5. **Budget for compliance costs**: Regulatory compliance (including NCA fees, NEMA EIA fees, county approval charges, fire safety consulting, and electrical testing) can add 2–5% to total project cost. 6. **Plan for ongoing compliance**: Post-construction obligations include annual environmental audits, fire safety inspections, electrical installation renewals, and OSHA workplace safety compliance. ## Conclusion Kenya's regulatory framework for data centre construction is multi-layered, involving **county governments, the National Construction Authority, NEMA, KEBS, fire authorities, and occupational safety bodies**. While the framework was not designed specifically for data centres (and this creates some friction) it provides a comprehensive baseline of safety, environmental, and quality assurance requirements. Successful data centre developers in Kenya are those who combine **international design standards** (Uptime Institute, TIA-942, NFPA) with **diligent local regulatory compliance**, engaging early and consistently with the relevant authorities throughout the project lifecycle. As Kenya's data centre market continues to grow, there is increasing advocacy for a more **data-centre-specific regulatory framework** that would streamline approvals, provide clearer technical standards, and reduce the current reliance on general construction codes. Until such a framework emerges, operators must navigate the existing system with thorough preparation and expert guidance. ## Frequently Asked Questions ### What approvals are required to build a data centre in Nairobi? Building a data centre in Nairobi requires multiple approvals: a change of user and building plan approval from Nairobi City County, registration of the project with the National Construction Authority (NCA), an Environmental Impact Assessment (EIA) licence from NEMA, fire safety clearance from the Nairobi County Fire Brigade, electrical installation certification from a licensed electrical engineer under KEBS standards, and occupational health and safety registration with the Directorate of Occupational Safety and Health Services (DOSHS). The process typically takes 6 to 18 months depending on project scale and location. ### Does Kenya have specific building codes for data centres? Kenya does not have a dedicated building code exclusively for data centres. Instead, data centre construction is governed by the general Kenya Building Code, county government by-laws, and applicable Kenyan standards from KEBS. However, data centre operators typically design and build to international standards such as the Uptime Institute's Tier I–IV classifications, TIA-942 (Telecommunications Infrastructure Standard for Data Centres), and NFPA 75/76 (US fire protection standards for IT equipment), and then ensure compliance with Kenyan requirements on top of those international benchmarks. ### What fire safety regulations apply to data centres in Kenya? Data centres in Kenya must comply with the Kenya Fire Brigade Act (Cap 56), county government fire safety by-laws, and the Kenya Building Code's fire safety provisions. Key requirements include fire detection and alarm systems, automatic suppression systems (such as clean agent systems using FM-200 or Novec 1230, which are safe for electronic equipment), fire-rated structural elements, emergency evacuation routes, fire hydrant connections, and regular fire safety audits. For data centres pursuing international certification, NFPA 75 (Standard for the Fire Protection of Information Technology Equipment) and NFPA 76 (Standard for the Fire Protection of Telecommunications Facilities) are commonly adopted as supplementary standards. ### How long does the NEMA EIA process take for a data centre project? The NEMA Environmental Impact Assessment process for a data centre typically takes 3 to 6 months, though the timeline can vary based on project complexity, location sensitivity, and the completeness of the EIA report. The process involves engaging a registered Environmental Impact Assessor (EIA/EA expert), conducting environmental and social impact studies, preparing the EIA report, submitting it to NEMA for review, addressing any comments or requests for additional information, and attending a public hearing if required. For data centres located in sensitive zones (such as near water bodies, protected areas, or residential neighbourhoods) the review process may be extended. NEMA charges EIA review fees based on project capital cost, and data centres, given their significant power and cooling requirements, are generally classified as medium to high-impact projects. ### How do Kenya's building regulations compare to international data centre standards? Kenya's building regulations provide a baseline of structural safety, fire protection, electrical safety, and environmental compliance, but they do not address the specific operational requirements of data centres, such as redundant power, cooling capacity, raised access flooring, seismic resilience of server racks, or concurrent maintainability. International standards such as the Uptime Institute's Tier Classification System (Tier I through Tier IV), TIA-942, and the EU Code of Conduct on Data Centre Energy Efficiency provide the specialised technical framework that data centre operators actually design to. In practice, Kenyan data centre developers adopt these international standards and then layer on local regulatory compliance. The challenge is that Kenya's building code and approval processes were not designed with data centres in mind, which can create friction, for example, building inspectors may not be familiar with raised-floor ventilation requirements or clean agent fire suppression systems. ## Career Progression Paths in African Data Centres URL: https://data-centers-254.vercel.app/articles/career-progression-african-data-centres Cluster: Careers Published: 2026-08-28 Updated: 2026-08-28 Summary: Realistic career ladders in Africa's data centre industry, from cable technician to VP of Operations, with Kenyan salary ranges and certifications. Africa's data centre industry is undergoing explosive growth, and Kenya sits at the heart of this transformation. With Nairobi emerging as a major digital hub and investments from operators like Africa Data Centres, Liquid Intelligent Technologies, and global hyperscalers pouring into the region, the [career landscape for data centre professionals](/articles/data-centre-careers-kenya) has never been more promising. But what does a realistic career path actually look like, and how do you climb from cable puller to C-suite? This article maps out the complete career progression ladder in the African data centre industry, with specific salary data for the Kenyan market, certification milestones at each level, and practical advice on regional mobility across East Africa and beyond. ![Field technician working on aerial fibre connections on a utility pole](/images/aerial-fibre-technician-utility-pole.webp) *Data centre careers in Kenya span from hands-on technical roles to strategic leadership positions* ## The African Data Centre Career Ladder Overview The data centre industry in Africa (and Kenya specifically) follows a fairly consistent hierarchical structure, though the pace of advancement is often faster than in mature markets like Europe or North America. The talent shortage means capable professionals are promoted quickly, and new facilities being built across the continent create leadership vacancies that didn't exist five years ago. According to [Africa Data Centres' market reports](https://www.africadatacentres.com/news/), the continent needs approximately 10,000 additional skilled data centre professionals by 2030 to meet demand. Kenya alone, driven by the [national market outlook through 2030](/articles/kenya-data-centre-market-outlook-2025-2030), is expected to require over 2,000 new roles across technical, operational, and leadership functions. ## Entry-Level Roles (0–3 Years Experience) Entry-level positions form the foundation of any data centre career. These roles are hands-on, requiring physical presence in the facility, and they provide the essential ground-floor understanding of how data centres actually operate. ### Cable Technician The cable technician is often the first rung on the data centre career ladder. Responsibilities include installing, terminating, and testing structured cabling, both copper (Cat6/6a) and fibre optic, organising cable pathways, maintaining patch panel documentation, and ensuring compliance with TIA-942 cabling standards. **Requirements:** A diploma in electrical engineering, IT, or telecommunications. Fibre optic certification (such as CFOT from the Fibre Optic Association) is a strong advantage. Most training happens on the job. **Salary Range (Kenya):** KES 40,000–65,000 per month. ### NOC Junior Analyst (Network Operations Centre) NOC junior analysts monitor data centre systems around the clock, tracking network performance, server health, environmental conditions (temperature, humidity), and power systems. They escalate incidents according to standard operating procedures and maintain shift logs. **Requirements:** A degree or diploma in IT, computer science, or telecommunications. CompTIA Network+ or CCNA certification is preferred. Strong problem-solving abilities and willingness to work rotating shifts. **Salary Range (Kenya):** KES 50,000–75,000 per month. ### Data Centre Technician A broader role than cabling alone, data centre technicians handle server racking, hardware installation, basic troubleshooting, and preventive maintenance tasks. They work closely with both the facilities and IT teams. **Requirements:** Diploma or degree in IT or electrical engineering. Certifications like CompTIA A+ and basic familiarity with Linux command-line operations. **Salary Range (Kenya):** KES 45,000–70,000 per month. ![Structured cabling connecting server racks](/images/server-rack-patch-cabling.webp) *Cable technicians form the critical entry point into the data centre career ladder* ## Mid-Level Roles (3–8 Years Experience) Mid-level is where professionals specialise and begin to take ownership of significant systems or teams. This is also where certifications start to make a major difference in earning potential. ### Network Engineer Network engineers design, implement, and maintain the complex network infrastructure that connects data centre systems, from spine-leaf architectures to interconnections with internet exchange points like KIXP. They handle BGP peering, MPLS circuits, SD-WAN deployments, and network security. **Requirements:** CCNP or JNCIP certification. Degree in computer science, telecommunications, or related field. Experience with data centre network architectures, including overlay networks (VXLAN, EVPN). **Salary Range (Kenya):** KES 120,000–250,000 per month. At hyperscale cloud providers (AWS, Azure, Google), this can reach KES 300,000+ with benefits. ### Facilities Engineer Facilities engineers are responsible for the physical infrastructure, power distribution (UPS, generators, switchgear), cooling systems (CRAC/CRAH units, chillers, and increasingly liquid cooling), fire suppression, and building management systems (BMS). This is one of the most critical and undersupplied roles in the African market. **Requirements:** Degree in mechanical or electrical engineering. EPI CDCP (Certified Data Centre Professional) certification. Experience with power and cooling systems in mission-critical environments. **Salary Range (Kenya):** KES 100,000–200,000 per month. Specialists with power systems expertise can command KES 180,000–280,000. ### Operations Manager Operations managers oversee the day-to-day running of a data centre facility. They manage shift teams, coordinate maintenance windows, handle incident response, manage vendor relationships, and ensure SLA compliance with customers. **Requirements:** 5+ years in data centre operations. EPI CDCS (Certified Data Centre Specialist) or Uptime Institute ATS certification. PMP or PRINCE2 is an advantage. Strong leadership and communication skills. **Salary Range (Kenya):** KES 150,000–350,000 per month, depending on facility size and operator. ![Site crew in protective gear reviewing an active build site](/images/construction-workers-site-review.webp) *Operations managers oversee critical infrastructure across multiple African markets* ### Compliance and Regulatory Officer With the Kenya Data Protection Act 2019 in full effect and increasing regulatory scrutiny across the continent, compliance officers ensure data centres meet all legal and industry standards, from data residency requirements to physical security audits and environmental reporting. **Requirements:** Degree in law, IT, or information security. Knowledge of the Kenya Data Protection Act, GDPR principles, and industry standards like ISO 27001 and SOC 2. **Salary Range (Kenya):** KES 120,000–250,000 per month. ## Senior and Director-Level Roles (8–15+ Years Experience) Senior roles combine deep technical expertise with strategic business leadership. At this level, professionals are shaping the direction of data centre operations across entire regions or organisations. ### Senior Operations Manager / Head of Data Centres This role typically oversees multiple facilities or a large-scale campus. Responsibilities include strategic capacity planning, capex/opex budgeting, team development across multiple sites, P&L accountability, and executive stakeholder management. **Requirements:** 10+ years of progressive data centre experience. CDCE (Certified Data Centre Expert) from EPI or Uptime Institute ATD. MBA or equivalent business qualification is increasingly expected. **Salary Range (Kenya):** KES 350,000–700,000 per month, plus performance bonuses and benefits. ### VP of Operations / Director of Infrastructure At the VP or Director level, professionals are responsible for the entire infrastructure strategy of an organisation. This includes decisions about new facility builds, technology selection, sustainability initiatives, and workforce planning across a region. **Requirements:** 12–15+ years in data centre or critical infrastructure. Proven track record of managing large teams and budgets. Strong industry network across Africa. **Salary Range (Kenya):** KES 600,000–1,500,000+ per month. At pan-African operators, total compensation including stock, bonuses, and allowances can exceed KES 2,000,000. ### Chief Technology Officer (CTO) The CTO defines the technology roadmap for the organisation, from adoption of liquid cooling for AI workloads to cloud-hybrid strategies and edge computing deployments. This is the pinnacle of the technical career path. **Requirements:** Deep expertise across data centre disciplines. Strategic vision and the ability to align technology investments with business objectives. Typically 15+ years of experience. **Salary Range (Kenya):** KES 1,000,000–3,000,000+ per month at major operators, with total packages significantly higher at listed companies. ### Managing Director / Country Manager For those who pivot from pure technology into general management, the MD or Country Manager role leads the entire business unit (including sales, operations, finance, and people) for a specific market or region. **Requirements:** A blend of technical credibility and strong business development skills. MBA from a recognised institution. Extensive industry relationships. **Salary Range (Kenya):** KES 1,500,000–4,000,000+ per month, with significant performance-linked components. ![Officials and crew at a REREC pole handover event in Kenya](/images/rerec-pole-handover-kenya.webp) *Senior leadership in African data centres requires both technical depth and business acumen* ## Career Paths by Employer Type Not all data centre careers look the same. The path you take depends significantly on the type of organisation you join. ### Colocation and Dedicated Data Centre Operators Companies like **Africa Data Centres** (part of Cassava Technologies with facilities in Nairobi, Mombasa, and across the continent), **PAIX Data Centres**, and **Liquid Intelligent Technologies** offer the most traditional career ladders. Progression is linear (from technician to engineer to manager to director) with clear milestones tied to certifications and facility expansions. These operators provide the broadest exposure to all aspects of data centre operations. ### Hyperscale Cloud Providers **AWS**, **Microsoft Azure**, and **Google Cloud** are ramping up their presence in Africa. AWS operates from Cape Town with planned Nairobi expansion, while Microsoft Azure has multiple regions serving the continent. Careers here follow the tech company model, with globally standardised job families, stock-based compensation, and opportunities for international transfers. The [surge in AI and GPU computing](/articles/gpu-computing-kenya-ai-needs-data-centres) is driving rapid hiring across these platforms. ### Telecom-Integrated Operators **Safaricom** runs significant data centre infrastructure as part of its telecom ecosystem, and careers here offer the advantage of understanding the intersection of data centres, mobile networks, and enterprise services. Similarly, **Airtel Kenya** and the larger **Airtel Africa** group provide data centre-adjacent career paths. ### Enterprise IT Banks (KCB, Equity, Co-operative Bank), government agencies, and large corporates operate their own data centres or significant server rooms. Career progression in enterprise IT tends to be slower and less specialised, but offers stability and the ability to move between sectors. ## Complete Career Progression Table | Level | Role | Typical Experience | Key Certifications | Salary Range (KES/month) | |-------|------|--------------------|--------------------|--------------------------| | Entry | Cable Technician | 0–2 years | CFOT, CompTIA A+ | 40,000–65,000 | | Entry | NOC Junior Analyst | 0–2 years | CompTIA Network+, CCNA | 50,000–75,000 | | Entry | DC Technician | 0–3 years | CompTIA A+, Linux+ | 45,000–70,000 | | Mid | Network Engineer | 3–7 years | CCNP, JNCIP | 120,000–250,000 | | Mid | Facilities Engineer | 3–7 years | CDCP, vendor certs | 100,000–200,000 | | Mid | Operations Manager | 5–8 years | CDCS, PMP | 150,000–350,000 | | Mid | Compliance Officer | 3–7 years | CDPSE, ISO 27001 LA | 120,000–250,000 | | Senior | Snr Operations Manager | 8–12 years | CDCE, ATS | 350,000–700,000 | | Senior | VP of Operations | 12–15+ years | CDCE, MBA | 600,000–1,500,000+ | | Senior | CTO | 15+ years | Multiple | 1,000,000–3,000,000+ | | Senior | Managing Director | 15+ years | MBA, industry | 1,500,000–4,000,000+ | ## Regional Mobility: Working Across Africa n One of the most compelling aspects of a data centre career in Kenya is the potential for **regional and continental mobility**. Unlike many industries where geographic movement requires starting over, data centre skills are highly transferable, a UPS system in Nairobi operates on the same principles as one in Lagos, Kigali, or Johannesburg. ### East African Opportunities The **East African Community (EAC)** Common Market protocol facilitates the movement of skilled professionals between Kenya, Tanzania, Uganda, Rwanda, Burundi, South Sudan, and the Democratic Republic of Congo. Data centre operators actively leverage this, Africa Data Centres has staffed its Kigali facility partly with professionals from its Nairobi hub, while Liquid Intelligent Technologies moves network and infrastructure engineers across its East African footprint. **Tanzania** is investing in digital infrastructure around Dar es Salaam, with new colocation facilities emerging. **Uganda** has seen growth in data centre capacity in Kampala, driven by the country's fintech and government digitalisation programmes. **Rwanda** has positioned itself as a tech hub with investments from operators like Africa Data Centres, creating opportunities for Kenyan professionals willing to relocate. ### Pan-African Movement For those willing to travel further, **Nigeria** (Lagos and Abuja) represents the largest data centre market on the continent by capacity, while **South Africa** (Johannesburg and Cape Town) hosts the most mature ecosystem. **Egypt** (Cairo) and **Morocco** (Casablanca) are emerging as North African hubs. Operators like [Africa Data Centres and PAIX have facilities across these markets](https://www.africadatacentres.com/news/), and internal transfers are common for high-performing employees. The [Uptime Institute Global Data Centre Survey](https://uptimeinstitute.com/resources/research) consistently highlights that talent mobility is one of the top three challenges for African data centre operators, which means professionals with experience across multiple African markets command significant premiums. ### Visa and Work Permit Considerations For East African Community citizens, work permits within the EAC are relatively straightforward. For movement beyond the EAC, most data centre operators handle work permit applications as part of internal transfer processes. Specialised skills in short supply (such as senior facilities engineers and power systems specialists) generally receive expedited processing in most African countries. ## Growing Demand for Specialised Roles n The African data centre market is evolving beyond traditional roles. Several specialised positions are seeing surging demand that didn't exist five years ago. ### Cooling Engineers As data centres in Kenya and across Africa face rising power densities (particularly with AI and GPU workloads) the demand for engineers who understand precision cooling, liquid cooling systems, and free cooling strategies is growing rapidly. Engineers with expertise in chiller plant optimisation and tropical climate cooling design are particularly valued. ### Power Systems Specialists Kenya's unique energy mix (with significant geothermal, hydro, and growing solar capacity) creates demand for specialists who understand both traditional data centre power systems (UPS, generators, ATS) and renewable energy integration. The Kenya Power and Lighting Company (KPLC) tariff structures for large commercial and industrial customers add another layer of complexity that local expertise addresses. ### Compliance and Data Sovereignty Officers The Kenya Data Protection Act 2019, combined with increasing data localisation pressures across the continent, has created strong demand for professionals who understand both the technical and legal aspects of data sovereignty. This role requires a rare combination of IT infrastructure knowledge and regulatory expertise. ### AI and HPC Infrastructure Engineers The global AI boom is reaching Africa. With operators deploying GPU clusters and high-performance computing infrastructure, engineers who can design, deploy, and manage AI-optimised data centre environments (including liquid cooling for high-density racks, InfiniBand networking, and GPU workload scheduling) are in exceptionally high demand. ## How to Accelerate Your Career Progression For professionals looking to move up the data centre career ladder faster in Kenya, several strategies stand out. **Get certified early.** As covered in our guide on [landing a data centre job in Kenya](/articles/how-to-get-job-kenyan-data-centre), certifications like CDCP, CCNA, and CompTIA Network+ can accelerate your progression by 1–2 years. They demonstrate commitment and fill knowledge gaps that on-the-job experience alone might not address. **Choose your employer strategically.** Joining an operator during an expansion phase (when Africa Data Centres was building out its Nairobi campus, for example) creates fast-track opportunities that stable operations don't offer. **Develop cross-functional skills.** The most successful data centre leaders understand both IT and facilities. A network engineer who also understands power and cooling, or a facilities engineer who can discuss network architecture, will always outpace a single-discipline specialist. **Build your continental network.** Attend events like the Africa Data Centres Summit, Datacloud Africa, and the East Africa ICT conference. Relationships built at these events often lead to cross-border opportunities. **Pursue regional assignments.** When your employer offers a secondment to another country (whether Rwanda, Uganda, or Nigeria) take it. Cross-market experience is one of the strongest differentiators for senior roles. ## Frequently Asked Questions ### What is the starting salary for a data centre technician in Kenya? Entry-level data centre technicians in Kenya typically earn between KES 40,000 and KES 70,000 per month, depending on the employer and specific role. Cable technicians and NOC junior analysts at operators like Africa Data Centres or Safaricom usually start around KES 50,000–60,000. This figure rises significantly with certifications such as CompTIA Network+ or CDCP, which can push starting offers closer to KES 70,000–80,000. Enterprise IT departments and colocation providers tend to pay at the lower end, while hyperscale cloud providers like AWS and Azure offer competitive packages that may include additional benefits like stock units and training budgets. ### How long does it take to reach a senior data centre role in Kenya? The timeline varies, but most professionals reach a mid-level position (Network Engineer, Facilities Engineer, or Operations Manager) within 4–7 years of starting. Advancing to a senior role like Senior Operations Manager or Director typically takes 8–14 years of progressive experience, relevant certifications (CDCS, CDCE, CCNP, PMP), and demonstrated leadership. The rapidly growing market in Kenya accelerates this timeline compared to more mature markets, professionals who join a growing operator like Africa Data Centres or Liquid Intelligent Technologies during an expansion phase often get promoted faster due to the demand for experienced leadership. ### Can Kenyan data centre professionals work in other African countries? Absolutely. Regional mobility is one of the biggest advantages of a data centre career in Kenya. Major operators like Africa Data Centres (part of Cassava Technologies), Liquid Intelligent Technologies, and PAIX Data Centres have facilities across multiple African countries. Kenyan professionals are particularly valued in East African markets, Tanzania, Uganda, Rwanda, and Ethiopia are all expanding their digital infrastructure. The East African Community (EAC) Common Market protocol facilitates work permit mobility. Additionally, companies like Africa Data Centres have deployed staff from their Nairobi hubs to support new facility builds in Kigali, Lagos, and Cape Town. ### Which data centre operators in Kenya offer the best career growth? Africa Data Centres (the largest pan-African operator) offers the broadest career growth due to their multi-country footprint and ongoing expansion across the continent. Safaricom provides strong career paths within Kenya's largest telecom ecosystem. Liquid Intelligent Technologies combines data centre operations with their fibre and cloud businesses, offering cross-functional growth. PAIX Data Centres, backed by the PAIX Group with links to the Teraco ecosystem, offers exposure to international best practices. For those targeting cloud-native careers, AWS (via the Cape Town and upcoming Nairobi regions) and Microsoft Azure provide globally recognised career frameworks. ### What specialised data centre roles are in highest demand in Kenya? The highest-demand specialised roles in Kenya include cooling and HVAC engineers (critical in Nairobi's climate and for new liquid cooling deployments), power systems specialists (particularly with expertise in UPS systems, diesel generators, and Kenya's renewable energy integration like geothermal), and compliance officers familiar with the Kenya Data Protection Act 2019 and emerging regional frameworks. With the [growth of AI workloads requiring GPU infrastructure](/articles/gpu-computing-kenya-ai-needs-data-centres), high-performance computing (HPC) engineers and AI infrastructure specialists are also seeing surging demand. Sustainability and ESG reporting roles are emerging as operators pursue green certifications. ## Cloud Services in Kenya: AWS, Azure, Google Compared URL: https://data-centers-254.vercel.app/articles/cloud-services-kenya-compared Cluster: Cloud Computing Published: 2026-08-28 Updated: 2026-09-28 Summary: No global cloud provider runs a region inside Kenya. Compare AWS, Azure and Google Cloud on latency, data residency and pricing for Kenyan workloads. When a Kenyan bank decides to move its core banking platform to the cloud, when a startup chooses where to host its application, or when a government agency evaluates options for its digital services, the same question arises: which cloud provider is right for us? The answer is rarely straightforward. It depends on latency requirements, data residency regulations, budget constraints, technical requirements, and the organisation's long-term strategy. ![Cutaway infographic of an AI data centre's systems](/images/whats-inside-ai-data-center.webp) Kenya's cloud market in 2025 is served by a mix of global hyperscale providers operating remotely, local telecom operators offering cloud services from Kenyan data centres, and a growing number of managed service providers that help enterprises navigate the complexity. Understanding the landscape (who offers what, where the actual servers are, and what the trade-offs are) is essential for making informed decisions. ## The Global Hyperscalers in Kenya The three global cloud providers, Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP), all serve Kenyan customers, but none of them operates a cloud region within Kenya's borders. This single fact shapes the entire competitive landscape. ### Amazon Web Services (AWS) AWS is the dominant cloud provider in Africa by market share, and Kenya is no exception. AWS launched its Africa (Cape Town) region in April 2020, becoming the first hyperscaler to have a physical presence on the African continent. The Cape Town region has three availability zones, providing the redundancy that enterprise customers require. From Nairobi, accessing AWS Cape Town involves traversing approximately 4,000 kilometres of submarine and terrestrial fibre, with typical round-trip latency of 50–80 milliseconds. For most business applications (web applications, APIs, databases serving internal users) this latency is acceptable and often imperceptible. For real-time applications like high-frequency trading, multiplayer gaming, or interactive video, it can be noticeable. AWS has been proactive in building its Kenyan ecosystem. The company operates AWS Direct Connect partnerships with data centre operators in Nairobi, allowing dedicated, private network connections between customer equipment in Nairobi and AWS Cape Town. These Direct Connect connections bypass the public internet, providing more consistent latency and bandwidth. AWS also runs active training and partner programmes in Kenya, working with local universities and technology companies to build cloud skills. AWS's strength is breadth: with over 200 fully featured services spanning compute, storage, databases, machine learning, analytics, Internet of Things, and serverless computing, AWS offers the most comprehensive cloud platform available. For organisations with complex or diverse cloud needs, this breadth is a significant advantage. The ecosystem of third-party tools, consulting partners, and pre-built solutions that has developed around AWS is also the largest of any cloud provider. ### Microsoft Azure Microsoft Azure has been the most aggressive of the three hyperscalers in building a local presence in Kenya, not through a cloud region, but through partnerships and programmes that give it an on-the-ground advantage. Microsoft operates Azure regions in Johannesburg and Cape Town in South Africa, and in Bahrain for the Middle East. Latency from Nairobi to Johannesburg is comparable to AWS Cape Town, at 50–80 milliseconds. ![Corridor inside a data centre hall](/images/hero-server-hall.webp) Azure's unique advantage in Kenya is Microsoft's deep enterprise relationships. Microsoft products (Windows Server, Active Directory, Microsoft 365, SQL Server, Dynamics) are already deeply embedded in Kenyan banks, government agencies, and enterprises. Azure's integration with these existing systems makes it the path of least resistance for organisations that are already in the Microsoft ecosystem. A bank running SQL Server on-premises can migrate to Azure SQL Database with minimal application changes. A government agency using Microsoft 365 can extend its identity management to Azure Active Directory (now Microsoft Entra ID) for cloud workloads. Azure also benefits from Microsoft's investment in the Kenyan developer ecosystem. The Microsoft Africa Development Centre (ADC) in Nairobi employs hundreds of engineers working on global Microsoft products, and the company's partnership programmes with Kenyan universities, its AI Centre of Excellence, and its engagement with the Kenyan government's digitalisation agenda give it visibility and relationships that AWS and Google Cloud cannot easily match. ### Google Cloud Platform (GCP) Google Cloud is the third hyperscaler and, in Kenya, the smallest of the three by market share. Google Cloud's nearest region to Kenya is also in Johannesburg (launched 2023), with additional regions in Europe and the Middle East. Google Cloud's strength is in data analytics, machine learning, and open-source technologies, areas where Google's internal expertise, built through running its own global services (Search, YouTube, Gmail), translates into superior cloud offerings. For Kenyan organisations, Google Cloud's appeal is primarily in specific use cases: organisations running Kubernetes (Google created and open-sourced Kubernetes), companies building data lakes and analytics pipelines (BigQuery is widely regarded as the best cloud data warehouse), and startups building AI/ML applications (Vertex AI, TensorFlow, and Google's pre-trained models). Google Cloud often has the most aggressive pricing for compute-intensive workloads, which can make it attractive for cost-sensitive startups and research institutions. ## The Local Players While the hyperscalers dominate the conversation, local cloud providers play an important role in Kenya's market, particularly for organisations with strict data residency requirements or those that prefer local support and relationship management. ### Safaricom Cloud Safaricom, Kenya's largest telecom operator, offers cloud services from its own data centres in Nairobi. Safaricom Cloud provides virtual machines, storage, backup, networking, and some managed services, positioning itself as the local alternative to global providers. Its primary value propositions are data residency (data stays in Kenya, on Kenyan soil), low latency (sub-5ms for Kenyan users), local support in English and Swahili, and the ability to bundle cloud with connectivity (fibre and mobile) in a single contract. ![KICC tower and the Nairobi skyline at night](/images/nairobi-skyline-night-kicc.webp) Safaricom Cloud's limitations are real. Its service catalogue is much narrower than the hyperscalers, it lacks the breadth of managed services, AI/ML platforms, serverless computing, and advanced analytics that AWS and Azure offer. Its ecosystem of third-party tools and consulting partners is smaller. And its global reach is, by definition, limited to Kenya. For a Kenyan bank that needs to serve customers across East Africa, or a startup that plans to expand to Nigeria, Safaricom Cloud cannot provide the multi-region deployment that hyperscalers offer. Despite these limitations, Safaricom Cloud has a genuine market. Government agencies subject to data sovereignty requirements, banks that want their core systems in Kenya, and SMEs that value the simplicity of a single local provider all find value in Safaricom's offering. ### Africa Data Centres as a Cloud Enabler Africa Data Centres (ADC), the largest colocation provider in Africa, plays an indirect but important role in the cloud market. By providing carrier-neutral data centre facilities in Nairobi, ADC enables global cloud providers and their partners to establish a physical presence in Kenya without building their own data centres. AWS Direct Connect, Azure ExpressRoute, and Google Cloud Interconnect can all be delivered through ADC facilities, providing Kenyan customers with dedicated, lower-latency connections to cloud regions in South Africa and beyond. ### Managed Service Providers A growing ecosystem of Kenyan managed service providers (MSPs) helps enterprises navigate cloud adoption. Companies like Phase2, Internet Solutions, and various local IT firms offer cloud consulting, migration, and managed services that sit on top of the hyperscaler platforms. These MSPs provide the local expertise and ongoing support that hyperscalers cannot deliver directly, bridging the gap between global cloud platforms and Kenyan enterprise needs. ## Comparing the Options: Key Decision Factors ### Latency Latency is often the first factor organisations consider, and it is where the gap between local and global providers is most tangible. Hosting in a Nairobi data centre provides sub-5ms latency for users in Nairobi, and sub-20ms for users anywhere in Kenya. Accessing cloud services in South Africa adds 50–80ms. For most business applications, this is acceptable. For real-time applications, it matters. ### Data Residency and Sovereignty ![Aerial view of a data centre campus with rooftop solar](/images/dc-environment-sustainability.webp) Kenya's Data Protection Act (2019) places requirements on how personal data is processed and where it can be transferred. While the Act does not absolutely prohibit cross-border data transfer, it requires that transfers be to jurisdictions with adequate data protection standards. For government agencies handling citizen data, for banks regulated by the Central Bank of Kenya, and for healthcare providers handling patient records, the safest approach is to keep data in Kenya. This is a clear advantage for local providers like Safaricom Cloud and for colocation in Kenyan data centres. ### Pricing Cloud pricing is notoriously complex, and direct comparisons between providers are difficult because each offers different services, different pricing models, and different discount structures. As a general guide, compute pricing (virtual machines with equivalent specifications) is broadly similar across AWS, Azure, and GCP for on-demand instances, with the main differentiator being reserved instance and committed-use discounts. Local providers like Safaricom Cloud are competitive for basic virtual machines but may be more expensive for specialised services where hyperscalers achieve economies of scale. ### Ecosystem and Services AWS leads in service breadth (200+ services), Azure leads in enterprise integration and Microsoft ecosystem compatibility, and Google Cloud leads in data, AI/ML, and open-source tooling. Local providers offer a narrower catalogue but may excel in specific local needs (local compliance, local support, integration with local connectivity). ## The Emerging Trend: Hybrid and Multi-Cloud The most significant trend in Kenya's cloud market is the adoption of hybrid and multi-cloud strategies. Rather than choosing a single provider, many organisations are using a combination: keeping latency-sensitive and data-sensitive workloads in local data centres while running batch processing, development, and disaster recovery in global cloud regions. This approach maximises the strengths of each platform while mitigating the weaknesses of any single provider. For Kenyan organisations, the hybrid model is particularly practical. It allows banks to keep core banking systems in Nairobi for low latency and regulatory compliance, while using AWS or Azure for development, testing, data analytics, and disaster recovery. It allows government agencies to keep citizen data onshore while leveraging global cloud capabilities for non-sensitive workloads. And it reduces vendor lock-in by distributing workloads across multiple providers. The data centre infrastructure being built in Kenya, carrier-neutral facilities with Direct Connect capabilities, interconnection to multiple cloud providers, and high-speed fibre connectivity, is the physical foundation that makes these hybrid strategies possible. As the cloud market matures, the quality of local data centre connectivity will increasingly determine how effectively Kenyan organisations can leverage global cloud services while meeting local requirements. ## Colocation in Kenya: Why Companies Rent, Not Build URL: https://data-centers-254.vercel.app/articles/colocation-data-centre-kenya Cluster: Data Centres Published: 2026-08-27 Updated: 2026-09-28 Summary: Colocation lets Kenyan banks, telcos and fintechs house critical systems without building. How Nairobi's carrier-neutral market works and how to choose. Colocation lets you place your own servers inside a professionally built and managed facility. You bring the hardware; the provider supplies the space, power, cooling, connectivity, and physical security. For companies in Kenya that need control over their infrastructure without the capital outlay of building a data centre, colocation is the standard solution. ![Blue-lit server racks in a data hall](/images/hero-server-hall.webp) Banks, mobile network operators, fintech platforms, and government agencies across Kenya use colocation to house their critical systems. The model has matured significantly in Nairobi over the past decade, with multiple Tier III facilities now offering carrier-neutral space along the Mombasa Road corridor. ## What Is Colocation? Colocation (sometimes shortened to "colo") is an arrangement where a company rents physical space in a data centre to house its own servers, storage, and networking equipment. The company retains full ownership and control of its hardware and software. The data centre operator provides the facility infrastructure: rack space, electrical power, cooling, physical security, fire suppression, and internet connectivity. Understanding [what a data centre is](/articles/what-is-a-data-centre) is the first step to understanding colocation. A data centre is a purpose-built facility designed to house IT equipment with reliable power, cooling, and security. Colocation is one of several ways to use that facility. This model is distinct from both cloud computing and building your own data centre. In cloud computing (IaaS), you rent virtualised compute resources from providers like AWS, Azure, or local providers. You do not see or touch the physical hardware. In a self-built facility, you own and operate everything, the building, the power systems, the cooling, the security. Colocation sits between these two extremes: you own the gear, someone else runs the building. The term "carrier-neutral" comes up frequently in colocation discussions. A carrier-neutral facility allows you to connect to any internet service provider, cloud platform, or network carrier you choose. You are not locked into the operator's preferred connectivity. This is a critical distinction in the Kenyan market, where connectivity choice directly affects performance and cost. ### How Colocation Works in Practice The basic unit of colocation is the rack, a standard 42U steel enclosure that houses your servers, switches, and storage devices. Providers typically offer space in fractions: a quarter-rack (approximately 10U of space), a half-rack (approximately 20U), or a full rack. Larger deployments may take entire caged areas or private suites within the facility. When you sign a colocation agreement, you ship your servers to the facility or have your team install them. The provider racks the equipment, connects power cables to their distribution units, and runs network cables from your equipment to a meet-me room where your chosen ISP or carrier terminates their circuits. From that point, your systems are live in a professional facility with redundant power, cooling, and physical security. You retain full access to your equipment. Most facilities allow 24/7 physical access with appropriate authorisation. For tasks that do not require your presence (cable swaps, hard drive replacements, server reboots) the provider's remote hands team acts on your instructions. ## Why Companies Choose Colocation Over Building Their Own Building a data centre in Nairobi is a capital-intensive undertaking that few organisations can justify. A modest Tier III facility with 1 MW of IT load capacity costs upwards of USD 15–20 million to construct, before factoring in land acquisition in Nairobi's industrial zones. Add ongoing operational costs (dedicated facilities staff, power utility contracts, generator fuel, insurance, compliance) and the total cost of ownership becomes prohibitive for all but the largest enterprises. Colocation eliminates most of this. The provider has already absorbed the construction and operational capital. You pay a monthly or annual fee for the space and resources you actually use. ### Capital expenditure versus operational expenditure This is the primary financial driver. Building your own facility requires significant upfront capital expenditure (CapEx): land, construction, electrical infrastructure, cooling systems, fire suppression, security systems, and compliance certifications. Colocation converts this to operational expenditure (OpEx), a predictable monthly cost that scales with your needs. For a Kenyan bank needing to house 50 servers, the choice between spending KES 500 million on a private facility or KES 5–10 million per year on colocation space is straightforward. The bank's capital is better deployed in its core business. ### Reliability and redundancy A properly built colocation facility provides redundancy that is expensive to replicate in-house. Tier III facilities, which represent the standard for Kenyan colocation, offer N+1 redundancy on power and cooling, meaning any single component can fail without affecting operations. This includes dual utility power feeds, diesel generators with automatic transfer switches, UPS systems, and precision air conditioning units. Building this level of redundancy into a private facility requires specialised engineering and ongoing maintenance that most organisations lack the expertise to manage. ### Connectivity and peering Nairobi's colocation facilities sit at the intersection of multiple fibre networks. The Mombasa Road corridor, where most facilities are concentrated, is the primary route for submarine cable traffic from the coast. Kenya connects to the global internet through [submarine cable landing stations in Mombasa](/articles/submarine-cables-landing-mombasa), including the EASSy, TEAMS, SEACOM, and DARE1 cables. Traffic from these cables flows north to Nairobi, where the major data centres aggregate it. Colocation gives you direct access to the [Kenya Internet Exchange Point (KIXP)](/articles/kixp-internet-exchange-point-kenya), multiple ISPs (Safaricom, Wananchi, Liquid Intelligent Technologies, POA Internet, and others), and direct connections to cloud on-ramps for AWS, Azure, and Google Cloud. Peering at KIXP means your traffic to other Kenyan networks stays local, reducing latency and lowering your bandwidth costs. A company building its own facility would need to contract and manage these connections independently, running dedicated fibre from the nearest exchange or cable landing point. In colocation, the cross-connect infrastructure is already in place, and multiple carriers compete for your business within the same building. ![Server racks in a Nairobi data centre hall](/images/dc-gpu-cluster-2.webp) ### Data residency and regulatory compliance The Central Bank of Kenya (CBK) requires financial institutions to maintain data within the country. The National Treasury has issued similar guidance for government data. Colocation provides a straightforward mechanism for compliance: your servers are physically in Kenya, in a known facility, under your control. This is a significant advantage over public cloud, where data residency depends on the provider's infrastructure decisions. With colocation, there is no ambiguity, your hardware is in your rack, in your chosen facility. ### Scalability Colocation scales with your needs. Start with a quarter-rack. Expand to a half-rack, a full rack, or multiple racks as your infrastructure grows. Most providers can accommodate this without requiring you to relocate. In a self-built facility, you either over-provision (paying for space and power you do not yet use) or face a costly expansion when you outgrow the initial capacity. ### Speed to deployment Building a data centre from scratch takes 18–36 months in Kenya, factoring in land acquisition, regulatory approvals, construction, and commissioning. Provisioning colocation space takes days to weeks, depending on the provider's current availability. For organisations that need to deploy infrastructure quickly (whether responding to a new regulation, launching a service, or recovering from a facility failure) this time difference is decisive. ### Access to specialised expertise Operating a data centre requires skills that fall outside most organisations' core competencies: electrical engineering for power systems, mechanical engineering for cooling, physical security management, and facilities maintenance. Colocation providers employ these specialists. Your team focuses on managing servers and applications, not maintaining diesel generators and UPS batteries. ## Colocation Providers in Kenya Kenya's colocation market is served by several providers, each with different strengths, facility locations, and target customers. ### iXAfrica iXAfrica operates NBOX1.1, a purpose-built Tier III data centre located off Mombasa Road in Nairobi. The facility offers carrier-neutral colocation with connectivity to multiple submarine cable systems, ISPs, and the KIXP. NBOX1.1 is the first facility in East Africa to achieve the Uptime Institute Tier III Design certification, giving it a credibility advantage in enterprise procurement processes where independent verification matters. The facility has capacity for significant expansion, with plans for additional data halls. This is relevant for organisations evaluating long-term scalability, a provider with expansion plans is more likely to accommodate your growth than one operating at or near capacity. iXAfrica's colocation offering includes full and half rack options, with power allocations ranging from 2 kW to 10 kW per rack. The facility provides 24/7 remote hands support, biometric access control, and a Network Operations Centre (NOC). Their pricing is competitive for enterprise customers, and their carrier-neutral approach makes them a popular choice for organisations that need connectivity flexibility. ### Africa Data Centres Africa Data Centres, a pan-African operator and subsidiary of the Liquid Intelligent Technologies group, operates multiple facilities across Kenya. Their primary Nairobi facility offers Tier III colocation with the backing of Liquid's extensive fibre network across the continent. Africa Data Centres differentiates through scale and geographic reach. They operate over 20 facilities across Africa, giving them operational experience and procurement advantages that smaller operators cannot match. For organisations operating across East Africa, they offer consistent colocation environments in multiple countries, simplifying regional deployments. Their Nairobi facility provides carrier-neutral space, though their Liquid heritage means seamless integration with Liquid's connectivity services for customers who want a bundled solution. This can be an advantage for organisations that want a single contract covering both space and connectivity, but customers who prefer to choose their own carriers should verify that no commercial pressure is applied. They also operate a facility in Mombasa, positioned near the submarine cable landing stations. This Mombasa facility is particularly relevant for content delivery networks, telecommunications companies, and cloud providers that need proximity to the international cable landing points. Organisations that need both Nairobi and Mombasa presence can negotiate multi-site agreements with Africa Data Centres, simplifying vendor management and potentially securing volume pricing. ### Telecom Operator Facilities Safaricom, Airtel, and the now-merged Telkom Kenya have historically operated their own data centre facilities. While primarily built for internal use, some of these operators offer limited colocation or hosting services to enterprise customers. These arrangements tend to be less carrier-neutral, connectivity is typically skewed toward the operator's own network. ### Smaller and Emerging Providers Several smaller operators provide colocation and hosting services in Nairobi. These include local systems integrators and managed service providers that operate server rooms and small data halls. While useful for cost-sensitive customers or specific use cases, these facilities often lack the formal tier certifications, redundancy, and carrier neutrality of the purpose-built facilities. ### Pricing in the Kenyan Market Colocation pricing in Kenya has become more competitive as new facilities have come online, but it remains higher than equivalent space in markets like South Africa or Europe, reflecting smaller scale, higher power costs, and the relatively limited number of providers. A full 42U rack in a Nairobi Tier III facility typically ranges from KES 50,000 to KES 150,000 per month, depending on the provider, power allocation, and contract length. Half-rack options generally fall between KES 30,000 and KES 80,000 per month. Quarter-rack space is available from approximately KES 15,000 to KES 40,000 per month. Power is the largest variable cost. Standard racks include 4–5 kW of power. High-density configurations for GPU workloads or dense compute clusters command significant premiums. Cross-connect fees add one-time costs of KES 10,000 to KES 50,000 per connection, plus potential monthly recurring charges. Contracts typically run for 1–3 years, with longer commitments attracting better per-rack pricing. Organisations should negotiate power overage rates, remote hands response times, and annual escalation clauses as part of the contract discussion. It is also worth negotiating Service Level Agreements (SLAs) that define guaranteed uptime, response times for support requests, and financial penalties when the provider fails to meet those commitments. ### What to Expect During Onboarding Once you select a provider and sign a contract, the onboarding process typically involves: submitting your equipment list for review, completing security clearance and access registration for your team, scheduling a delivery and installation window, and coordinating cross-connect orders with your network providers. The entire process from contract signature to production deployment usually takes two to six weeks, depending on the complexity of your deployment and the lead times for cross-connects from your chosen carriers. Reputable providers assign a dedicated account manager or onboarding coordinator who manages this process. If a provider treats onboarding as your problem rather than a shared process, it is a warning sign for how they will handle ongoing support. ![Corridor between server racks in a colocation data centre](/images/worldbank-dc-corridor.webp) ## Colocation vs Cloud vs Build-Your-Own: A Direct Comparison The table below compares the three primary approaches to hosting IT infrastructure, specifically for Kenyan organisations. | Factor | Colocation | Cloud (IaaS) | Build Your Own | |---|---|---|---| | **Hardware ownership** | You own and control | Provider owns | You own and control | | **Upfront capital cost** | None (OpEx model) | None (OpEx model) | Very high (CapEx) | | **Monthly cost at scale** | Moderate and predictable | High at sustained workloads | Low (after initial build) | | **Control over hardware** | Full control | No control | Full control | | **Scalability** | Moderate (rack-based) | Very high (elastic) | Limited (requires expansion) | | **Compliance (data residency)** | Direct control | Depends on provider region | Direct control | | **Connectivity flexibility** | High (carrier-neutral) | Provider-managed | High (you choose) | | **Facility reliability** | Provider's tier rating | Provider's responsibility | Your responsibility | | **Staff requirements** | Remote hands available | Minimal | Dedicated facilities team | | **Best suited for** | Regulated enterprises, banks, telcos | Startups, variable workloads | Very large enterprises, government | Cloud computing offers unmatched convenience and elasticity, making it ideal for startups and organisations with variable workloads. But at sustained scale (hundreds of servers running 24/7) colocation is almost always more cost-effective than cloud. Banks, mobile operators, and large enterprises with predictable, always-on workloads typically find colocation delivers better performance per shilling spent. The hybrid model is increasingly common in Kenya. An enterprise might run its core banking platform on owned servers in colocation (for control, compliance, and cost efficiency) while using AWS or Azure for development environments, disaster recovery, and seasonal capacity spikes. This approach gets the benefits of both models without the drawbacks of either alone. ## How to Select a Colocation Provider in Kenya Choosing a colocation provider is a long-term commitment. Relocating servers between facilities is costly, risky, and disruptive. The decision deserves the same rigour as selecting any critical infrastructure partner. ### Facility tier and certification Insist on a minimum of Tier III design, preferably with Uptime Institute certification. This guarantees N+1 redundancy on power and cooling, and concurrent maintainability, meaning the provider can perform maintenance on any single component without shutting down your servers. Ask to see the certification documentation. Several Kenyan facilities claim to be "Tier III" without independent verification. ### Power capacity and pricing Understand exactly how power is billed. Some providers include a fixed power allocation per rack (e.g., 4 kW per rack) in the monthly fee, with additional charges for overage. Others bill power separately based on metered consumption. Clarify whether the quoted price includes utility power, generator backup, and UPS, and what happens during extended outages (how much generator fuel is stored on-site and how long it can sustain full load). Kenya's [power infrastructure](/articles/kenya-power-infrastructure-data-centres) affects colocation operations directly. Kenya Power (KPLC) provides utility power, but outages remain common enough that generator runtime matters. Ask providers about their diesel storage capacity and fuel supply agreements. A facility that can run on generators for 48 hours without resupply is significantly more resilient than one designed for 12 hours. For high-density deployments (AI workloads, GPU clusters), confirm the facility can deliver the per-rack power you need. Standard colocation racks are typically provisioned for 4–8 kW. High-density requirements of 15–30 kW per rack may need specialised arrangements, including high-density cooling and larger power feeds. Not all Kenyan facilities can accommodate these requirements today. ### Connectivity and carrier neutrality A carrier-neutral facility is non-negotiable for most enterprises. Verify that you can connect to your preferred ISPs, cloud providers, and the KIXP without restrictions. Ask about cross-connect fees, some providers charge significant one-time or monthly fees for each connection, which can add up quickly if you need connections to multiple providers. Confirm the facility's proximity to [submarine cable landing stations](/articles/submarine-cables-landing-mombasa) and the diversity of its fibre entry points. A facility with a single fibre path is vulnerable to a single cut. ![Security cameras monitoring a colocation facility](/images/dc-security-camera.webp) ### Physical security Evaluate the security layers systematically. A robust colocation facility implements security in concentric rings: - **Perimeter:** Fencing, barriers, and vehicle access control - **Building:** Mantraps at entry points, CCTV with retention policies (typically 90 days minimum), security personnel - **Data hall:** Biometric or card-based access restricted to authorised personnel, CCTV covering every row - **Rack level:** Individual rack locks (you provide your own padlock), cage options for larger deployments For financial institutions and government agencies, physical security is a formal procurement requirement aligned with CBK guidelines or other regulatory frameworks. Request the facility's security policy documentation and incident response procedures as part of your evaluation. ### Support and remote hands Remote hands service (the provider's on-site staff performing physical tasks on your equipment at your direction) is essential for organisations that do not maintain permanent staff at the facility. Clarify the scope of remote hands (is it included in the base price or billed hourly?), response times, and the technical competency of the on-site team. Tasks like server reboots, cable management, and hardware replacement should be routine for the provider's staff. ### Scalability and contract flexibility Review the provider's capacity roadmap. Can they accommodate your growth over the next 3–5 years without requiring you to move? Understand the contract terms: minimum commitment periods, notice periods for expansion or contraction, and exit clauses. Avoid long lock-in contracts unless the pricing and terms are significantly favourable. ### Compliance and audits If your organisation is subject to regulatory requirements (PCI-DSS for payment processors, CBK guidelines for banks, ISO 27001 for general information security), confirm that the facility supports your compliance needs. This may include providing audit reports, supporting physical access logs, and maintaining their own compliance certifications. ### Location and accessibility Most Kenyan colocation facilities are concentrated along Mombasa Road in Nairobi's Industrial Area and surrounding zones. This location is not accidental, it sits on the primary fibre route from Mombasa and offers proximity to other technology infrastructure. However, the Mombasa Road corridor is also prone to traffic congestion and, in some sections, flooding during heavy rains. Consider how your team will reach the facility. If your engineers need to visit regularly for hardware installation or maintenance, a facility 45 minutes from your office versus 15 minutes has real operational implications. Evaluate road access during peak hours and in adverse weather. Some facilities offer shuttle services or are located in business parks with better access infrastructure. For organisations with staff based outside Nairobi, confirm whether the provider offers remote hands services that can handle routine physical tasks without requiring your presence. ![Data centre security systems](/images/dc-biometric-access-2.webp) ## Common Mistakes When Choosing Colocation Organisations new to colocation often make avoidable errors that prove costly over the life of the contract. **Underestimating power requirements.** Servers are becoming more power-dense. A rack that fits 20 servers today may draw 8–10 kW, exceeding the standard allocation. Plan for your next hardware refresh, not just your current deployment. **Ignoring cross-connect costs.** The rack price is only part of the total cost. Each network connection, cloud on-ramp, and peering arrangement carries its own fees. A deployment requiring connections to five ISPs, two cloud providers, and the KIXP can accumulate significant cross-connect charges. **Choosing a non-carrier-neutral facility.** Locking into a single provider's connectivity limits your negotiating leverage and makes it difficult to switch providers or adopt hybrid architectures. Always confirm carrier neutrality before signing. **Not conducting a physical site visit.** Photographs and specifications do not tell the full story. Visit the facility. Inspect the cable management, check the data hall temperature, review the visitor access process, and talk to the on-site engineers. The condition of the facility tells you as much about the operator as their marketing materials. **Signing overly long contracts without exit provisions.** Technology needs change. Business requirements shift. Mergers, acquisitions, and technology transitions can dramatically alter your infrastructure needs. A three-year contract with no flexibility to reduce space or exit early can become a liability. Ensure your contract includes reasonable expansion, contraction, and termination provisions. **Failing to plan for hardware refreshes.** Servers typically have a 3–5 year lifecycle. When you replace them, you need to decommission old equipment, install new equipment, and potentially reconfigure cabling and power. Discuss hardware refresh procedures with your provider upfront. Some facilities offer staging areas where you can build and test new configurations before cutting over. ## The Role of Colocation in Kenya's Digital Economy Kenya's position as a regional technology hub depends on reliable digital infrastructure. Colocation facilities are a foundational layer of this infrastructure. They host the servers that run Kenya's banking systems, mobile money platforms, e-government services, and enterprise applications. Without colocation, many of these systems would either run in inadequate private facilities or be hosted offshore, undermining data sovereignty and increasing latency for local users. The [licensing framework for data centres in Kenya](/articles/kenya-data-centre-licensing-framework) is evolving to address the growing importance of these facilities. Organisations selecting a colocation provider should verify that the operator holds all required licences from the Communications Authority of Kenya and relevant county government approvals. The growing demand for colocation reflects broader trends. Kenya's fintech sector continues to expand, with mobile lending platforms, payment gateways, and digital banking services all requiring local server infrastructure. The government's push for digital services creates demand for secure, compliant hosting environments. Kenya's power infrastructure, while improving, still makes self-generation and backup power a necessity, colocation providers absorb this complexity and cost. The arrival of hyperscale cloud providers in Nairobi (through on-ramps at colocation facilities) creates new hybrid deployment models where organisations use colocation for core systems and cloud for burst capacity. This hybrid approach is becoming the default architecture for mid-to-large enterprises in Kenya: keep regulated and high-performance workloads on owned hardware in colocation, and use cloud for development, testing, and variable workloads. The [data centre tier ratings](/articles/data-centre-tier-ratings-explained) that define facility reliability are directly relevant to colocation selection. Organisations evaluating colocation providers should understand what each tier guarantees and why it matters for their specific workloads. Kenya's colocation market will continue to grow as more organisations recognise that owning servers does not require owning the building that houses them. The [Kenya data centre directory](/directory) provides a comprehensive listing of facilities and providers to support your evaluation. For most Kenyan enterprises, the question is not whether to use colocation, but which provider and which facility best matches their requirements for reliability, connectivity, scalability, and cost. ## Colocation Data Centres in Nairobi: A Buyer's Guide URL: https://data-centers-254.vercel.app/articles/colocation-data-centres-nairobi-buyers-guide Cluster: Data Centres Published: 2026-09-16 Updated: 2026-09-18 Summary: How to lease rack space in a Nairobi data centre: what Tier III means, which facilities are operational, what to check, and the questions to ask before signing. Somewhere in your office there is a metal cupboard humming with servers. It is too hot in that corner, the power fails on the wrong afternoons, and the only person who knows how to restart the machine is on leave. Colocation is the grown-up version of that cupboard: you keep your servers, but they live in a professional building whose entire job is power, cooling, security and connectivity. Nairobi now has several such buildings to choose from, which is good news, and choosing between them is the hard part. This guide walks the decision in plain English: what exactly you are renting, what the Tier label means, which Nairobi facilities are operational as of September 2026, and a checklist of questions to ask before you sign anything. If you are completely new to the concept, start with our explainer on [what colocation means in Kenya](/articles/what-is-colocation-kenya), then come back here for the buying part. ![Server hall of a Nairobi data centre with rows of racks](/images/hero-dc-nairobi.webp) ## What you are actually renting Strip away the jargon and a colocation deal has four ingredients. First, space: measured in racks or U. A rack is a standard cabinet, and a U (one rack unit) is a slice of that cabinet about 4.4 centimetres tall. A typical full rack offers 42U, and servers are built to stack into those slots like wide books on a shelf. Second, power: delivered per rack or per kilowatt, backed by batteries and generators so a grid failure becomes a non-event. Third, cooling: the machines shed heat constantly, and the facility's chillers and airflow systems fight that battle so you never see it. Fourth, connectivity: fast links to the internet and, crucially, to other networks in the same building. Everything else in the contract, security, fire suppression, access control, support staff, exists to protect those four ingredients. Understanding this split matters because prices are quoted differently by different operators: some bill per rack per month, some per kilowatt of reserved power, and the add-ons (extra power, extra connections, hands-on support) are where totals quietly grow. ![A person using a biometric fingerprint access reader](/images/dc-biometric-access.webp) ## The Tier label, decoded Every facility brochure says Tier III, so it is worth knowing what the word is actually promising. The rating comes from the Uptime Institute, an independent body, and it grades how fault-tolerant a building's infrastructure is. Our full guide to [what Tier ratings actually measure](/articles/data-centre-tier-ratings-explained) covers all four levels, but the practical summary is this. Tier III means concurrent maintainability: any component of the power or cooling chain can be taken out for maintenance without your servers going dark. The facility is engineered for roughly 99.982 percent uptime, which is about one hour of downtime a year. The mechanism behind that promise is redundancy, spelled N+1 in the brochures: one more unit than strictly needed, so a failing battery string, chiller or generator is a maintenance ticket rather than an outage. A [UPS](/articles/ups-backup-power-kenyan-data-centres) (uninterruptible power supply, a room of batteries) bridges the milliseconds between a grid failure and the generators picking up the load. Tier II promises less and Tier IV promises more, with more redundancy and fault tolerance than III. For most Kenyan enterprises, Tier III is the sensible floor: it matches the reliability of the systems most businesses actually run, without paying for Tier IV's industrial-grade extras. ![UPS battery strings inside a data centre power room](/images/dc-ups-batteries.webp) ## Who is actually open in Nairobi right now Numbers first, then names. Kenya's data centre market is small by global standards but concentrated in Nairobi, and the facilities below are operational as of 16 September 2026, per our directory's verification process, which tracks 27 facilities across the country. **Africa Data Centres NBO1** sits at Sameer Business Park, Block A, Mombasa Road, an address confirmed by Africa Data Centres' own site and by PeeringDB's independent listing (September 2025). It is the most densely connected data centre building in Kenya, with four internet exchanges on site, meaning local traffic can be exchanged inside the building rather than detouring abroad. We profile the operator in our piece on [Africa Data Centres' operations in Kenya](/articles/africa-data-centres-kenya-operations). The same group, together with Digital Realty, launched the new **NBO2** facility in September 2026, with about 6.4MW of commissioned capacity reported by trade press (ET Datacenters, 10 September 2026; more in our [NBO2 launch](/articles/digital-realty-nbo2-launch-nairobi) analysis). **iXAfrica NBOX1** is the large Nairobi campus marketing itself as East Africa's first hyperscale, carrier-neutral, AI-ready data centre, with an overall design capacity the operator puts at 22.5MW (iXAfrica, fetched 16 September 2026). Design capacity means what the campus is built to reach, not what is switched on today, so ask what is live when you tour. **iColo Nairobi One**, part of Digital Realty, and **PAIX Data Centres'** Nairobi facility round out the commercial options, alongside earlier-generation facilities and the Kenyatta University-based academic node. Our directory pages carry the per-facility detail, including [ADC NBO1](/directory/africa-dc-nairobi-1) and [iColo NBO1](/directory/icolo-nbo1). One note on names you may see in the market: **Raxio**, a pan-African operator, has announced Kenya as a target market, but as of 16 September 2026 its first Nairobi facility has not opened yet, so for planning purposes it is an upcoming option rather than an open one today. We unpack that situation separately in [Raxio vs Africa Data Centres in Kenya](/articles/raxio-vs-africa-data-centres-kenya-comparison). ![Server racks with neatly organised cabling](/images/racks-cabling.webp) ## The questions that actually matter Brochures grade themselves, so bring your own test. These are the questions that separate a serious facility from a shiny one, and the answers should be specific enough to check. **Power, twice over.** Ask how many independent power paths reach your rack, how many generators sit behind the building, how long the fuel lasts at full load, and when the UPS batteries were last load-tested. The answers should sound routine to them and clear to you. **Cooling, honestly sized.** Ask the maximum kilowatts your rack can draw before cooling becomes a problem. AI-ready claims mean little if your three servers will share air with a rack running hot next door. **Connectivity, counted.** Ask which internet providers and exchange points are physically in the building, and what a cross-connect (a cable from your rack to another network in the same facility) costs per month. Carrier neutrality is only real if several providers are genuinely on site. **Security, walked.** Ask for the full journey a visitor takes from the front door to your rack: mantrap, ID checks, biometric readers, cameras, escort rules. Then ask when it was last audited or certified. **Support, in writing.** Ask what remote hands (technicians who act on your behalf) includes, what it costs, and the response-time promise in the service level agreement. A facility is only as good as the 2 a.m. answer to a dead power supply. ![Diagram of a standard 42U server rack layout](/images/diagram-server-rack-42u.webp) ## Sizing the decision: rack, cage, or neither? Three honest paths exist for a Kenyan organisation outgrowing the office cupboard: 1. **Colocation, starting small.** A quarter or half rack, two to five servers. Monthly cost sits in the range you will discover by quoting, since Kenyan operators rarely publish prices. 2. **A private cage or suite.** This makes sense when you own many machines and need physical separation from other tenants. 3. **Staying out of hardware entirely.** If your actual need is one or two virtual machines, a [Nairobi VPS paid with M-Pesa](/articles/vps-hosting-kenya-mpesa-payment) is cheaper and faster to start, and nothing is lost by beginning there. The economics tip toward colocation when machines are numerous, workloads run around the clock, or compliance rules demand that you control the hardware. Kenya's data protection regime, discussed in our [GPU cloud in Kenya](/articles/gpu-cloud-infrastructure-kenya) guide, is one reason some enterprises prefer physical control of the boxes holding personal data. The economics tip away from colocation when workloads are light, bursty, or experimental, because idle racks earn nobody anything. Nairobi's colocation market has never offered more choice than it does in late 2026: a densely connected veteran on Mombasa Road, a hyperscale-grade campus, a freshly launched NBO2, and focused boutique operators. That competition is good for buyers. Do the site visits, ask the five questions, compare three quotes line by line, and the right facility tends to announce itself. ## Frequently asked questions ### What is colocation in simple terms? Colocation (colocation data centre hosting) means renting space, power, cooling and security inside a professional data centre while keeping ownership of your own servers. The operator provides the building and the infrastructure; you provide and manage the machines. It is the middle ground between running servers in your office and renting fully managed cloud servers. ### How much does colocation cost in Nairobi? Kenyan operators rarely publish colocation prices, so expect to request a quote. Pricing is usually quoted per rack per month, or per kilowatt of power reserved, with cross-connects and remote hands billed separately. As a sizing rule, a half rack suits a small business with a handful of servers, while growing companies typically start at a full rack or a private cage. Get quotes from at least three Nairobi facilities and compare what is included, because two similar-sounding quotes can cover very different things. ### What does Tier III mean, and do I need it? Tier III is a Uptime Institute rating describing a facility built for concurrent maintainability, which means any part of the power or cooling system can be taken out for maintenance without switching your servers off. Tier III facilities are designed for about 99.982 percent uptime, which works out to roughly one hour of downtime per year. Most serious Nairobi facilities are built to this standard. If your systems must stay online through business hours, Tier III or better is the sensible floor. ### What does carrier neutral mean, and why does it matter? A carrier neutral data centre is one that does not force you to buy internet or network services from the facility's owner. Instead, many internet providers and networks install their equipment in the same building, and you connect to whichever you choose. This matters because competition lowers prices, multiple providers mean your connection survives one provider's outage, and networks that meet inside the building exchange traffic faster. In Nairobi this shows up as internet exchange points sitting inside facilities, keeping local traffic local. ### Can I start with less than one full rack? Yes. Most operators sell from a quarter rack or half rack upwards, and some offer single-server hosting. A quarter rack is a small locked slice of one cabinet, enough for two to five servers. That said, if you only need one or two virtual machines rather than physical servers, a Nairobi VPS is cheaper and easier to start with; move into colocation when you need your own hardware, more control, or steady workloads that make owning the machine economical. ## Colocation Pricing in Kenya: What a Rack Costs URL: https://data-centers-254.vercel.app/articles/colocation-pricing-kenya-explained Cluster: Business Guide Published: 2026-09-16 Updated: 2026-09-16 Summary: How Kenyan colocation is priced: per rack, per kilowatt, retail versus wholesale, what drives the bill, and honest benchmark ranges to judge quotes against. Colocation pricing has a gift for hiding in plain sight. The headline rate is real, but it answers a question nobody actually asked. What a Kenyan business really wants to know is: what will this cost me per month, all-in, for the amount of computing I need, with the reliability I am being promised? Getting from the headline to the answer requires understanding four pricing models, three hidden cost centres, and the Kenyan specifics, power tariffs, dollar linkage, redundancy economics, that shape every quote you will receive in Nairobi or Mombasa. This article is the pricing companion to our [Nairobi buyer's guide](/articles/colocation-data-centres-nairobi-buyers-guide). It explains how the industry charges, gives honest orientation ranges from published benchmarks, and shows where Kenyan costs depart from global ones. ![Labelled diagram of a 42U server rack with components](/images/diagram-server-rack-42u.webp) ## The four ways colocation is sold **Per rack unit (per U).** The smallest unit of sale: you rent a slice of a rack, a 1U firewall, a 2U storage node, and the facility handles everything else. Per-U pricing is the retail end of the market and the natural first step for a Kenyan SME moving from an office server cupboard. It is simple, but it is also the most expensive per unit of capacity, because you are buying convenience in slices. **Per rack.** A full rack, the classic 42U cabinet, is the standard mental unit, and published benchmarks keep it honest: internationally a standard 3 to 5 kilowatt rack runs roughly USD 300 to 1,000 per month at the base (ServerMania, 6 January 2026), and about USD 900 to 2,500 all-in per month depending on market, redundancy and what the price includes (QuoteColo, 27 March 2026). Nairobi quotes sit inside those global bands, tending toward the upper half because Tier III power infrastructure in Kenya is genuinely costly to build and run. **Per kilowatt.** The enterprise language. You commit to a power draw, say 100 kilowatts across a cage, and pay a rate per kilowatt per month, with electricity either included or metered on top (DataCenterHawk, 12 February 2026). This is the model that reflects reality, because power is what the facility must provision in redundancy, cooling and UPS capacity. A buyer who understands the per-kilowatt quote can compare any two facilities on earth with one calculation. **Wholesale.** Above a few hundred kilowatts, the conversation stops being about racks at all. Wholesale deals, powered shells or build-to-suit campuses, price megawatts and multi-year commitments. This is where Kenya's hyperscale conversations live, and where the grid's capacity limits, the same ones that stalled the Microsoft and G42 campus, shape what is even available. ## The three hidden cost centres Whatever the model, three line items decide whether a contract is actually cheap. **Electricity.** The biggest operational cost in any facility, typically 30 to 40 percent of opex. In Kenya, industrial consumers pay all-in effective rates in the range of KES 15 to 20 per kilowatt-hour once demand charges and levies are counted (see our [tariffs article](/articles/kenya-power-tariffs-data-centres) for the breakdown). When a quote includes power, you are buying that cost bundled and smoothed; when power is metered to you, the facility's efficiency, its PUE, becomes your bill. A facility at PUE 1.4 moves noticeably less of your electricity into cooling than one at 1.8, and over a year on a loaded rack, that difference is real money. **Cross-connects.** The monthly fee for each cable between you and a carrier, a partner or another tenant. A single connection is trivial; a bank with eight network providers, a DR link and two cloud on-ramps is paying a meaningful recurring line item. Facilities in carrier-rich buildings price cross-connects with confidence because their ecosystem is the product; ask for the full schedule before you model your bill. **Escalations and currency.** Kenyan enterprise contracts commonly carry annual escalation clauses and dollar linkage, because the facility's own costs, imported equipment, and often power contracts, move with the shilling. The question to ask is not "what does it cost today" but "what does the contract do at a ten percent currency move in year three". ![Server racks with status lights in a data hall](/images/dc-servers-racks.webp) ## What drives Nairobi prices above other markets Kenya is not an expensive market by global standards, but it is priced above the cheapest emerging markets, and the reasons are legible. Building Tier III infrastructure, dual power paths, concurrent-maintainable cooling, N+1 or better on everything, costs the same imported capital whether the site is in Nairobi or Ohio, and Kenya's financing costs make the capital dearer. Reliable utility power at KES 15 to 20 per kilowatt-hour all-in is mid-range globally, not cheap, which is why geothermal economics matter so much to the country's data centre ambitions. And the market is still young: smaller absolute demand means facilities spread fixed costs over fewer racks than a Frankfurt operator does. Against that, Kenya offers something the cheapest markets do not: actual interconnection. The carrier ecosystem, the exchange point and the cable landings mean a Nairobi facility is not just racks, it is the shortest path to East African users. Pricing a Nairobi rack against a Johannesburg or Lagos rack purely on rate-per-kilowatt misses that the Nairobi rack may carry a third of the network latency bill to reach regional customers. ## How to compare two quotes in fifteen minutes Bring every quote down to the same three numbers and the marketing falls away. **All-in monthly cost at your real load.** Not at the sales sheet's load, at what you will actually draw, including the metered power at the quoted rate. Ask the facility to model it; a good one will. **Cost per included certainty.** What uptime does the SLA promise and what redundancy backs it? A 99.982 percent Tier III commitment with concurrent maintainability is a different product from a 99.9 percent best-effort room, and the price gap is the cheapest insurance a downtime-prone business can buy. Our [SLA article](/articles/data-centre-sla-uptime-guarantees) explains what the nines are worth. **Total cost of change.** Cross-connects beyond allowance, remote hands rates, move-in fees, escalation percentages. Contracts are won and lost in this section, because it is where year two surprises live. ## The Kenyan ladder, from M-Pesa to megawatts It helps to see colocation as one rung on a ladder the Kenyan market prices honestly. At the bottom, VPS and cloud slices rented monthly, sometimes paid by M-Pesa, are the retail cloud end. Above that, per-U and per-rack retail colocation. Above that, per-kilowatt enterprise cages. At the top, wholesale megawatts. A growing Kenyan business typically climbs that ladder: out of the office cupboard into per-U, into racks when hardware multiplies, into a cage when compliance and scale demand it. Each rung buys more control and more responsibility, and each is priced by the same underlying physics, kilowatts delivered without interruption. The honest summary: colocation in Kenya is not cheap, because reliability is not cheap anywhere and Kenya's inputs are mid-priced. What it can be, if you read the contract like this article does, is predictable, and predictability is what a rack is actually for. ## Dangote's Lamu Power Plant and Kenya Data Centres URL: https://data-centers-254.vercel.app/articles/dangote-lamu-refinery-power-plant-data-centres Cluster: Energy Published: 2026-09-26 Updated: 2026-09-26 Summary: Aliko Dangote announced a 1,000 MW power plant at the proposed Lamu refinery, with 500 MW potentially for the grid. What it means for Kenya's data centre power problem. Africa's richest man has offered Kenya half of a very large power plant. Announcing details during President William Ruto's late-September 2026 visit to his Lekki refinery in Nigeria, Aliko Dangote said the proposed Lamu refinery complex would generate about 1,000 megawatts of electricity and sell roughly 500 megawatts to the Kenyan government, as reported by Kenyans.co.ke on 26 September 2026. "We are going to produce about 1,000 megawatts in Lamu and have 500 megawatts to sell to the Kenyan government," Dangote said. For a country whose largest announced data centre project stalled largely on grid constraints, half a gigawatt of new generation is the kind of number that demands attention, and scrutiny. ![President William Ruto and Aliko Dangote walking past refinery structures at the Dangote Petroleum Refinery in Lekki, Nigeria](/images/dangote-ruto-lekki-refinery-walk.webp) This article was published on 26 September 2026, days after the announcement. It is a developing story: the refinery has not broken ground, financing is not closed, and no power purchase agreement is public. Every figure below is staged the way DC254 stages everything, as an announcement with a named source, not as infrastructure that exists. ## What Was Actually Announced The facts as reported are straightforward. Dangote framed electricity generation as part of the refinery investment, an "entry point" for further businesses. "What this investment will do to Kenya's economy is not only the refinery. The refinery is actually the gate. Once you open the refinery, you will be shocked at how many people will come to invest in Kenya," he said during the visit. President Ruto backed the project, said his administration had already secured land for its development, and described it as an industrial-scale-up opportunity rather than a refinery alone. ![Aerial view of storage tanks and pipelines at the working Dangote refinery complex in Lekki, Nigeria](/images/dangote-lekki-tank-farm-aerial.webp) The project's reported shape, assembled from coverage across Kenyans.co.ke, The Star and international outlets, includes a refinery complex backed by the Africa Finance Corporation, an investment value reported in the range of Ksh 1.94 trillion to Ksh 2.20 trillion (roughly $15 billion to $17 billion), and processing capacity reported between about 220,000 and 700,000 barrels per day depending on the outlet, a discrepancy worth flagging rather than smoothing over. The project was initially earmarked for Tanga in Tanzania before shifting to Lamu, citing infrastructure, logistics and market considerations. Reports in late September suggested a launch or groundbreaking window at the end of September 2026, which, if it happens, will be the first hard evidence that the project is moving. ## The Detail That Matters: It Is an LNG Plant One technical detail from the earlier reporting changes how this story should be read. The Africa Report, writing on 3 September 2026, described Dangote Industries planning to double the capacity of an LNG-fired power plant at the proposed Lamu refinery from 500 megawatts to 1,000 megawatts, with an eye on Tanzanian gas supply. That means the 1,000 megawatts now being discussed is not renewable baseload. It is fossil generation, and its fuel would be imported or regionally sourced natural gas. That detail matters because Kenya's electricity story, and by extension its data centre story, has been built on the opposite proposition. Kenya's grid is roughly 90 percent renewable, anchored by Rift Valley geothermal that supplies roughly 45 percent of generation, with hydro and wind carrying most of the rest. That green grid is a genuine commercial edge in a market where hyperscalers and enterprises face mounting scrutiny over the carbon cost of compute. Half a gigawatt of LNG baseload would not end that edge, but it would dilute the grid-level mix that the edge rests on. ![Geothermal steam wells and plant on the Kenyan Rift Valley floor](/images/kenya-geothermal-plant-2.webp) ## The Scale, in the Context That Counts Put the announced numbers against the grid DC254 tracks. Kenya's installed generation capacity is roughly 3,000 megawatts. The Microsoft and G42 campus, announced in 2024 at around $1 billion and roughly 100 megawatts of IT load, stalled in 2026 after President Ruto himself said the project would have needed about a third of the country's installed capacity, and the National Treasury never approved its funding concept note. Against that arithmetic, 500 megawatts of new supply is material: about one-sixth of today's installed capacity, from a single project. For data centres specifically, the comparison is starker still. DC254's directory currently tracks 27 verified Kenyan facilities with 42.9 megawatts of built capacity at operational sites and 10.5 megawatts of published in-service IT load, inside a 230 megawatt announced pipeline. The half of Dangote's plant destined for the grid is roughly twelve times the built capacity of every operational Kenyan data centre combined. That is the scale gap between the compute Kenya actually runs and the generation now being promised to support what comes next. Two caveats keep the excitement honest. First, generation is not capacity: megawatts produced at Lamu only matter to a data centre if they reach it, which requires a contracted power purchase agreement, transmission investment and dispatch priority, none of which is public. Second, an announcement tied to an unbuilt refinery inherits every schedule risk the refinery carries, and African infrastructure history is full of refinery-scale projects that stayed announcements. Our [Kenya data centre directory](/directory) keeps live and announced capacity strictly separated for exactly this reason. ## Why a Data Centre Site Is Covering a Refinery Because the constraint this announcement speaks to is the one that decides Kenya's data centre decade. Power, not demand, not land, and not capital, is the binding constraint in the directory's own records: the Microsoft-G42 stall was a power story, and operator build-outs across Nairobi are sized by what the grid can supply. If 500 megawatts of firm capacity actually entered the system, it would change what projects are financeable, starting with the stalled gigawatt-scale ambitions. There is also a geographic angle. Lamu sits on Kenya's coast, near but not adjacent to the infrastructure that anchors Kenya's digital economy. Kenya's submarine cables land at Mombasa, the landing stations and fibre routes that carry traffic run through Mombasa and Nairobi, and the market's interconnection, skills and customers are concentrated inland. Our analysis of [why data centres cluster in Nairobi and Mombasa](/articles/why-data-centres-cluster-nairobi-mombasa) explains why cheap power alone has never moved the market: a refinery-town power surplus, however large, needs wires and fibre to become compute. ![High-voltage transmission pylons on Kenya's backbone grid](/images/kenya-transmission-pylons.webp) The third angle is competitive. Ethiopia is exporting regional hydro power to Kenya across a 2,000 MW interconnector and plans to double those exports, a story we track in [our Ethiopia power exports analysis](/articles/ethiopia-power-exports-kenya-data-centres). Olkaria's [geothermal corridor](/articles/olkaria-geothermal-data-centre-corridor) idea pulls compute to the steam fields instead of stretching the grid. Dangote's LNG plant adds a third model: big coastal fossil baseload beside an industrial anchor tenant. Kenya's data centre power future may end up being all three, and the mix will shape both cost and carbon claims. ## The Deal Structure and the Open Questions The commercial architecture around the project is still forming. Reporting in August 2026 indicated Dangote offered Kenya a 10 percent stake worth about $500 million, with a combined 30 percent offered to East African countries, and that Kenya was considering financing its contribution through a local facility. Dangote has said the project would be financed through debt and shareholder funds, with support from the Africa Finance Corporation. Whether any government stake, offtake agreement or power purchase arrangement is concluded, and on what terms, will say more about the project's credibility than any announcement event. ![Labelled artistic impression of the proposed Lamu refinery complex layout](/images/lamu-refinery-layout-illustration.webp) The environmental and heritage questions are equally part of the record. Lamu is a UNESCO World Heritage site, and the area carries the history of the 1,050 MW Amu Power coal plant, cancelled after the National Environment Tribunal revoked its licence in 2019. LNG is a different proposition from coal on emissions, but the permitting history means any large thermal plant at Lamu should be expected to face organised opposition and litigation. Investors, including data centre operators who might one day contract that power, price that risk. ## What to Watch Next Four markers will tell you whether this becomes infrastructure. Watch for a signed power purchase agreement naming the offtaker and the tariff, the single most verifiable signal that the 500 megawatts are real. Watch for financing close, not a launch ceremony, since debt and shareholder commitments are the difference between a groundbreaking photo and a construction site. Watch for the transmission plan that moves Lamu power to demand centres. And watch the fuel chain, because an LNG plant is only firm capacity if the gas supply contracts behind it are firm. Until those land, treat the 1,000 megawatts the way our methodology treats every announced number: as intent, dated and sourced, not as capacity. ## Under Attack: Would You Even Notice a DC Breach? URL: https://data-centers-254.vercel.app/articles/data-centre-attack-scenarios-kenya Cluster: Security Published: 2026-09-16 Updated: 2026-09-16 Summary: Five realistic attack scenarios at a Kenyan data centre: how each unfolds, how fast anyone notices, who responds, and what tenants would actually feel. Here is the uncomfortable truth about data centre attacks: most of them are boring from the inside, and the dangerous ones are silent. The movies show sirens and code scrolling. Reality shows a threshold crossed on a graph at 03:14, a guard reviewing a badge log, or, in the worst cases, nothing at all until a customer asks why their files open in notepad. This article walks five realistic attack scenarios through a Kenyan data centre, one at a time. For each: how it unfolds, who notices and how fast, what the response looks like, and what tenants would actually feel. The scenarios are composites built from documented Kenyan incidents and standard facility operations, and the detection mechanics described are what separates a professional facility from a server room with a lock. ![Server hardware and network equipment inside a running data hall](/images/dc-server-chip-4-wide.webp) ## Scenario one: the flood on the shared pipe **How it unfolds.** At 03:14 on a Sunday, traffic hitting the facility's edge triples, then multiplies. It is a distributed denial of service attack of the kind Anonymous Sudan rained on Kenya's government platforms in July 2023, and the facility's shared upstream links are filling with junk. **Who notices, and when.** Everyone who is watching, immediately. Bandwidth graphs cross alarm thresholds within minutes; the network operations centre gets a pager alert; upstream providers' engineers start seeing the same flood from their side. This is the loudest attack class in existence. If anything, the risk is the opposite of silence: the noise drowns legitimate alerts. **The response.** The NOC calls upstreams to confirm scrubbing is engaged, verifies that customer prefixes are being announced correctly, watches filtering take the junk out upstream, and communicates status to affected tenants. Within an hour the flood is usually still visible on graphs but no longer a customer problem. **What tenants feel.** Brief slowness, or nothing. The July 2023 national attacks showed the difference infrastructure makes: platforms with upstream filtering degraded gracefully while unprepared services went dark (BBC, 28 July 2023). At a well-connected facility, a DDoS is a Tuesday. ## Scenario two: the defacement that goes public first **How it unfolds.** An attacker finds an unpatched admin panel on a public-facing web property and swaps its homepage for a message and a ransom demand, exactly as happened to president.go.ke in July 2026, when five Bitcoin was demanded and the site was defaced with an anti-government message (The Record, 21 July 2026). **Who notices, and when.** Sometimes the public notices first. A defacement is designed to be seen; screenshots spread before the webmaster's coffee loads. Monitoring that checks page content and integrity flags it within minutes, but the embarrassing detection channel is a journalist's phone call. **The response.** Take the property offline, preserve logs as evidence, restore from a known-clean backup, patch the entry point, and reopen. The presidency's site was restored within about two days, which is the right shape: offline fast, back clean, investigate in parallel. **What tenants feel.** Nothing at all, if the defaced property is not theirs. That is the quiet danger: defacement of a shared management interface or the facility's own website can be the visible tip of a deeper intrusion, which is why a professional response treats the defacement as evidence, not just an embarrassment. ## Scenario three: the ransomware that encrypts a tenant's world **How it unfolds.** A tenant's administrator opens a phishing attachment on a Tuesday. By Thursday, the attacker has the tenant's VPN credentials, has spent two nights mapping the network, and begins encrypting every file server and backup the tenant owns. The data centre itself is untouched; the disaster is entirely inside the tenant's environment. **Who notices, and when.** The tenant notices first, when files refuse to open, often days after the intruder got in. The facility's monitoring sees perfectly healthy infrastructure: power stable, cooling stable, network forwarding normally. A compromised virtual machine may only surface as odd traffic patterns, if the facility inspects tenant traffic at all, which most colocation providers deliberately do not, because tenant traffic is tenant business. **The response.** The facility's role is containment support: freezing the tenant's access if asked, providing traffic captures, connecting the tenant to the incident response team, and handling its own perimeter so the attacker cannot pivot into shared services. The tenant's role is recovery: isolate, restore from offline backups, rebuild, and meet its own notification duties. **What tenants feel.** Everything, if it is their environment. This scenario is also where the facility's contract terms matter: who responds, who is liable, what remote hands cost at 4am, and whether the facility can help or only watch. Ransomware is the clearest reason the split of responsibilities must be understood before the incident, not during it. ![A surveillance camera watching over equipment in a secure room](/images/dc-security-camera-2.webp) ## Scenario four: the insider who never trips an alarm **How it unfolds.** A person with legitimate credentials, a contractor, a junior administrator, a departed employee whose account was never disabled, starts querying data they do not need for their job. Nothing crashes. No graph moves. Over weeks, sensitive records quietly leave. **Who notices, and when.** Possibly nobody, ever, unless three specific things exist: logging of what accounts touched what, a baseline of normal behaviour to compare against, and a human who reviews the exceptions. This is the attack class that detection infrastructure exists for, because it defeats every other control. Kenya's registry breach of January 2025 showed how valuable bulk records are; insiders know better than anyone which databases hold them. **The response.** Disable the account, preserve the logs, scope the access with forensic help, and start the legal clock. Here Kenya's frameworks bite: the Computer Misuse and Cybercrimes Act 2018 covers unauthorised access by insiders, and the Data Protection Act 2019 imposes notification duties when personal data is exposed (Bowmans, 23 January 2024). **What tenants feel.** Usually nothing, until the report lands, which is exactly what makes this scenario expensive. Trust is the control here, and trust is built from least-privilege accounts, door logs, escort policies and audits that happen when nothing is wrong. ## Scenario five: sabotage at the cooling plant **How it unfolds.** Someone with access to the plant room stops the chillers, or a compromised building management system ramps cooling down while reporting everything normal. In a packed hall, air temperatures climb; server intake goes hot; thermal protection starts throttling and shutting hardware within minutes. **Who notices, and when.** The building management system alarms on temperature and equipment state within minutes, if its sensors are independent of the system being manipulated, and that "if" is the entire security lesson of the scenario. Environmental monitoring is a security control, not just an operations tool; a facility whose alarms can be silenced from inside the system they watch is a facility trusting its attacker's politeness. **The response.** Operators confirm physically at the plant, start the recovery sequence, and treat it as a deliberate act until proven otherwise, which means access logs, CCTV review and police involvement. Generators and UPS systems ride through the electrical side, but no UPS cools a hall; the cooling chain is its own availability story. **What tenants feel.** A hot aisle, then emergency shutdowns if the event outruns the response. Well-run facilities document maximum hang times and automate the shutdown sequence so hardware survives the event cleanly. Sabotage is rare, but Kenya's 2023 attacks proved that determined opponents will aim at whatever hurts most, and for a data centre, cooling is second only to power. ![Industrial chiller units on a data centre roof](/images/dc-chillers-roof-2.webp) ## So, would we notice? Score the five scenarios on noticeability: the flood, minutes; the defacement, minutes to hours, possibly by the public; the ransomware, days, by the tenant; the insider, maybe never; the sabotage, minutes, if the sensors are honest. The pattern is the answer to the question in the title. You notice exactly the attacks your instrumentation is built to notice, and nothing else. That is why serious facilities converge on the same kit: 24/7 eyes on correlated logs from firewalls, badge readers, CCTV and building systems; thresholds and baselines tuned so anomalies surface; rehearsed response plans with phone numbers, not just policies; and reporting lines to KE-CIRT/CC and the data protection regulator that are practised before they are needed. Kenya's national response team detects hundreds of millions of threat events per quarter, most of them automated and unremarkable, and the reason that number does not translate into national outages is exactly this: detection, layered deep, with humans behind it. The most realistic danger in Kenya is not the attack nobody can stop. It is the attack nobody was watching for, in a building whose staff mistook silence for safety. ## Data Centre Cabling Standards and Practices in Kenya URL: https://data-centers-254.vercel.app/articles/data-centre-cabling-standards-kenya Cluster: Infrastructure Published: 2026-08-28 Updated: 2026-08-28 Summary: A guide to structured cabling standards, copper and fibre best practices, testing protocols, and certification for data centres in Kenya. ![Structured cabling connecting server racks](/images/server-rack-patch-cabling.webp) Structured cabling is the nervous system of any data centre, and getting it wrong in Kenya carries significant operational and financial consequences. From the high-density halls of Africa Data Centres in Nairobi to growing colocation facilities in Mombasa, the quality of cabling infrastructure directly determines network reliability, troubleshooting speed, and the ability to attract enterprise tenants. This guide examines the **data centre cabling standards Kenya** operators follow, the practical challenges of implementation, and the emerging requirements driven by AI workloads. ## Key Cabling Standards Governing Kenyan Data Centres ### TIA-942: The Global Benchmark The **TIA-942** standard (currently at Revision D, published by the Telecommunications Industry Association) is the single most influential cabling standard in Kenyan data centres. It defines the telecommunications infrastructure for data centres, including cabling topology, design, and installation practices. TIA-942 organises data centre spaces into a hierarchical model: - **Entrance Room (ER):** Where external carrier circuits (such as those from [Liquid Intelligent Technologies](https://www.liquid.tech/) or Safaricom) enter the facility - **Main Distribution Area (MDA):** The central hub housing the Main Cross-Connect, core switches, and the primary patch panel fields - **Horizontal Distribution Area (HDA):** Serves individual rows or zones of equipment racks - **Equipment Distribution Area (EDA):** The rack-level area where servers, storage, and network devices connect Kenyan Tier III facilities like **Africa Data Centres' Nairobi campus** and **PAIX (Pan African Internet Exchange)** have adopted TIA-942's structured approach, which ensures scalability and simplified troubleshooting. When operators invest in [data centre design and construction](/articles/data-centre-design-build-kenya), TIA-942 compliance is a fundamental specification. ### ISO/IEC 11801: The International Counterpart The **ISO/IEC 11801** standard, particularly Part 5 (focused on data centres), provides a complementary framework. While TIA-942 is North American in origin, ISO/IEC 11801 is the international equivalent and is often referenced by European-backed operators. In Kenya, **GDCA (Global Data Centre Africa)** and multinational banking tenants frequently request ISO/IEC 11801 compliance. The standard defines generic cabling for customer premises and specifies performance classes: Class D (Cat5e), Class E (Cat6), Class EA (Cat6a), Class F (Cat7), and Class FA (Cat7a). Kenyan new-build facilities typically target **Class EA minimum** for copper and **OM4 or OS2 for fibre**. ## The Cabling Hierarchy: Backbone, Horizontal, and Patch ### Backbone Cabling (MDF to IDF) ![Chart matching breaker ratings to copper cable sizes](/images/diagram-cable-gauge-breaker-chart.webp) Backbone cabling connects the Main Distribution Frame to Intermediate Distribution Frames or Horizontal Distribution Areas. In Kenyan data centres, backbone runs predominantly use **fibre optic cable** due to the distances involved and the bandwidth requirements. - **Single-mode fibre (OS2):** Supports distances up to 10 kilometres at 10G and is used for campus-scale interconnections, such as between the Africa Data Centres facility in Westlands and carrier PoPs along Waiyaki Way - **Multimode fibre (OM3/OM4):** Supports distances up to 100–150 metres at 10G/40G and is common for intra-building backbone links - **OM5 wideband multimode fibre:** Emerging in Kenyan facilities planning for 400G upgrades, supporting multiple wavelengths over a single fibre strand ### Horizontal Cabling (to Racks) Horizontal cabling runs from the HDA patch panels to individual rack positions. In Kenya, **Cat6a** (augmented Category 6) is the current mainstream standard, supporting 10GBASE-T at frequencies up to 500 MHz over distances up to 100 metres. Cat7 is used selectively in premium facilities but requires specialised GG45 or TERA connectors that are harder to source locally. ### Patch Panels and Cable Management Proper patch panel density and labelling are non-negotiable. Kenyan colocation providers like [colocation services in Kenya](/articles/colocation-data-centre-kenya) operators use high-density patch panels (48-port in 1U or 96-port in 2U) to maximise rack space. Cable management accessories, including horizontal and vertical cable managers, bend radius guides, and Velcro ties (never cable ties on data cables), are essential for maintaining airflow pathways and enabling rapid troubleshooting. ![Fibre optic termination equipment in a data centre](/images/dc-fibre-optics.webp) ## Copper vs Fibre: Practical Considerations for Kenya | **Parameter** | **Cat6a Copper** | **Single-Mode Fibre (OS2)** | **Multimode Fibre (OM4)** | |----------------------|-------------------------------------------|------------------------------------------|------------------------------------------| | **Max Distance** | 100 m | 10+ km | 150 m (at 40G) | | **Max Speed** | 10 Gbps | 100 Gbps+ | 100 Gbps (short reach) | | **Cost per Metre** | KES 300–800 | KES 150–400 | KES 200–500 | | **Connector** | RJ45 | LC/SC | LC/SC | | **Power Consumption**| Higher (active electronics) | Lower | Lower | | **Best Use** | Server-to-ToR switch, horizontal runs | MDF-to-IDF backbone, carrier links | Intra-hall backbone, HDA interconnects | ### Sourcing Challenges in Kenya One of the persistent challenges for Kenyan data centre operators is **sourcing certified cabling and components locally**. While basic Cat6 patch cords are widely available in Nairobi's electronics markets (notably Luthuli Avenue and Kirinyaga Road), **certified Cat6a, Cat7, and OM4/OS2 fibre components** are predominantly imported from manufacturers like CommScope, Panduit, Corning, and Belden. Lead times for specialised components can range from 2–8 weeks, which impacts project timelines for new builds. Africa Data Centres and Liquid Intelligent Technologies typically maintain strategic inventory buffers, but smaller operators may face delays. The Kenya Bureau of Standards (KEBS) requires imported cabling to meet relevant IEC standards, and customs documentation must include test certificates from accredited laboratories. ## The Role of Certified Installers in Kenya ### BICSI RCDD and Manufacturer Certifications The **BICSI RCDD** (Registered Communications Distribution Designer) credential is the gold standard for data centre cabling design globally, and its importance is growing in Kenya. RCDD-certified professionals understand the full scope of TIA-942 compliance, including pathway design, grounding and bonding, and cable performance specifications. In the Kenyan market, key installation partners include: - **Dimension Data (now part of NTT):** BICSI-certified teams that have deployed cabling for multiple banking sector data centres - **CSquared:** The fibre infrastructure company provides backbone cabling for several Nairobi facilities - **IntraVision Technologies:** A local integrator with CommScope and Panduit certifications serving East African enterprises - **Safaricom Business:** Offers structured cabling services for enterprise data rooms and small colocation deployments Manufacturer-specific certifications from **CommScope (SYSTIMAX)**, **Panduit**, and **Corning** ensure installers follow approved practices for terminating and testing their respective cabling systems. These certifications are particularly important for warranty claims, a 25-year system warranty from CommScope, for example, is only valid if installation is performed by a certified partner. ## Cable Management in High-Density Environments ![Tangle of unmanaged cables around a distribution cabinet](/images/poor-cable-management-example.webp) Kenyan data centres are increasingly deploying **high-density computing**, with some racks drawing 15–20 kW to support GPU clusters for AI workloads. High-density environments demand meticulous cable management for several reasons: - **Airflow preservation:** Poorly managed cables block under-floor or overhead airflow, creating hot spots. Given that [data centre cooling systems](/articles/data-centre-cooling-systems-explained) already work hard against Nairobi's ambient temperatures, cable obstructions compound the cooling challenge significantly - **Maintenance speed:** In a well-organised facility, a technician should be able to trace any cable from patch panel to server in under five minutes. Disorganised cabling can turn a 15-minute task into a multi-hour exercise - **Capacity planning:** Structured cabling with proper labelling allows facilities to identify unused ports and plan for expansion without deploying new cable runs Best practices adopted by leading Kenyan facilities include: - **Colour-coded cables** by function (e.g., blue for Ethernet, yellow for fibre, red for management networks) - **Pre-terminated trunk cables** to reduce on-site termination errors and speed up deployment - **Overhead cable trays** rather than under-floor routing where possible, separating cabling from cooling airflow - **Automated infrastructure management (AIM)** systems using RFID-tagged patch cords for real-time port mapping ## Nairobi's Climate and Cabling Longevity Nairobi sits at approximately 1,795 metres above sea level, with an average temperature of 19–25°C and relative humidity frequently exceeding 65%. These environmental factors directly affect cabling infrastructure: - **Humidity and corrosion:** High humidity accelerates oxidation of copper contacts, particularly at RJ45 connectors in poorly sealed environments. Gold-plated contacts are essential, and regular inspection schedules should include visual checks for verdigris on exposed connectors - **Heat and cable jacket degradation:** PVC cable jackets become brittle under sustained heat exposure. **Low-Smoke Zero-Halogen (LSZH)** jackets are now the standard specification in Kenyan Tier III facilities, offering better flame resistance and reduced toxic fume emission, a critical safety consideration - **UV exposure:** Facilities with rooftop or exposed cable runs (more common in edge computing deployments in smaller Kenyan towns) must use UV-resistant outdoor-rated cables - **Pest resistance:** Termites and rodents are a real threat to ground-level cable runs in Kenya. Armoured cables and proper conduit sealing are necessary for any at-grade or below-grade installations ## 400G Ethernet and Cabling for AI Workloads ![Long row of server racks with managed fibre cabling in a data centre hall](/images/worldbank-dc-server-row.webp) The rapid growth of **GPU computing and AI workloads** in Kenya (driven by financial services, agritech, and government digitisation initiatives) is accelerating the adoption of **400G Ethernet (IEEE 802.3bs)**. This has profound implications for cabling infrastructure: - **Fibre is mandatory for 400G:** No copper standard supports 400G speeds. Single-mode fibre (OS2) is preferred for distances beyond 100 metres, while OM4 multimode supports 400G at 100 metres using parallel fibre (8-fibre MPO) configurations - **MPO connectors:** Multi-fibre Push On (MPO) connectors (typically 12-fibre or 24-fibre) are essential for 400G parallel optic transceivers. Kenyan facilities upgrading to 400G must invest in MPO patch panels and testing equipment - **Cable bend radius:** Higher-speed fibre is more sensitive to macrobending losses. Installers must maintain minimum bend radius specifications rigorously, particularly in high-density patch panel areas - **Power over Fibre (PoF):** While still emerging, Power over Fibre technology is being evaluated for remote optical transceivers in Kenyan edge data centres, potentially reducing the need for separate power cables to some networking equipment Africa Data Centres' Nairobi facility has already begun provisioning 400G-capable cabling paths in anticipation of growing AI demand, and Liquid Intelligent Technologies has announced plans to upgrade its East African backbone to support 400G interconnects between Nairobi, Mombasa, and Dar es Salaam. ## Testing, Certification, and SLA Implications ### Fluke Testing Protocols Every cable link in a professional Kenyan data centre should be tested and certified using a **Fluke Networks DSX CableAnalyzer** (or equivalent) before being placed into service. The testing process verifies: - **Wiremap:** Correct pin-to-pin connectivity - **Insertion Loss:** Signal attenuation across the link - **Return Loss:** Signal reflected back toward the source - **Near-End Crosstalk (NEXT) and Far-End Crosstalk (FEXT):** Interference between adjacent pairs - **Channel vs Permanent Link testing:** Channel tests include patch cords at both ends; permanent link tests cover the fixed cabling only ### Certification Documentation Certification reports serve multiple critical purposes: 1. **Warranty validation:** Manufacturers like CommScope and Panduit require Fluke test reports to honour 20–25 year system warranties 2. **SLA compliance:** Colocation customers, particularly Kenyan banks operating under Central Bank of Kenya (CBK) data governance requirements, demand certified cabling documentation as part of their due diligence 3. **Troubleshooting baseline:** Test reports from initial installation provide a performance baseline for future comparison if connectivity issues arise 4. **Handover documentation:** When facilities change ownership or management (as occurred with several acquisitions in the Kenyan market), certified cabling records ensure continuity Kenyan operators that maintain rigorous Fluke testing programmes include **Africa Data Centres**, **IXAfrica**, and **PAIX Nairobi**. Smaller facilities sometimes skip comprehensive testing to reduce costs, a practice that frequently leads to intermittent connectivity issues and customer dissatisfaction. ## Real-World Cabling Practices in Kenyan Facilities ### Africa Data Centres, Nairobi The flagship Nairobi campus (formerly the Telkom Kenya facility along Muthangari Drive) features a fully structured cabling design following TIA-942-B topologies. The facility uses a combination of **CommScope SYSTIMAX 360 fibre** for backbone runs and **Cat6a copper** for horizontal distribution. Cable management employs overhead ladder racking with horizontal wire managers at each rack, and all installations are Fluke-certified with full documentation available to colocation tenants. ### PAIX Nairobi The Pan African Internet Exchange facility in Nairobi serves as a critical interconnection point for East African carriers. Its cabling infrastructure prioritises **high-density fibre patching** with MPO-based trunk assemblies supporting the KIXP (Kenya Internet Exchange Point) peering fabric. PAIX maintains colour-coded cabling with strict separation between carrier, peering, and management networks. ### IXAfrica As one of Nairobi's newer carrier-neutral data centres, IXAfrica has invested in **pre-terminated cabling systems** to accelerate deployment times for new tenants. The facility targets TIA-942-C compliance and offers 400G-ready fibre paths as standard in its premium colocation tiers. ## Frequently Asked Questions ### What cabling standard is most commonly used in Kenyan data centres? TIA-942 is the most widely adopted cabling standard in Kenyan data centres, particularly among Tier III facilities like Africa Data Centres and PAIX Nairobi. ISO/IEC 11801 is also referenced, especially by multinational colocation operators. Both standards define the structured cabling architecture, including backbone and horizontal cabling subsystems, that Kenyan facilities follow to achieve international compliance and attract enterprise clients. ### Is it better to use copper or fibre optic cabling in a Kenyan data centre? Both have their place. Copper (Cat6a/Cat7) is cost-effective for short runs up to 100 metres and is commonly used for horizontal cabling to individual server racks. Single-mode fibre (OS2) is preferred for backbone cabling between the Main Distribution Frame (MDF) and Intermediate Distribution Frames (IDF) due to its longer reach and higher bandwidth. Multimode fibre (OM3/OM4) is used for shorter backbone runs within the same hall. For new AI and GPU deployments, fibre is increasingly the default. ### How does Nairobi's climate affect data centre cabling? Nairobi's high humidity (averaging 60–80%) and warm temperatures can accelerate cable jacket degradation, particularly for PVC-sheathed cables. Low-smoke zero-halogen (LSZH) rated cables are strongly recommended to mitigate toxic fume risks. Condensation can also form on exposed patch cords if the cold aisle is significantly cooler than the ambient environment. Proper environmental controls and raised-floor cable routing help manage these risks effectively. ### Why is Fluke testing and certification important for Kenyan colocation providers? Fluke DSX cable testing provides independent, standardised verification that every cable link meets its rated performance (e.g., Cat6a at 10GBASE-T). This certification documentation is essential for colocation providers to guarantee service-level agreements, support warranty claims from cable manufacturers, and demonstrate compliance to enterprise tenants. In Kenya, colocation customers (particularly banks and telecom operators) routinely request Fluke test reports before signing leases. ### What certifications should a data centre cabling installer have in Kenya? The most recognised credentials include BICSI RCDD (Registered Communications Distribution Designer), manufacturer-specific certifications from CommScope, Panduit, and Corning, and CompTIA Network+ for foundational networking knowledge. In Kenya, Africa Data Centres and Liquid Intelligent Technologies typically require BICSI-certified installers for their facility builds. The East Africa Data Centre Association also advocates for structured cabling training through local technical institutions. *For further reading on the telecommunications infrastructure standards referenced in this article, consult the [TIA-942-D Telecommunications Infrastructure Standard for Data Centres](https://www.tia.org/standards/tia-942-d) and the [ISO/IEC 11801-5:2017 standard for data centre cabling](https://www.iso.org/standard/70701.html).* ## Data Centre Careers in Kenya: Jobs and Certifications URL: https://data-centers-254.vercel.app/articles/data-centre-careers-kenya Cluster: Careers Published: 2026-08-26 Updated: 2026-08-27 Summary: Kenya's data centre industry needs facility engineers, cooling specialists, and network architects. The roles, certifications, and entry paths. Kenya's data centre industry is creating jobs that most Kenyans do not know exist. When a young person thinks about a career in technology, they think of software development, cybersecurity, or digital marketing. They do not think about designing cooling systems that prevent servers from melting, managing power distribution for buildings that consume more electricity than small towns, or maintaining the fibre optic networks that connect a country to the global internet. Yet these are the roles that keep Kenya's digital infrastructure running, and they are in growing demand. The data centre industry in Kenya employs hundreds of people across a range of specialisations, from entry-level technicians to senior facility managers with decades of experience. As the industry grows (driven by increasing digitalisation, cloud adoption, and the potential for AI infrastructure investment) the demand for skilled professionals will only increase. For anyone considering a career in this space, understanding the roles, the required qualifications, and the pathways to entry is essential. ![Construction crew in protective gear reviewing a build site, the industry Kenya's data centre careers are built on](/images/construction-workers-site-review.webp) ## The Core Roles: What People Actually Do ### Facility Engineer Facility engineers are responsible for the physical infrastructure of the data centre: power systems, cooling, fire suppression, and the building itself. This is a broad role that encompasses electrical engineering, mechanical engineering, and building management. In smaller facilities, one person may cover all of these. In larger ones, the role is split into specialists. Day-to-day work includes monitoring power and cooling systems, responding to alarms, performing preventive maintenance, and coordinating with utility providers. During facility expansions or upgrades, facility engineers design the infrastructure modifications needed to support additional capacity. **Qualifications**: Degree in electrical or mechanical engineering. Industry certifications like CDCP, or vendor certifications from Schneider Electric, Vertiv, or Eaton, are highly valued. ### Network Engineer Network engineers manage the data centre's connectivity, both the internal network that connects servers to each other and the external network that connects the facility to the internet, submarine cables, and client networks. This role requires deep knowledge of networking protocols (BGP, OSPF, MPLS), fibre optic technologies, and network equipment from vendors like Cisco, Juniper, and Arista. In a colocation facility like iXAfrica or Africa Data Centres, network engineers also manage cross-connects between different clients and ensure that each customer's network is properly isolated and performing optimally. **Qualifications**: Degree in telecommunications, IT, or computer science. CCNA, CCNP, or equivalent certifications are typically required. ### Operations Manager Operations managers oversee the day-to-day running of the facility. They coordinate between engineering teams, manage shift schedules, handle client onboarding, and ensure that service level agreements (SLAs) are met. This is a senior role that requires both technical understanding and strong management skills. In practice, the operations manager is the person who decides what happens when something goes wrong, and they need enough technical knowledge to make the right call quickly. **Qualifications**: Typically 5-10 years of experience in data centre operations, often starting in an engineering role. An MBA or operations management qualification is an advantage. ### Cabling Technician Cabling technicians install, maintain, and troubleshoot the structured cabling that connects everything in the data centre. This includes copper and fibre optic cables, patch panels, and cable management systems. It is one of the most accessible entry points into the industry. Good cabling work requires precision, patience, and attention to detail. A single poorly terminated fibre connector can cause signal loss that degrades performance for an entire rack of servers. **Qualifications**: Diploma in IT, telecommunications, or electrical engineering. Vendor certifications for fibre optic termination (e.g., Corning, CommScope) are valued. ## Certifications That Matter The data centre industry has a well-established certification framework that helps professionals demonstrate their knowledge and advance their careers. ### CDCP, Certified Data Centre Professional The CDCP, offered by the Infrastructure Masons Academy (formerly the Uptime Institute), is the most widely recognised entry-level data centre certification globally. It covers data centre design concepts, infrastructure components, operations best practices, and maintenance procedures. The certification requires passing a multiple-choice exam and is valid for three years. For anyone in Kenya looking to enter the data centre industry, the CDCP is the single most valuable certification to obtain. It is recognised by all major operators and demonstrates a foundational understanding of how data centres work. ### Vendor-Specific Certifications Manufacturers like Schneider Electric, Vertiv, and Eaton offer certifications for their specific products and systems. These are particularly valuable for engineers who will work with specific equipment. For example, a facility using Schneider Electric UPS systems and Prisma management software would benefit from engineers who hold Schneider's Data Centre Certified Associate (DCCA) certification. ### Networking Certifications Cisco's CCNA and CCNP remain the gold standard for network engineers globally, including in Kenyan data centres. For fibre optic specialists, certifications from the Fibreoptic Industry Association (FIA) or vendor-specific programmes from Corning or CommScope are relevant. ### Safety Certifications Working in data centres involves high-voltage electrical systems, confined spaces, and heavy equipment. Safety certifications, including occupational health and safety training, are mandatory for most roles and are often a legal requirement. ![Comparison of network engineer and cybersecurity engineer roles](/images/diagram-network-vs-cyber-career.webp) ## The Salary Landscape Data centre salaries in Kenya vary significantly by role, experience, and employer. International operators (iXAfrica, Africa Data Centres, and potentially Microsoft) tend to pay above market rates, while local operators and government facilities may pay less. | Role | Entry Level (KES/mo) | Mid-Level (KES/mo) | Senior (KES/mo) | |---|---|---|---| | Cabling Technician | 40,000–60,000 | 60,000–100,000 | 100,000–150,000 | | Network Engineer | 60,000–100,000 | 100,000–200,000 | 200,000–400,000 | | Electrical Engineer | 60,000–100,000 | 120,000–250,000 | 250,000–500,000 | | Cooling Engineer | 60,000–100,000 | 120,000–250,000 | 250,000–500,000 | | Operations Manager |, | 150,000–300,000 | 300,000–600,000+ | | Facility Manager |, | 200,000–400,000 | 400,000–800,000+ | These figures are approximate and based on industry reports and job listings. Senior roles at international operators can exceed these ranges, particularly when including benefits like medical insurance, housing allowances, and professional development budgets. ## How to Enter the Industry ![Students learning computer skills in a Kenyan classroom](/images/classroom-ict-training-kenya.webp) The most common pathways into Kenya's data centre industry are: **Engineering graduates.** A degree in electrical, mechanical, or telecommunications engineering is the most direct entry point. Most large operators run graduate programmes or hire junior engineers. Adding a CDCP certification significantly improves employability. **IT professionals transitioning.** Network engineers, systems administrators, and IT support staff can transition into data centre roles, particularly in networking and operations. The key gap to fill is understanding the physical infrastructure, power, cooling, and facilities management. **Trade and vocational.** Cabling technicians, electricians, and HVAC technicians can enter the industry with relevant trade qualifications and on-the-job training. These roles offer a practical pathway to senior positions through experience and additional certifications. **Military and telecom veterans.** People who have worked on communications infrastructure in the military or at telecom companies (Safaricom, Telkom Kenya) often have directly transferable skills. The [data centre tier ratings](/articles/data-centre-tier-ratings-explained) framework is important knowledge for anyone entering the field, understanding the reliability standards that facilities are built to will help you understand why things are designed the way they are. ![Field technician working on aerial fibre connections on a utility pole](/images/aerial-fibre-technician-utility-pole.webp) ## The Outlook Kenya's data centre industry is at an inflection point. The combination of regulatory clarity from the [licensing framework](/articles/kenya-data-centre-licensing-framework), growing demand from cloud and AI, and the country's geothermal energy advantage creates a strong foundation for job growth. The constraint, as always, is power supply, but as grid infrastructure improves, the industry will create hundreds of skilled, well-paying jobs. For anyone considering this career path, the timing is good. The industry is small enough that capable professionals are noticed quickly, but growing fast enough that there is real demand. Start with the CDCP certification, build practical skills through training or entry-level roles, and develop specialisation in an area like power, cooling, or networking. The career path from there (to senior engineer, operations manager, and eventually facility director) is well-established and well-compensated. ![Nairobi skyline lit up at night](/images/nairobi-skyline-night.webp) ## Data Centre Certifications Worth Having in Kenya URL: https://data-centers-254.vercel.app/articles/data-centre-certifications-kenya Cluster: Careers Published: 2026-08-28 Updated: 2026-09-28 Summary: Guide to data centre certifications in Kenya, CDCP, CDCS, CCNP, AWS, Azure, PMP, with KES costs, local trainers, exam requirements, career impact. In Kenya's rapidly expanding data centre industry, certifications are more than just credentials on your CV, they are the currency of career advancement. Whether you're a [recent graduate looking to break into the sector](/articles/how-to-get-job-kenyan-data-centre) or an experienced professional targeting a leadership role, the right certifications can accelerate your journey significantly. This comprehensive guide covers every data centre-relevant certification available to Kenyan professionals, organised by speciality area. For each certification, you'll find the cost in Kenyan Shillings, where to take training locally, exam requirements, and the real career impact as observed in the Kenyan market. ![Electrical switchgear and UPS cabinets](/images/dc-ups-batteries-wide.webp) *Certifications are the fastest way to advance your data centre career in Kenya* ## Infrastructure Certifications Infrastructure certifications are the most directly relevant to day-to-day data centre operations in Kenya. They cover the design, build, and management of the physical facility, power, cooling, cabling, and physical security. ### EPI CDCP, Certified Data Centre Professional The **CDCP** from [EPI](https://www.epi-ap.com/) is widely regarded as the foundational certification for anyone working in a data centre in Africa. It covers all major disciplines: design concepts, power systems, cooling, cabling, physical security, fire protection, and operations management. - **Cost in Kenya:** KES 120,000–160,000 (training + exam bundled) - **Exam Requirements:** 2-day instructor-led course followed by a 1-hour exam (40 multiple-choice questions, 65% pass mark) - **Availability in Kenya:** Regularly offered in Nairobi through Strathmore @iLabAfrica and EPI's certified training partners. Online instructor-led options are also available. - **Career Impact:** The CDCP is often listed as a minimum requirement for mid-level operations roles at Africa Data Centres, Liquid, and PAIX. It can increase starting salary offers by 15–25% and is essentially mandatory for operations manager positions. ### EPI CDCS, Certified Data Centre Specialist The **CDCS** is the mid-tier EPI certification, designed for professionals with 2+ years of data centre experience. It dives deeper into design, operations, and maintenance, with a focus on optimising availability, efficiency, and capacity. - **Cost in Kenya:** KES 150,000–190,000 (training + exam) - **Exam Requirements:** 3-day course, 1.5-hour exam (60 questions, 65% pass mark). CDCP certification is a prerequisite. - **Availability in Kenya:** Offered in Nairobi 3–4 times per year through Strathmore and EPI partners. - **Career Impact:** The CDCS is the mark of a serious mid-career professional. It's frequently required for Operations Manager and Senior Engineer roles. Professionals with CDCS typically earn 25–35% more than those with only CDCP. ### EPI CDCE, Certified Data Centre Expert The **CDCE** is EPI's top-tier certification, targeting senior professionals involved in strategy, design, and governance of data centre portfolios. It covers business continuity, risk management, financial modelling, and multi-site management. - **Cost in Kenya:** KES 200,000–250,000 (training + exam) - **Exam Requirements:** 3-day course, 2-hour exam (60 questions, 70% pass mark). CDCS is a prerequisite. - **Availability in Kenya:** Less frequently offered locally, may require attending sessions in Johannesburg, Dubai, or taking the online version. - **Career Impact:** The CDCE positions you for Director and VP-level roles. It's held by a small but growing number of senior professionals in the Kenyan market and signals readiness for strategic leadership. ### Uptime Institute ATD, Accredited Tier Designer The [Uptime Institute](https://uptimeinstitute.com/education/) **ATD** certification validates the ability to design data centre facilities to meet specific Tier Standard requirements. It is globally recognised and particularly valued for professionals involved in new facility design and build projects. - **Cost in Kenya:** KES 350,000–500,000 (training + exam, often including travel if taken outside Kenya) - **Exam Requirements:** 2-day course plus exam. No formal prerequisites, but significant design experience is recommended. - **Availability in Kenya:** Not regularly offered in Nairobi. Most Kenyan professionals attend sessions in Johannesburg, London, or take the virtual format. - **Career Impact:** The ATD is highly prestigious and relatively rare in the Kenyan market, making it a significant differentiator. It's particularly valuable for professionals at consulting firms (Arup, WSP, SMEC) working on data centre design projects in East Africa. ### Uptime Institute ATS, Accredited Tier Specialist The **ATS** focuses on the operational side of Tier Standards, how to maintain and operate facilities to their designed Tier level. It's ideal for operations managers and senior engineers responsible for uptime performance. - **Cost in Kenya:** KES 350,000–500,000 (training + exam) - **Exam Requirements:** 2-day course plus exam. - **Availability in Kenya:** Similar to ATD, primarily available virtually or in South Africa. - **Career Impact:** Strong credential for senior operations roles. The Uptime Institute brand carries significant weight with international colocation operators and hyperscale customers. ![Corridor inside a data centre hall](/images/hero-server-hall.webp) *Certified professionals command higher salaries and greater responsibility in Kenyan data centres* ## Networking Certifications Networking is the nervous system of any data centre. Whether it's managing the internal spine-leaf fabric or external peering at KIXP, networking certifications are among the most consistently valuable credentials in the industry. ### Cisco CCNA, Cisco Certified Network Associate The **CCNA** remains the most recognised entry-level networking certification globally and is widely valued in Kenyan data centres. It covers network fundamentals, IP connectivity, security fundamentals, automation, and programmability. - **Cost in Kenya:** KES 35,000–55,000 (exam only: approximately KES 18,000–22,000) - **Exam Requirements:** Single exam (200-301), 105 minutes, no formal prerequisites - **Availability in Kenya:** Exams available at Pearson VUE test centres in Nairobi (including at Strathmore). Training available through numerous local providers. - **Career Impact:** The CCNA is considered a minimum qualification for network-related roles in Kenyan data centres. It opens doors to NOC analyst and junior network engineer positions. ### Cisco CCNP Enterprise The **CCNP Enterprise** is the professional-level Cisco certification, covering advanced routing, switching, wireless, and security topics. It's particularly relevant for data centre network engineers working with Cisco ACI (Application Centric Infrastructure). - **Cost in Kenya:** KES 80,000–130,000 (training for both exams; exams cost approximately KES 18,000–22,000 each) - **Exam Requirements:** Two exams, a core exam (350-401) and one concentration exam - **Availability in Kenya:** Exams at Pearson VUE Nairobi. Training available locally and online. - **Career Impact:** CCNP holders in Kenya can expect salary premiums of 30–50% over CCNA-only professionals. It's a strong qualification for senior network engineer roles at operators and enterprises. ### Juniper JNCIA **JNCIA** (Juniper Networks Certified Associate) is gaining relevance as Juniper equipment appears in more African data centre builds, particularly through partnerships with colocation operators. - **Cost in Kenya:** KES 15,000–20,000 (exam only) - **Exam Requirements:** Single exam, available via Pearson VUE - **Availability in Kenya:** Exams available at Pearson VUE Nairobi - **Career Impact:** Niche but valuable, especially at facilities deploying Juniper fabric architecture. ### CompTIA Network+ **CompTIA Network+** is a vendor-neutral networking certification that provides excellent foundational knowledge. It's often recommended as a stepping stone before CCNA. - **Cost in Kenya:** KES 35,000–50,000 (exam only: approximately KES 22,000–28,000) - **Exam Requirements:** Single exam (N10-008), no formal prerequisites - **Availability in Kenya:** Pearson VUE Nairobi, extensive online study resources - **Career Impact:** Good entry point for [data centre careers in Kenya](/articles/data-centre-careers-kenya), particularly for those transitioning from general IT or fresh graduates. ![Chart of common network protocols including HTTP, DNS, SSH and TCP](/images/diagram-network-protocols.webp) *Networking certifications remain the backbone of data centre career advancement* ## Power and Cooling Certifications Power and cooling represent the largest operational cost in any Kenyan data centre. Professionals with specialised certifications in these areas are in high demand and short supply. ![UPS and switchgear line-up in a data centre power room](/images/ups-switchgear-power-room.webp) ### Certified Energy Manager (CEM) The **CEM**, offered by the Association of Energy Engineers (AEE), covers energy management principles including power systems, efficiency optimisation, and renewable energy integration, all critical for data centres in Kenya where energy costs and reliability are major concerns. - **Cost in Kenya:** KES 150,000–200,000 (training + exam) - **Exam Requirements:** 4-day training programme followed by exam. A relevant engineering degree plus experience is recommended. - **Availability in Kenya:** Occasionally offered in Nairobi through AEE-certified training providers. Online options are available. - **Career Impact:** The CEM is particularly valuable in the Kenyan context where data centre operators are focused on reducing PUE (Power Usage Effectiveness) and integrating renewable energy sources like geothermal power. ### Vendor-Specific Certifications: Schneider Electric Schneider Electric is a dominant player in data centre power and cooling infrastructure across Africa. Their certifications include: - **Schneider Electric Certified Data Centre Infrastructure Specialist**, Covers APC by Schneider Electric UPS, cooling, and rack systems - **Cost in Kenya:** KES 80,000–150,000 (varies by programme) - **Availability:** Offered through Schneider's partner network in Kenya, including training events at distributor facilities - **Career Impact:** Highly practical, these certifications directly apply to the equipment installed in most Kenyan data centres ### Vendor-Specific Certifications: Vertiv Vertiv (formerly Emerson Network Power) offers certifications covering precision cooling, power management, and monitoring systems that are widely deployed across East African data centres. - **Cost in Kenya:** KES 60,000–120,000 - **Availability:** Through Vertiv's regional training programmes, sometimes offered in Nairobi - **Career Impact:** Vertiv-certified technicians and engineers are valued at facilities where Vertiv equipment is the primary infrastructure ### Vendor-Specific Certifications: Eaton Eaton provides certifications for UPS systems, power distribution units, and electrical safety that are relevant to data centre operations. - **Cost in Kenya:** KES 50,000–100,000 - **Availability:** Online and occasional in-person sessions - **Career Impact:** Useful complement to broader certifications for power systems specialists ![Electrical switchgear and UPS cabinets](/images/dc-ups-batteries-wide.webp) *Power and cooling certifications are increasingly valuable as data centre densities rise* ## Security Certifications With the Kenya Data Protection Act 2019, increasing cybersecurity threats, and the physical security requirements of [data centre facilities](/articles/data-centre-security-explained), security certifications are increasingly important. Note that facility-level ISO 27001 certification, which our [ISO 27001 guide](/articles/iso-27001-data-centre-kenya) covers, is earned by the operator and is a different track from the personal certifications below. ### CISSP, Certified Information Systems Security Professional The **CISSP** from (ISC)² is the gold standard for information security professionals. While not data-centre-specific, it's highly valued for data centre security architects and compliance leaders. - **Cost in Kenya:** KES 80,000–120,000 (exam only: approximately KES 55,000–70,000; training courses add significantly) - **Exam Requirements:** Requires 5 years of cumulative paid work experience in two or more of the eight CISSP domains - **Availability in Kenya:** Exams at Pearson VUE Nairobi. Training available online and through local cybersecurity firms like Serianu. - **Career Impact:** CISSP holders command premium salaries in the Kenyan market, typically KES 250,000–500,000+ per month for experienced professionals. It's increasingly required for security leadership roles at data centre operators. ### CompTIA Security+ **Security+** provides foundational cybersecurity knowledge and is an excellent starting point for professionals moving into data centre security roles. - **Cost in Kenya:** KES 35,000–55,000 (exam only: approximately KES 22,000–28,000) - **Exam Requirements:** No formal prerequisites, though Network+ is recommended - **Availability in Kenya:** Pearson VUE Nairobi, extensive online training options - **Career Impact:** Strong entry-level security certification, valued for NOC security analysts and junior security roles ### Certified Ethical Hacker (CEH) The **CEH** from EC-Council focuses on offensive security, understanding how attackers think to better defend systems. It's relevant for data centre security teams conducting vulnerability assessments and penetration testing. - **Cost in Kenya:** KES 60,000–100,000 (training + exam) - **Exam Requirements:** 2 years of information security experience recommended - **Availability in Kenya:** Available through EC-Council's training partners and online - **Career Impact:** Valuable complement to CISSP for professionals targeting security architect roles ## Cloud Certifications As cloud services expand across Kenya and East Africa, cloud certifications are becoming essential for data centre professionals, especially those working at or with hyperscale providers. ### AWS Certified Solutions Architect, Associate The **AWS Solutions Architect Associate** is one of the most in-demand cloud certifications globally. It validates the ability to design distributed systems on AWS. - **Cost in Kenya:** KES 55,000–100,000 (exam: KES 18,000–22,000; training varies) - **Exam Requirements:** 1 year of hands-on experience with AWS recommended - **Availability in Kenya:** Exams available online via Pearson VUE. Free digital training through AWS Training. Instructor-led training available through providers like Strathmore. - **Career Impact:** AWS certification is particularly valuable as AWS expands its African presence. Certified professionals can access the AWS Partner network, which is a significant career accelerator. ### Microsoft Certified: Azure Administrator Associate (AZ-104) The **Azure Administrator Associate** validates skills in managing Azure cloud services, including storage, compute, networking, and identity, all relevant to hybrid cloud data centre environments. - **Cost in Kenya:** KES 50,000–90,000 (exam: KES 15,000–20,000; training varies) - **Exam Requirements:** 6 months of Azure administration experience recommended - **Availability in Kenya:** Exams via Pearson VUE. Microsoft Learn provides free preparation materials. - **Career Impact:** Strong demand in Kenya's enterprise market, where Microsoft's ecosystem is dominant among banks, government agencies, and large corporations. ### Google Cloud Associate Cloud Engineer The **Google Cloud Associate Cloud Engineer** certification validates the ability to deploy applications, monitor operations, and manage enterprise solutions on Google Cloud. - **Cost in Kenya:** KES 50,000–85,000 (exam: KES 15,000–18,000; training varies) - **Exam Requirements:** 6+ months of hands-on Google Cloud experience recommended - **Availability in Kenya:** Online exams via Pearson VUE. Google Cloud Skills Boost offers free and paid training. - **Career Impact:** Growing relevance as Google Cloud expands its African infrastructure and partnerships. ## Project Management Certifications Project management certifications are essential for professionals moving into leadership and management roles, particularly those overseeing data centre builds, expansions, and major upgrades. ### PMP, Project Management Professional The **PMP** from the Project Management Institute (PMI) is the most recognised project management certification globally. For data centre professionals, it's particularly valuable for those managing construction projects, technology migrations, and facility expansions. - **Cost in Kenya:** KES 100,000–180,000 (exam: KES 40,000–55,000; training courses KES 60,000–120,000) - **Exam Requirements:** 36 months of project leadership experience with a four-year degree, or 60 months with a secondary school diploma. 35 hours of project management education. - **Availability in Kenya:** Exams at Pearson VUE Nairobi. Training available through multiple providers including Strathmore, PMI Kenya Chapter events, and online platforms. - **Career Impact:** The PMP is a significant differentiator for operations managers and directors. It's often listed as a requirement for senior roles at Africa Data Centres, Liquid, and in consulting firms working on data centre projects. ### PRINCE2 **PRINCE2** (Projects IN Controlled Environments) is widely used in Commonwealth countries and by organisations with UK or European ties. It's structured, process-based, and particularly valued at organisations with a British or European operational heritage. - **Cost in Kenya:** KES 80,000–150,000 (Foundation + Practitioner) - **Exam Requirements:** No formal prerequisites for Foundation level - **Availability in Kenya:** Offered by training providers in Nairobi and online - **Career Impact:** Valuable at organisations like PAIX Data Centres (with its Teraco/UK links) and European-backed investors in African data centres ## Local Training Providers in Kenya Kenya has a growing ecosystem of training providers for data centre certifications: **Strathmore University (@iLabAfrica):** The leading local provider for EPI certifications (CDCP, CDCS, CDCE) and a Pearson VUE test centre. They also offer cloud and cybersecurity training aligned to industry certifications. **KENET (Kenya Education Network):** Primarily serves the education sector but facilitates training programmes for member institutions and occasionally opens sessions to industry professionals. **Liquid Intelligent Technologies Academy:** Offers internal and partner-facing training programmes covering networking, cloud, and data centre operations. **Safaricom's Learning and Development:** Internal programmes for staff, but Safaricom also sponsors external certifications for employees. **Online Platforms:** Udemy, Coursera, CBT Nuggets, and Pluralsight offer extensive preparation courses. While not Kenya-specific, they're widely used by Kenyan professionals for self-study before sitting proctored exams at Pearson VUE centres in Nairobi. ## Scholarship and Employer Sponsorship Options The cost of certifications can be a barrier, but several avenues exist for financial support: ### Employer Sponsorship Most established data centre operators in Kenya have training budgets. **Africa Data Centres** has a formal professional development framework that includes certification sponsorship tied to performance reviews. **Liquid Intelligent Technologies** runs a structured talent development programme. **Safaricom** provides education assistance benefits. The key is presenting a clear business case, demonstrate how the certification directly benefits your team's performance and the organisation's objectives. ### Vendor Programmes **AWS** offers free digital training through AWS Skill Builder, and their Re/Start programme has been piloted in African markets. **Microsoft** provides free Azure Fundamentals training and exam vouchers through various community programmes. **Google Cloud** offers scholarships through their Google Career Certificates programme. ### Professional Bodies The **PMI Kenya Chapter** occasionally offers discounted training and exam vouchers. The **ISACA Kenya Chapter** provides resources for cybersecurity certifications. **IEEE Kenya Section** hosts workshops and sometimes facilitates certification preparation. ### Government and Development Programmes Programmes under the **Kenya ICT Board** (now ICT Authority) and various development partners occasionally fund certification training as part of digital skills initiatives. The **World Bank's Digital Development Partnership** has supported ICT training programmes in East Africa that include data centre-relevant certifications. ## Certification ROI in the Kenyan Market The return on investment for data centre certifications in Kenya is strong, but varies by certification type and career level: - **Infrastructure certs (CDCP/CDCS):** 20–40% salary premium; payback within 6–12 months - **Networking certs (CCNP):** 30–50% salary premium; payback within 6–9 months - **Cloud certs (AWS/Azure):** 25–45% salary premium; payback within 3–6 months (lower cost, high demand) - **Security certs (CISSP):** 40–60% salary premium; payback within 12–18 months (higher investment) - **Project management (PMP):** 20–35% salary premium; payback within 9–15 months The bottom line: in Kenya's fast-growing data centre market, certifications are not optional for serious professionals. They are the most reliable investment you can make in your career, and the [career progression paths](/articles/career-progression-african-data-centres) in this industry are significantly smoother (and faster) for certified professionals. ## Frequently Asked Questions ### Which data centre certification should I get first in Kenya? For most professionals entering the Kenyan data centre industry, the **EPI CDCP (Certified Data Centre Professional)** is the best starting certification. It provides a comprehensive foundation covering all aspects of data centre design, operations, and management, from power and cooling to cabling and security. The CDCP is widely recognised across Africa by operators like Africa Data Centres, Liquid Intelligent Technologies, and PAIX. It costs approximately KES 120,000–160,000 including training, and courses are regularly offered in Nairobi. For those on a tighter budget, **CompTIA Network+** (KES 35,000–50,000) is an excellent alternative that builds foundational networking knowledge relevant to any data centre role. ### Are international data centre certifications worth the cost in Kenya? Yes, international certifications consistently deliver strong return on investment in the Kenyan market. Professionals with CDCP or CDCS certifications typically earn 20–40% more than uncertified peers in equivalent roles. For [data centre career progression](/articles/career-progression-african-data-centres), certifications are often non-negotiable at the mid-level and above, operations managers at Africa Data Centres and Liquid Intelligent Technologies are typically required to hold at least CDCS. Cloud certifications (AWS, Azure) are particularly valuable because hyperscale providers are expanding into Kenya and East Africa, and they require certified professionals for their partner ecosystem. The key is to choose certifications aligned with your career path and the specific demands of the Kenyan market. ### Where can I take data centre certification training in Kenya? Several options exist in Kenya. **Strathmore University** through its @iLabAfrica centre regularly hosts EPI data centre certification training (CDCP, CDCS, CDCE) and is one of the most reliable local providers. **KENET** (the Kenya Education Network) occasionally facilitates training sessions for its member institutions. **Liquid Intelligent Technologies** runs internal and partner-facing certification programmes. For vendor-specific certifications, **APMG International** partners with local training firms to deliver courses in Nairobi. For self-study, online platforms like Udemy, Coursera, and CBT Nuggets offer quality preparation material, though you'll still need to sit the proctored exam. Many international training providers like EPI and the Uptime Institute also offer live online instructor-led training that you can attend from Kenya. ### Can my employer sponsor my data centre certification in Kenya? Many employers in the Kenyan data centre industry do sponsor certifications, particularly at established operators. **Africa Data Centres** has a structured professional development programme that includes certification sponsorship for high-performing employees. **Safaricom** provides training budgets for technical staff, and **Liquid Intelligent Technologies** offers certification support as part of its talent development framework. Enterprise IT departments at banks (KCB, Equity) and large corporations may also sponsor relevant certifications. The approach is to present your manager with a clear business case, show how the certification will improve your contribution to the team, reference industry benchmarks, and explain the certification's relevance to upcoming projects. Some employers require you to sign a retention agreement committing to stay for 12–24 months after certification. ### How much do cloud certifications cost in Kenya? Cloud certification costs in Kenya vary by provider and level. **AWS Certified Solutions Architect – Associate** costs approximately KES 18,000–22,000 for the exam alone, with training courses adding KES 40,000–80,000 if taken through a provider. **Microsoft Certified: Azure Administrator Associate (AZ-104)** costs around KES 15,000–20,000 for the exam. **Google Cloud Associate Cloud Engineer** runs approximately KES 15,000–18,000. Professional-level exams (like AWS Solutions Architect Professional) cost roughly 50% more. These costs reflect exam fees only; instructor-led training from local providers like Strathmore or online platforms adds to the total investment. AWS and Microsoft both offer free digital training through their respective platforms, which many Kenyan professionals use for self-study before sitting the paid exam. ## Data Centre Cooling: How Servers Stay Cool in Kenya URL: https://data-centers-254.vercel.app/articles/data-centre-cooling-systems-explained Cluster: Data Centres Published: 2026-08-27 Updated: 2026-08-27 Summary: Cooling can eat up to 40% of a data centre's energy. From CRAC units to liquid cooling, how Kenyan facilities keep servers at optimal temperature. ![Data centre cooling systems](/images/dc-cooling.webp) *Precision air conditioning and cooling systems are the second-largest energy consumer in any data centre* Every server running inside a [what is a data centre](/articles/what-is-a-data-centre) is essentially a controlled heat source. Processors, memory modules, power supplies, and storage drives all convert electrical energy into thermal energy as a byproduct of computation. The fundamental challenge of data centre operations is removing that heat fast enough to keep hardware within safe operating ranges, typically 18–27°C as defined by [ASHRAE Data Centre Cooling Guidelines](https://www.ashrae.org/). Cooling is not optional infrastructure. It is the system that determines whether a facility can reliably operate at all. This article breaks down every major cooling technology used in data centres today, explains how each affects Power Usage Effectiveness, and examines what Kenyan operators are deploying given local climate conditions and grid realities. ## Why Cooling Is the Second-Largest Energy Consumer In a typical data centre, IT equipment accounts for roughly 50–60% of total electricity consumption. Cooling systems consume another 30–40%. The remaining 10–15% covers lighting, security, and ancillary systems. That makes cooling the single largest overhead category and the primary lever for improving operational efficiency. The relationship between cooling and overall efficiency is measured through PUE. A facility with a PUE of 2.0 spends exactly as much on cooling and other overhead as it does on actual computing. A facility with a PUE of 1.2 directs only 20% of its total power to non-IT loads, and the majority of that 20% is cooling. Every 0.1 improvement in PUE translates to meaningful cost savings at scale. ![UPS battery strings and switchgear in a data centre power room](/images/technician-ups-battery-room.webp) *Power and cooling are the two largest energy consumers, together they determine a facility's PUE, and the power half is won in rooms like this* Understanding the [Kenya power and data centres](/articles/kenya-power-infrastructure-data-centres) landscape is critical here. Kenya's electricity tariffs are among the highest in East Africa, so cooling inefficiency directly erodes the commercial viability of any colocation or hyperscale facility operating in Nairobi, Mombasa, or elsewhere in the country. ## How Do Data Centre Cooling Systems Work? All data centre cooling systems perform the same basic function: absorb heat from IT equipment and reject it outside the facility. The differences lie in the medium used to transfer heat (air, water, or refrigerant), the mechanism driving the transfer (fans, pumps, compressors, or natural convection), and the level of precision in temperature and humidity control. The cooling chain typically works in stages. First, heat is captured at the server level, either by air flowing over heat sinks or by cold plates or immersion fluid in direct contact with components. That heat is then transported to a heat rejection system (chiller, cooling tower, or dry cooler) which transfers it to the external environment. Each stage in this chain introduces energy consumption and potential points of failure. ## The Six Major Cooling Types Compared | Cooling Type | How It Works | Typical PUE Impact | Best For | Complexity | Cost | |---|---|---|---|---|---| | **CRAC/CRAH (Precision Air)** | Chilled air blown under raised floor or through ducts into server intakes | 1.5 – 1.8 | Low to medium density (5–15 kW/rack) | Low | Low–Medium | | **Free Cooling (Air-Side)** | Outside air filtered and ducted directly into the white space when ambient conditions allow | 1.2 – 1.5 | Temperate climates, medium density | Low–Medium | Low | | **Free Cooling (Water-Side)** | Cooling towers or dry coolers use cool ambient air to chill water without compressors | 1.2 – 1.4 | Most climates, medium to high density | Medium | Medium | | **Rear-Door Heat Exchangers** | Passive or active heat exchangers mounted on the back of racks, cooling exhaust air | 1.3 – 1.5 | High-density retrofit (15–30 kW/rack) | Medium | Medium | | **Direct Liquid Cooling** | Cold plates or direct-to-chip cooling loops circulate chilled water to CPUs and GPUs | 1.1 – 1.3 | Very high density (30–100+ kW/rack) | High | High | | **Immersion Cooling** | Entire servers submerged in dielectric fluid that absorbs and carries away heat | 1.03 – 1.1 | Extreme density (50–100+ kW/rack), HPC, AI | Very High | Very High | Each technology occupies a specific niche based on rack power density, climate, and budget. Most commercial colocation facilities in sub-Saharan Africa rely on precision air cooling with some form of free cooling supplement. ![Cutaway infographic of an AI data centre's systems](/images/whats-inside-ai-data-center.webp) *Modern high-density servers generate significantly more heat than older equipment, requiring more sophisticated cooling* ## CRAC and CRAH Units: The Industry Standard Computer Room Air Conditioning (CRAC) units and Computer Room Air Handler (CRAH) units are the most widely deployed cooling systems in data centres worldwide, and Kenya is no exception. Both deliver conditioned air to the white space, but they differ in how they produce cold air. ![Illustrated overview of data centre rack components and airflow](/images/diagram-dc-rack-overview.webp) A CRAC unit contains its own refrigeration cycle (compressor, condenser, and evaporator) similar to a split-system air conditioner but with far tighter temperature and humidity control. It can operate independently, which makes it straightforward to deploy. However, the onboard compressor makes it less energy-efficient than alternatives that separate the cooling production from the air delivery. A CRAH unit, by contrast, does not have a compressor. It receives chilled water from a central chiller plant and uses a coil and fan to transfer that cooling to the data centre air. Because the chiller plant can be optimised separately (and because water transfers heat far more efficiently than refrigerant in small distributed units) CRAH-based systems generally achieve better PUE than CRAC-based systems in facilities above 500 kW of IT load. Most [data centre tier ratings](/articles/data-centre-tier-ratings-explained) at Tier II and above in East Africa use CRAH units fed by centralised chiller plants with N+1 or 2N redundancy. This provides the reliability and precision that enterprise and hyperscale customers demand. ## Hot Aisle and Cold Aisle Containment Regardless of whether a facility uses CRAC or CRAH units, the arrangement of airflow within the white space has an enormous impact on cooling effectiveness. Without containment, the cold air supplied to servers mixes with the hot exhaust air before it reaches the intakes. This mixing forces cooling systems to work harder and supply more air than would otherwise be necessary. ![Structured cabling connecting server racks](/images/server-rack-patch-cabling.webp) *Hot aisle and cold aisle containment improves cooling efficiency by preventing hot and cold air from mixing* Hot aisle/cold aisle containment solves this problem by physically separating the two air streams. Server racks are arranged in alternating rows. Cold air is delivered to the front of the racks (the cold aisle), and hot exhaust air is collected from the rear (the hot aisle). Doors, plastic curtains, or rigid enclosures seal the aisles to prevent recirculation. The results are significant. Facilities that implement containment typically see a 10–25% reduction in cooling energy consumption. Supply air temperature can be raised (because it is not diluted by hot air), which allows chillers to operate more efficiently or free cooling to be used for more hours per year. In Nairobi's climate, containment combined with elevated supply temperatures can enable air-side economiser operation for a substantial portion of the year. ## Free Cooling: Leveraging Kenya's Climate Free cooling is not a single technology but a design philosophy: use the external environment as a heat sink whenever possible, and only run mechanical refrigeration when ambient conditions make it unavoidable. There are two main approaches. Air-side economisers draw outside air through filters and directly into the data centre white space when the outside temperature and humidity fall within acceptable ranges. This is the simplest and cheapest form of free cooling, but it requires robust filtration to prevent dust, pollen, and pollutants from entering the white space, a real concern in Nairobi and other urban African environments. Water-side economisers use cooling towers or dry coolers to chill the water loop without engaging the chiller compressors. Because water is a much better heat transfer medium than air, water-side economisers can operate effectively at higher ambient temperatures than air-side systems. In Nairobi, where average temperatures range from 20–25°C and nighttime temperatures regularly drop below 18°C, water-side free cooling can displace mechanical cooling for a significant number of hours annually. The [Schneider Electric Data Centre Efficiency](https://www.se.com/ww/en/work/products-services/offerings/data-centers/) research indicates that facilities in climates similar to Nairobi's can achieve free cooling for 40–60% of annual operating hours using water-side economisers. This is one of the strongest operational advantages Kenyan data centres have over facilities in hotter climates like Lagos or Dubai. ## Liquid Cooling: The Future of High-Density Computing As server power densities increase (driven by GPU-intensive AI workloads, high-performance computing, and converged infrastructure) air cooling is reaching its practical limits. Moving air is inherently inefficient because air has very low heat capacity compared to liquids. Water, for example, transfers heat roughly 1,000 times more effectively than air by volume. Direct liquid cooling circulates chilled water through cold plates mounted directly on processors and other high-heat components. The water absorbs heat at the component level and carries it away to a heat rejection system. This approach can handle rack densities of 30–100 kW or more, far beyond what air cooling can manage without enormous airflow volumes. Immersion cooling takes this further by submerging entire server boards in a dielectric (non-conductive) fluid. The fluid absorbs heat from all components simultaneously and is then circulated through a heat exchanger. Immersion cooling achieves some of the lowest PUE values in the industry (often below 1.1) because nearly all heat is captured directly with minimal thermal resistance. While liquid cooling is still rare in Kenyan data centres today, the rapid growth of cloud computing, AI, and edge computing in East Africa is likely to drive adoption over the next five to ten years. Facilities that plan for liquid cooling infrastructure during initial construction (including floor load capacity, piping routes, and heat rejection systems) will be better positioned to serve high-density customers. ## Rear-Door Heat Exchangers: A Practical Middle Ground For facilities with existing air-cooled infrastructure that need to support higher density racks without a complete cooling overhaul, rear-door heat exchangers (RDHx) offer a compelling solution. These units mount directly on the back of standard server racks and contain a cooling coil connected to a chilled water loop. As hot exhaust air leaves the servers, it passes through the RDHx coil and is cooled before it enters the room. This means the heat is captured and removed at the rack level, before it can contribute to overall room heat load. Active RDHx units include fans to assist airflow, while passive units rely on the server fans alone. Rear-door heat exchangers are particularly attractive for colocation facilities in Kenya because they can be deployed rack-by-rack as customer density increases, without requiring changes to the room-level cooling infrastructure. This incremental deployment model matches the growth patterns of East African data centres, where demand builds gradually rather than in large hyperscale jumps. ## Humidity Control: The Overlooked Cooling Challenge Temperature is not the only environmental parameter that matters. ASHRAE recommends maintaining relative humidity between 20% and 80% in the data centre (with a tighter 40–60% range preferred for optimal equipment longevity). Air that is too dry increases the risk of electrostatic discharge. Air that is too humid risks condensation on cold surfaces and corrosion of circuit boards. In Nairobi, the main humidity challenge is seasonal variation. During the long rains (March–May) and short rains (October–December), humidity can rise significantly, requiring dehumidification. During drier months, humidification may be needed. Both processes consume energy. Modern CRAC and CRAH units integrate humidification and dehumidification, but these functions must be carefully controlled to avoid wasting energy fighting against ambient conditions that could otherwise be tolerated within ASHRAE's wider allowable range. ## What Cooling Systems Are Used in Kenyan Data Centres? The data centre market in Kenya is anchored by facilities in Nairobi, with additional presence in Mombasa. The major operators, including Africa Data Centres, PAIX (now part of Liquid Intelligent Technologies), and several carrier-neutral colocation providers, predominantly use precision air cooling with CRAH units fed by centralised chiller plants. Most Tier III facilities in Nairobi deploy N+1 or 2N chiller configurations with water-side economiser capability. This gives them the redundancy required by their [data centre tier ratings](/articles/data-centre-tier-ratings-explained) while also allowing them to leverage Nairobi's favourable climate for free cooling during cooler periods. Raised floors with perforated tiles remain the standard air delivery method, supplemented by hot aisle containment in more recently built or upgraded halls. Air-side free cooling is less common in Kenya's urban centres due to air quality concerns. Nairobi's dust levels, combined with seasonal pollution from vehicles and industry, make direct air-side economisers risky without very high-efficiency filtration systems, which themselves add pressure drop and fan energy. Water-side economisers avoid this issue entirely because the external air only contacts the cooling tower water, not the data centre air supply. Liquid cooling and immersion cooling are not yet mainstream in Kenyan facilities, but several operators are evaluating these technologies as they prepare for anticipated demand from AI and high-performance computing workloads. The modular data centre containers that some providers deploy to regional markets outside Nairobi are also beginning to incorporate direct liquid cooling options. ## Cooling Efficiency and PUE: The Numbers That Matter PUE remains the industry's primary metric for cooling efficiency, and understanding the relationship between cooling technology and PUE is essential for both operators and customers. A facility using only CRAC units with no containment and no free cooling will typically achieve a PUE of 1.8–2.0. Adding hot aisle containment might bring that down to 1.6–1.7. Upgrading to a CRAH system with water-side economisers and containment can achieve PUE values of 1.4–1.5. The most efficient facilities globally (using liquid cooling, free cooling, and advanced containment) operate at PUE values below 1.2. ![Data centre cooling and power distribution equipment](/images/dc-cooling-crac-2.webp) *Cooling efficiency directly impacts a data centre's carbon footprint and operating costs* For Kenyan operators, the economics are stark. At commercial electricity rates in Kenya, every 0.1 improvement in PUE on a 1 MW facility translates to approximately KES 15–20 million in annual energy savings. Over a 10–15 year facility lifecycle, the cumulative savings from cooling efficiency investments easily justify the upfront capital expenditure. ## The Path Forward for Cooling in East Africa Cooling technology in Kenyan data centres is evolving along three parallel tracks. First, existing air-cooled facilities are being optimised through better containment, elevated supply temperatures, and maximised free cooling hours. Second, new constructions are incorporating infrastructure that supports future liquid cooling deployment. Third, operators are investing in real-time monitoring and AI-driven cooling management systems that dynamically adjust cooling capacity to match actual IT load, eliminating the waste of overcooling. Nairobi's altitude (approximately 1,795 metres above sea level) and temperate climate are genuine competitive advantages for data centre operations. The thinner, cooler air at altitude improves both natural heat dissipation and cooling tower efficiency. Combined with Kenya's growing renewable energy portfolio (particularly geothermal and wind power) the country has the foundational conditions to develop some of the most operationally efficient data centres on the continent. The operators that will lead this market are those that treat cooling not as a cost to be minimised but as a system to be engineered for maximum performance per watt. In a region where electricity is expensive and demand for digital infrastructure is growing rapidly, cooling efficiency is not just an operational concern, it is a strategic imperative. --- **External sources:** - [ASHRAE Data Centre Cooling Guidelines](https://www.ashrae.org/) - [Schneider Electric Data Centre Efficiency](https://www.se.com/ww/en/work/products-services/offerings/data-centers/) ## Data Centre Design and Build in Kenya: Site to Handover URL: https://data-centers-254.vercel.app/articles/data-centre-design-build-kenya Cluster: Infrastructure Published: 2026-08-28 Updated: 2026-09-28 Summary: The full data centre design and build process in Kenya: site selection, feasibility, MEP design, commissioning, Kenya Power connection, and NEMA EIA. ## Introduction: Why Data Centre Design and Build Matters in Kenya Kenya's digital economy is expanding at a pace that demands robust, locally-built data centre infrastructure. With the government's Digital Economy Blueprint, the growing presence of hyperscale cloud providers, and increasing enterprise demand for [colocation services in Kenya](/articles/colocation-data-centre-kenya), the country is witnessing a data centre construction boom. But building a data centre is far more complex than constructing a standard commercial building, it requires specialised knowledge of power systems, cooling, structural engineering, and regulatory compliance specific to the Kenyan context. This guide walks through every stage of the data centre design and build process in Kenya, from initial site selection through to commissioning and handover. ![Server racks in a modern data centre hall](/images/whats-inside-ai-data-center-wide.webp) *Modern server rack installation during the fit-out phase of a Kenyan data centre build* ## Phase 1: Site Selection Criteria in Kenya Choosing the right location is the single most consequential decision in any data centre project. In Kenya, site selection must balance multiple technical, regulatory, and commercial factors. ### Power Availability and Grid Proximity **Reliable power access is non-negotiable.** Kenya Power (KPLC) operates the national grid, and data centre operators must evaluate proximity to high-voltage substations. Facilities in Nairobi's Industrial Area and along Mombasa Road benefit from relative grid stability, while locations further from the central transmission network may require costly dedicated lines. The [Kenya power infrastructure for data centres](/articles/kenya-power-infrastructure-data-centres) varies significantly across regions, with Nairobi enjoying the most robust supply. Dual-grid feeds from independent KPLC substations are essential for any facility targeting **Tier III or higher** classification. Operators must also evaluate the feasibility of on-site backup power, including diesel generator fuel storage and increasingly, solar and battery energy storage systems. ### Fibre Connectivity and Network Proximity A data centre without diverse fibre connectivity has limited value. Kenyan site selection must confirm proximity to multiple fibre duct routes from operators such as Liquid Intelligent Technologies, Safaricom, Telkom Kenya, Jamii Telecommunications, and the Kenya Data Centres backbone. Proximity to submarine cable landing stations in Mombasa (particularly the EASSy, Seacom, and TiGS cable systems) is critical for facilities serving international traffic. ### Flood Zones, Seismic Considerations, and Terrain **Kenya is not typically associated with major seismic activity**, but the Kenya Geological Survey has identified moderate seismic risk zones, particularly along the Rift Valley. The USGS seismic hazard maps for East Africa indicate that Nairobi falls within a low-to-moderate risk zone, but structural design must still comply with seismic loading requirements under the Kenyan building code. Flood risk is more immediate. Nairobi has experienced severe flooding events, notably in April 2024 when widespread floods displaced thousands and damaged infrastructure. Data centre sites must be evaluated against flood plain maps from the Ministry of Water and Irrigation, and the building's plinth level should be raised above historical flood lines. Mombasa's coastal locations face additional tidal surge and sea-level rise considerations. ### Land Ownership, Zoning, and Title Verification Land due diligence is critical in Kenya. The [data centre due diligence process](/articles/data-centre-due-diligence-kenya) should include verification of title deeds at the Ministry of Lands, confirmation that the land use zoning permits commercial or industrial development, and a search for any encumbrances or caveats. In Nairobi County, data centre developments typically fall under commercial or special-use zoning categories, and the county's physical planning department must approve the proposed use. ## Phase 2: Feasibility Studies Before committing significant capital, developers must conduct thorough feasibility studies covering technical, financial, regulatory, and market dimensions. ### Technical Feasibility This involves detailed assessments of power capacity (available kVA from KPLC and upgrade requirements), cooling feasibility (ambient temperature profiles, water availability for chillers), and structural load-bearing capacity of the site. For facilities in Nairobi, where average temperatures range from 13°C to 28°C, **free cooling opportunities** exist for much of the year, significantly improving Power Usage Effectiveness (PUE). ### Market and Financial Feasibility Developers must evaluate demand for colocation, hyperscale, or hybrid cloud services in the target market. The [Kenya data centre market outlook for 2025–2030](/articles/kenya-data-centre-market-outlook-2025-2030) projects significant growth driven by cloud adoption, 5G rollout, and AI workloads. Financial modelling should account for KPLC power tariffs (currently ranging from KES 12–25 per kWh for large commercial users), construction costs in Kenya (typically USD 800–1,500 per kW of IT load for Tier III), and expected revenue per rack or per kW. ## Phase 3: Design Phases ### Conceptual Design ![Diagram of a medium enterprise office rack design with security and switching layers](/images/diagram-office-rack-design.webp) The conceptual design phase establishes the facility's overall scope, target tier classification, capacity (in kW or MW), and architectural approach. Key decisions at this stage include whether to build a purpose-built facility or retrofit an existing commercial building, operators like **Africa Data Centres** and **PAIX** (Pan-African Internet Exchange) have taken both approaches in Kenya. The conceptual design produces initial floor plans, block layouts showing white space, electrical rooms, mechanical plant rooms, and support areas. It also defines the **resilience strategy**, N+1, 2N, or N+2 redundancy for critical systems. ### Detailed Design The detailed design phase translates the concept into construction-ready documentation. This includes: - **Architectural drawings** compliant with Nairobi County building codes and [data centre tier rating requirements](/articles/data-centre-tier-ratings-explained) - **Structural engineering** calculations for raised floors, heavy equipment (UPS systems, generators, chillers), and seismic loads - **Electrical engineering** designs covering utility feeds, switchgear, UPS systems, power distribution units (PDUs), generator systems, and earthing/grounding - **Mechanical engineering** designs for precision cooling, chilled water systems, air handling units, and ductwork ![Officials marking the launch of the NBO2 data centre in Nairobi](/images/nbo2-launch-ribbon-cutting.webp) *Power distribution infrastructure is a critical design element in any Kenyan data centre build, NBO2's September 2026 launch showed the end result of that design work* ### MEP Design: The Most Critical Discipline **Mechanical, Electrical, and Plumbing (MEP) design is the heart of data centre construction.** In the Kenyan context, MEP design faces unique challenges: - **Power quality**: Despite improvements, Kenya's grid can experience voltage fluctuations. The MEP design must incorporate robust power conditioning, automatic voltage regulators, and UPS systems with sufficient battery runtime (typically 10–15 minutes) to bridge until generator startup. - **Cooling efficiency**: Nairobi's mild highland climate allows for economiser-based cooling for 6–8 months annually, but design must also handle hot season peaks. Water-cooled systems face water scarcity challenges in some Nairobi locations. - **Fire suppression**: Kenya's fire safety regulations, enforced through the Kenya Fire Brigade Act and Nairobi County by-laws, typically require gas-based fire suppression (Novec 1230 or FM-200) in the white space, complemented by wet sprinkler systems in non-critical areas. ## Phase 4: Construction Considerations Specific to Kenya ![Construction cranes rising over the Nairobi skyline](/images/dc-business-investment.webp) ### Materials Availability and Import Logistics Many specialised data centre materials are not manufactured in Kenya and must be imported. This includes raised-floor panels, hot-aisle/cold-aisle containment systems, precision air-conditioning units, UPS systems, and automatic transfer switches (ATS). Lead times from manufacturers in Europe, China, and South Africa can range from 8 to 16 weeks, and **import duties and VAT (16%)** must be factored into project budgets. Standard construction materials (concrete, steel reinforcement, structural steel, and electrical cable) are readily available from Kenyan suppliers, though quality must be verified against **KEBS standards**. The [Kenya Bureau of Standards](https://www.kebs.org) certifies materials including cement (KS 02-542), steel reinforcement (KS 574), and electrical cables (KS IEC 60227). ### The Contractor Ecosystem Kenya's construction industry includes both large established firms and specialised subcontractors. For data centre projects, the contracting strategy typically involves a main contractor for civil and structural works, with specialist subcontractors for MEP installation. Finding contractors with specific data centre experience remains a challenge, while many firms have built industrial and commercial facilities, the precision requirements of data centre construction (tolerances for raised floors, cable routing, piping alignment) demand additional supervision and quality assurance. ### Regulatory Compliance During Construction **Nairobi County building approvals** require submission of architectural plans, structural calculations, and MEP drawings to the County's Department of Urban Planning. The approval process can take 2–6 months. During construction, county inspectors verify compliance at foundation, superstructure, and finishing stages. The [National Environment Management Authority (NEMA)](https://www.nema.go.ke) requires that the approved EIA conditions are adhered to throughout construction. This includes dust control, noise management, stormwater management, and waste disposal protocols. ## Phase 5: Commissioning Process Commissioning is the systematic process of verifying that all installed systems perform as designed. In data centre construction, commissioning is arguably the most critical quality assurance step. ### Integrated Systems Testing (IST) The commissioning process culminates in **Integrated Systems Testing (IST)**, which simulates real-world failure scenarios. This includes: - **Utility failure simulation**: Confirming that UPS systems pick up load within milliseconds and that generators start and assume full load within the specified timeframe (typically 30–60 seconds) - **Cooling failure simulation**: Verifying that the facility can maintain acceptable temperatures during a cooling system outage using thermal mass and emergency ventilation - **Fire suppression testing**: Confirming detection, alarm, and suppression system activation sequences ### Commissioning in the Kenyan Context Commissioning in Kenya requires coordination with KPLC for utility switch-over testing. KPLC engineers must participate in the synchronisation and load-transfer tests for dual-grid feeds. The **commissioning agent** (typically an independent third party) must verify that all systems meet the design intent, the specified tier requirements, and applicable Kenyan standards. ![Rooftop exhaust fans and HVAC ducting](/images/dc-chillers-roof-3.webp) *Cooling system installation during the MEP phase of data centre construction* ## Phase 6: Handover and Operations Transition The handover phase transfers the completed facility from the construction team to the operations team. This involves: - **As-built documentation**: Complete sets of drawings, manuals, and test certificates - **Training**: Operations staff training on all installed systems - **Warranty management**: Establishing warranty tracking for all equipment and construction elements - **Ongoing maintenance planning**: Development of preventive maintenance schedules aligned with manufacturer recommendations and tier requirements Operators like **Liquid Intelligent Technologies**, which operates multiple facilities across East Africa, have developed mature operations transition processes. **Africa Data Centres**, a Cassava Technologies subsidiary, has invested heavily in building local operations teams to manage their growing Kenyan portfolio. ## Key Kenyan Regulatory Bodies in the Build Process | Regulatory Body | Role in Data Centre Build | |---|---| | Nairobi County | Building plan approvals, occupation certificates, zoning compliance | | NEMA | Environmental Impact Assessment licence, construction environmental monitoring | | Kenya Power (KPLC) | Grid connection agreements, high-voltage supply design, metering | | KEBS | Materials standards certification, electrical equipment safety | | Kenya Fire Brigade | Fire safety plan approval, fire suppression system inspection | | Communications Authority of Kenya (CA) | Licensing for carrier-grade facilities | | National Construction Authority (NCA) | Contractor registration, construction quality oversight | ## Conclusion: Building Kenya's Digital Future The data centre design and build process in Kenya is complex but increasingly well-understood as the local industry matures. From site selection through commissioning, each phase requires careful attention to both international best practices and Kenya-specific regulatory, environmental, and infrastructure realities. As demand for digital infrastructure continues to grow (driven by cloud adoption, AI workloads, and Kenya's position as East Africa's technology hub) the construction pipeline is expected to accelerate. For developers and investors, understanding this build process is essential for delivering facilities that meet the performance, reliability, and compliance standards that modern digital services demand. ## Frequently Asked Questions ### How long does it take to build a data centre in Kenya from site selection to commissioning? The typical timeline ranges from 18 to 36 months depending on the facility's scale and tier classification. A Tier II or Tier III facility of 1–5 MW in Nairobi generally takes 18–24 months, while larger hyperscale deployments by operators like Africa Data Centres or Liquid Intelligent Technologies can extend to 30–36 months. The site selection and feasibility phase alone can consume 3–6 months, especially when securing Kenya Power high-voltage grid connections and NEMA environmental impact assessment approvals. Delays in Nairobi County building plan approvals and imported equipment lead times for specialised MEP components can add further months to the schedule. ### What KEBS standards apply to data centre construction in Kenya? Several KEBS standards are relevant to data centre builds. KS ISO/IEC 27001 governs information security management in facility design. KEBS also references international standards such as KS EN 50173 for structured cabling, KS IEC 60364 for electrical installations in buildings, and fire safety standards under KS CP 15 and the Kenya Fire Safety Act. For structural integrity, KS 641 (Code of Practice for Use of Concrete) and KS 02-300 (structural steel) apply. Data centre operators targeting Tier III certification also design against the Uptime Institute's Tier Standard Topology, which is adopted by reference in many Kenyan build specifications, though KEBS does not yet have a dedicated data centre construction standard. ### What is the Kenya Power grid connection process for a data centre? Data centres typically require dedicated high-voltage connections (33 kV or 132 kV) from Kenya Power. The process involves submitting a formal application to KPLC with load projections, single-line diagrams, and site plans. KPLC then conducts a system study to confirm grid capacity at the proposed location. For facilities drawing over 1 MW, KPLC may require the developer to fund nearby substation upgrades or dedicated transmission lines. The connection agreement, metering setup, and commissioning of the supply typically take 6–12 months. Dual-feed redundancy is essential for Tier III facilities, requiring two independent power feeds from separate KPLC substations. The cost of grid connection can range from KES 50 million to over KES 500 million depending on proximity to existing high-voltage infrastructure. ### Does a data centre project in Kenya require an Environmental Impact Assessment? Yes. Under Kenya's Environmental Management and Coordination Act (EMCA) 1999, data centre projects fall under the category of large commercial and industrial developments that require an Environmental Impact Assessment (EIA) by NEMA-accredited lead experts. The EIA process includes a project brief submission, scoping report, full EIA study, and public participation with affected communities. NEMA evaluates impacts on air quality (from backup diesel generators), noise pollution, water consumption, wastewater discharge, and electromagnetic fields. For facilities within Nairobi or its metropolitan area, additional county-level environmental approvals may apply. The EIA licence is typically a prerequisite for obtaining Nairobi County building plan approvals, making it a critical-path item in the project schedule. ### Who are the main data centre design and build contractors operating in Kenya? The Kenyan data centre construction ecosystem includes a mix of international specialists and local firms. Global data centre design firms such as Eaton, Schneider Electric, and Vertiv provide MEP design and equipment. For construction, firms like CPB Contractors (involved with Africa Data Centres), China Wu Yi, and local contractors including Gathoni & Partners, Spencon, and Interconsult have participated in data centre builds. Specialist MEP contractors handle the critical power and cooling installations. The ecosystem also includes commissioning agents, [data centre tier certification consultants](/articles/data-centre-tier-ratings-explained), and project management firms. As demand grows, more Kenyan engineering firms are building in-house data centre capabilities, though highly specialised roles like raised-floor installation, hot-aisle containment fabrication, and precision air-conditioning commissioning still rely partly on imported expertise. ## Data Centre Due Diligence in Kenya: A Checklist URL: https://data-centers-254.vercel.app/articles/data-centre-due-diligence-kenya Cluster: Business Guide Published: 2026-08-28 Updated: 2026-08-28 Summary: A due diligence checklist for choosing a Kenyan data centre: power, cooling, connectivity, security, and compliance, before you sign. Choosing a data centre is one of the most consequential infrastructure decisions a Kenyan organisation will make. Your servers, your data, and your customers' trust will reside in that facility for years. A bad choice (unreliable power, inadequate security, poor connectivity, or an operator in financial difficulty) can result in outages, data breaches, regulatory penalties, and reputational damage that far exceed the cost of the colocation contract itself. Due diligence is the process that prevents these outcomes. ![Server racks in a Nairobi data centre hall](/images/dc-gpu-cluster-2.webp) This guide provides a practical due diligence framework specifically designed for the Kenyan market. It covers every factor that an enterprise customer (a bank, a telecom company, a government agency, or a large corporate) should evaluate before placing equipment in a Kenyan data centre. While some of these factors are universal, others are specific to Kenya's infrastructure environment, regulatory context, and market dynamics. ## Phase 1: Pre-qualification Before investing time in detailed due diligence, narrow the field to facilities that meet your basic requirements. This pre-qualification phase eliminates facilities that cannot serve your needs regardless of how well they perform in other areas. ### Location and Accessibility Determine whether the facility's physical location works for your operations. If your IT team needs regular physical access to equipment, the facility should be within a reasonable distance of your office. For organisations in Nairobi's CBD, the Mombasa Road corridor is a 20–40 minute drive. For organisations in Westlands, a facility on Mombasa Road is less convenient than one in Westlands or Parklands. Consider accessibility during different times of day and in different conditions. Nairobi traffic is notoriously heavy during rush hours, and a facility that takes 20 minutes to reach at midday may take 90 minutes at 5 PM. If you need 24/7 access for emergency work, can your team reach the facility at any hour? Is there adequate parking? Is the surrounding area safe for staff travelling at night? ### Capacity Availability Confirm that the facility has the capacity you need, both now and for the foreseeable future. A facility that is 90% full may not be able to accommodate your growth. Ask about total capacity (racks, power, cooling), current utilisation, and expansion plans. A facility with expansion land or the ability to add power capacity is more future-proof than one that is already at its limits. ### Carrier Neutrality If you need to connect to multiple network providers, confirm that the facility is carrier-neutral. Ask which providers have equipment in the meet-me room. A facility that only connects to one or two providers limits your connectivity options and negotiating leverage. ## Phase 2: Technical Evaluation The technical evaluation is the most detailed phase of due diligence. It assesses the physical infrastructure that determines whether the facility can reliably house your equipment. ### Power Infrastructure Power is the single most important technical factor. Evaluate the following elements in detail. **Utility supply**: What is the facility's grid connection? Is it served by a dedicated Kenya Power feeder (preferred) or a shared feeder? What voltage level (11kV, 33kV)? What is the contracted demand? Has the facility experienced any grid outages in the past 12 months, and if so, how long and how frequently? **Generators**: How many diesel generators are installed? What is their total capacity relative to the facility's IT load? The industry standard is N+1 (one spare generator) at minimum. How much fuel is stored on-site, and how long can the generators run at full load? The minimum standard for enterprise use is 24 hours of fuel storage; 48–72 hours is preferred. Are there contracts for emergency fuel delivery? ![Security systems assessment during a facility audit](/images/dc-biometric-access-4.webp) **UPS systems**: What UPS technology is used (online double-conversion is the standard for enterprise facilities)? What is the battery runtime? The minimum is 10–15 minutes, enough time for generators to start and synchronise. Are the UPS systems redundant (N+1 or 2N)? What is the age and condition of the batteries? **Power distribution**: How is power distributed from the utility entry point to the racks? Look for redundant power paths (A and B feeds) to each rack. Ask about the power distribution unit (PDU) topology and whether each rack has dual PDUs on separate power paths. ### Cooling Infrastructure Cooling is the second most critical technical factor. **Cooling capacity**: What is the total cooling capacity, and how does it compare to the current and planned IT load? Cooling should have the same or greater redundancy as power (N+1 minimum). **Cooling technology**: What type of cooling system is used? Precision air conditioning (CRAC/CRAH) is standard. Does the facility use free cooling (outside air) to reduce energy consumption? Nairobi's climate allows free cooling for significant portions of the year, and facilities that take advantage of this have lower operating costs and PUE values. **Environmental monitoring**: Are temperature and humidity monitored continuously throughout the white space? Ask to see the building management system (BMS) and verify that sensors are placed at multiple points (not just at the return air of cooling units, but at rack inlets and hot aisles). ### Connectivity Connectivity evaluation focuses on diversity and redundancy. **Fibre entry diversity**: How many independent fibre routes enter the building? A single route creates a single point of failure, a road construction project could sever the only fibre connection and isolate the facility. Look for at least two physically diverse entry routes. **Meet-me room**: Visit the meet-me room. Which network providers have equipment present? Is the room well-organised, with proper cable management and labelling? Is there space for additional providers? **Cross-connect process**: How quickly can cross-connects be provisioned? What are the costs? Are there restrictions on which providers you can connect to? ![UPS and switchgear line-up in a data centre power room](/images/ups-switchgear-power-room.webp) ## Phase 3: Security Evaluation Security evaluation covers both physical and cybersecurity. [Our security guide](/articles/data-centre-security-explained) covers the details; here we focus on what to verify during due diligence. ### Physical Security Walk the security perimeter. Are all access points controlled? Are CCTV cameras covering all approaches, not just the main entrance? Ask about camera retention period (90 days minimum) and whether footage is monitored in real-time or recorded for later review. Experience the access control process. Request a visitor badge and walk through the process. Does it include biometric verification? Are there mantraps? Is access logged? A facility that makes it easy for you to walk in unchallenged is not adequately secured. Ask about security staffing levels. How many security personnel are on-site at different times of day? Are they employed directly by the data centre operator or by a security contractor? What training do they receive? ### Compliance and Certifications Request copies of all certifications. ISO 27001 (information security management) is the most relevant for most customers. SOC 2 Type II reports, while more common in North American markets, are increasingly expected by international organisations. If the facility claims a Tier rating, verify whether it is self-declared or certified by the Uptime Institute. Self-declared Tier ratings carry less weight than independent certification. ## Phase 4: Business and Contractual Evaluation The business evaluation assesses the operator's financial stability, the contract terms, and the SLA. ### Financial Stability A data centre operator in financial difficulty may cut corners on maintenance, staffing, or investment, all of which affect the reliability of your service. Request financial information or, for privately held companies, assess indicators of financial health: customer base (diverse or dependent on one or two large customers), investment in facility maintenance and upgrades, staff retention, and payment history with suppliers (if you can assess this indirectly). ### Service Level Agreement The SLA is the contractual backbone of the colocation relationship. Evaluate it carefully. **Uptime guarantee**: What uptime percentage is guaranteed? For enterprise use, 99.99% (approximately 52 minutes of downtime per year) is the standard. 99.9% (8.7 hours per year) may be acceptable for non-critical workloads. Anything below 99.9% is insufficient for enterprise use. **Service credits**: What financial compensation is provided if the SLA is not met? Are service credits applied automatically or only upon customer request? Are there caps on total service credits that limit your effective compensation? **Exclusions**: What is excluded from the SLA? Scheduled maintenance, force majeure events, and customer-caused outages are commonly excluded. Read the exclusions carefully, some SLAs exclude so much that the guarantee is effectively meaningless. ![Fingerprint being recorded on an ink pad](/images/dc-biometric-access-3.webp) ### Contract Terms Review contract duration, renewal terms, and termination conditions. Most enterprise colocation contracts are 3–5 years. Ensure that termination for cause is clearly defined and that you can exit the contract if the operator fails to meet material obligations. Consider whether the contract includes annual price escalation and, if so, whether the escalation is reasonable and predictable. ## Phase 5: The Site Visit The site visit is where due diligence becomes real. Do not accept a guided tour that shows only the lobby, a meeting room, and a glass-walled viewing corridor into the white space. A proper site visit should include access to the white space, the electrical rooms, the cooling plant, and the meet-me room. During the visit, observe the condition of the infrastructure. Are cable trays organised and labelled, or are cables hanging loosely? Is the white space clean and free of debris? Are floor tiles properly seated? Do the generators and UPS systems appear well-maintained? Is the cooling plant operating quietly and smoothly, or are there unusual noises or vibrations? Talk to the operations staff, not just the sales team. The operations team's knowledge and professionalism are better indicators of day-to-day facility quality than any sales presentation. Ask them about recent incidents, how they handle maintenance, and what their biggest operational challenges are. Due diligence is not a formality, it is the process that separates a good data centre decision from a costly mistake. In Kenya's growing but still developing market, the variation between facilities is significant, and the effort invested in thorough due diligence will pay for itself many times over the life of the contract. ## Data Centre Interconnection and Peering in Kenya: KIXP URL: https://data-centers-254.vercel.app/articles/data-centre-interconnection-peering-kenya Cluster: Infrastructure Published: 2026-08-28 Updated: 2026-10-06 Summary: How interconnection and peering work in Kenya: KIXP, peering vs transit economics, cross-connects, and why Nairobi is East Africa's hub. ## Introduction: Why Interconnection Is the Lifeblood of Data Centres A data centre is only as valuable as the networks it connects. In Kenya, where [fibre optic networks are expanding rapidly](/articles/fibre-optic-networks-kenya-data-centres) and submarine cables land at the Mombasa coast, interconnection (the ability to link networks, cloud platforms, and enterprises directly within and between facilities) has become the most important competitive differentiator for data centre operators. This article explains how data centre interconnection and peering work in Kenya, covering the Kenya Internet Exchange Point (KIXP), cross-connects within facilities, the economics of peering versus transit, and why Kenya is consolidating its position as East Africa's premier interconnection hub. ![Shore-based cable works on the Kenyan coast with a vessel offshore](/images/mombasa-cable-landing.webp) *Network switching infrastructure enabling interconnection and peering within a Kenyan data centre* ## What Is an Internet Exchange Point (IXP)? An **Internet Exchange Point (IXP)** is a physical infrastructure through which internet service providers (ISPs), content delivery networks (CDNs), cloud providers, and other networks exchange internet traffic directly with each other. Instead of traffic between two Kenyan networks (say, Safaricom and Telkom Kenya) travelling out of the country via submarine cables to Europe and back, an IXP allows that traffic to be exchanged locally within the same city. The benefits are substantial: - **Reduced latency**: Local traffic stays local, cutting round-trip times from 150–300 ms (via Europe) to under 10 ms - **Lower costs**: Networks avoid paying for international transit capacity on submarine cables - **Improved reliability**: Less dependency on undersea cable routes, which can be disrupted by cable cuts - **Better user experience**: Faster loading times for locally-hosted content and services ## KIXP: Kenya's Internet Exchange Point ### History and Growth The **Kenya Internet Exchange Point (KIXP)** was established in 2002 by the Telecommunications Service Providers Association of Kenya (TESPOK), making it one of the first IXPs in East Africa. At launch, it handled just a few hundred megabits per second. The growth curve since is steep: the Internet Society measured 19 Gbps of peak traffic by 2020, TESPOK announced a historic peak of 1.3 Tbps on 25 June 2024, and by October 2026 PeeringDB listed **144 member networks with roughly 2,985 Gbps of connected port capacity** - a growth of several orders of magnitude that reflects Kenya's dramatic internet expansion. KIXP is a neutral, not-for-profit exchange operated by [Packet Clearing House](https://www.pch.net) in partnership with TESPOK. It operates from multiple points of presence (PoPs) across Nairobi's major data centre facilities, ensuring that members can connect from whichever facility they operate in. ### The 100G Upgrade In 2024–2025, KIXP completed a significant infrastructure upgrade to **100 Gigabit Ethernet (100GbE) port capabilities** at its core switches. This upgrade was essential to accommodate the growing bandwidth demands of mobile network operators, streaming services, and cloud providers. The 100G capability means that large members like Safaricom and Liquid Intelligent Technologies can exchange traffic at much higher volumes without port congestion, and it positions KIXP to handle projected traffic growth as 5G adoption and video streaming increase in Kenya. ### KIXP Membership Profile KIXP's membership has evolved significantly. The current roster includes: - **Mobile network operators**: Safaricom, Airtel Kenya, Telkom Kenya - **ISPs and fibre providers**: Liquid Intelligent Technologies, Jamii Telecommunications, FAIBA (Wananchi Group), Poea Internet, AccessKenya - **Content providers and CDNs**: Google (via Global Cache), Netflix, Meta, Cloudflare - **Cloud and technology platforms**: AWS, Microsoft Azure, and others through local PoPs - **Government and education networks**: Kenya Education Network (KENET), government data centres - **Enterprise and financial services**: Banks and other enterprises with direct IXP connections for low-latency financial applications This diverse membership is what makes KIXP powerful, the more networks that participate, the more traffic can be kept local. ## Peering vs Transit: Understanding the Economics ### What Is Peering? **Peering** is a mutually beneficial arrangement where two networks agree to exchange traffic directly, typically without charging each other. At KIXP, peering allows a Safaricom customer accessing a Google service hosted locally to have that traffic exchanged directly at the IXP, without either party paying for international transit. Peering can be: - **Public peering**: Exchanging traffic over a shared IXP fabric (like KIXP) - **Private peering**: A direct connection between two networks, often via a cross-connect in a shared data centre ### What Is Transit? **Transit** is a commercial service where one network (the transit provider) agrees to carry another network's traffic to all destinations on the internet. Transit providers charge for this service, typically based on bandwidth capacity (per Mbps or per Gbps). For Kenyan networks, transit connects them to the global internet via submarine cables (EASSy, Seacom, and TiGS) landing in Mombasa. ### The Cost Case for Peering in Kenya The economics are compelling. **International transit from Kenya typically costs USD 50–200 per Mbps per month**, depending on the provider, capacity commitment, and route. By peering at KIXP, networks can offload a significant portion of their traffic, industry estimates suggest KIXP keeps **over 70% of locally-sourced traffic within Kenya**. For a mid-sized ISP with 10 Gbps of total traffic, if 7 Gbps can be peered locally at KIXP, the savings on transit costs alone can amount to **USD 350,000–1.4 million per month**. These savings flow through to lower retail broadband prices for consumers and more competitive enterprise connectivity offerings. ## Data Centre Cross-Connects: Direct Connections Within Facilities ### What Is a Cross-Connect? ![Technician splicing fibre at a patch panel on a data centre floor](/images/fibre-splicing-technician-data-hall.webp) A **cross-connect** is a physical cable (typically fibre optic or copper) that directly links one party's equipment to another's within the same data centre. Cross-connects are the fundamental building block of data centre interconnection. In practice, a cross-connect might connect: - A bank's server rack directly to Safaricom's network equipment for dedicated bandwidth - An enterprise to AWS Direct Connect for private cloud access - A content provider to multiple ISPs for optimal content delivery - Two enterprise partners for secure, low-latency data exchange ### Why Cross-Connects Drive Data Centre Value Data centres with rich interconnection ecosystems command premium pricing for [colocation services in Kenya](/articles/colocation-data-centre-kenya). When a customer can connect to ten different networks, three cloud platforms, and multiple enterprise partners through cross-connects in a single facility, the value proposition is far stronger than a facility with limited connectivity options. **Africa Data Centres**, **Liquid Intelligent Technologies**, and **PAIX** all market their interconnection capabilities as a key differentiator. The ability to offer direct cloud on-ramps (AWS Direct Connect, Azure ExpressRoute, Google Cloud Interconnect) is particularly important for attracting enterprise customers migrating workloads to the cloud. ### Cross-Connect Provisioning in Kenya Kenyan data centres typically offer cross-connect provisioning within **24 to 72 hours**, depending on the complexity and whether the interconnection requires configuration on the network side. Cross-connects are priced as a monthly recurring charge plus a one-time setup fee, with pricing varying by facility and cable type (single-mode fibre, multi-mode fibre, or copper). ## Kenya as East Africa's Interconnection Hub ### The Geographic Advantage Kenya's position as an interconnection hub is no accident. Several factors converge: - **Submarine cable landing**: Mombasa is the landing point for EASSy, Seacom, TiGS, and DARE1, giving Kenya direct connectivity to Europe, Asia, and the rest of Africa. [The choice between Nairobi and Mombasa for data centre location](/articles/nairobi-vs-mombasa-data-centre-locations) often comes down to balancing proximity to cable landing stations (Mombasa) against market proximity and ecosystem depth (Nairobi). - **Regional connectivity**: Kenya borders Uganda, Tanzania, Ethiopia, and South Sudan. Cross-border fibre links from these countries terminate in Nairobi, making it a natural aggregation point. - **Market size and sophistication**: Kenya has East Africa's largest and most developed digital economy, with the highest internet penetration rate in the region. ![Aerial view of the Port of Mombasa](/images/mombasa-port.webp) *Kenya's position at the crossroads of multiple submarine cables makes it a natural interconnection hub* ### KIXP's Regional Role KIXP's influence extends beyond Kenya's borders. Networks from Uganda, Tanzania, Rwanda, and Burundi peer at KIXP either through direct connections or via their Kenyan upstream providers. This regional role is supported by the **East African Backhaul System**, which provides high-capacity fibre links between Nairobi and neighbouring capitals. ## Major Peering Participants in Kenya's Ecosystem ### Telecommunications Operators **Safaricom** is by far the largest traffic contributor at KIXP, reflecting its dominance in Kenya's mobile and fixed-line markets. **Telkom Kenya** and **Airtel Kenya** are also significant participants, using KIXP to exchange traffic with each other and with content providers. **Liquid Intelligent Technologies** plays a dual role, as both a significant traffic source and a carrier providing connectivity between Kenyan data centres and the broader Liquid network spanning over 20 African countries. ### Content Delivery Networks and Cloud Providers The presence of **Google, Netflix, Meta, and Cloudflare** at KIXP has been transformative. Google's Global Cache nodes, deployed at KIXP and within Kenyan data centres, mean that YouTube, Google Search, and other Google services are served locally. Netflix similarly caches popular content at the exchange, dramatically improving streaming quality for Kenyan users. **AWS and Microsoft Azure** have established local interconnection points, enabling Kenyan enterprises to access cloud services with single-digit millisecond latency rather than the 100+ ms latency typical of accessing European cloud regions. ## Comparing KIXP with Africa's Other Major IXPs ### KIXP vs NAPAfrica (South Africa) **NAPAfrica**, operated by INX-ZA, is Africa's largest IXP by a significant margin. With over 300 members and peak traffic exceeding 2 Tbps across its Johannesburg and Cape Town nodes, NAPAfrica benefits from South Africa's larger economy, more mature telecom sector, and the presence of multiple submarine cable systems. KIXP, while smaller in absolute terms, has grown faster proportionally. The 100G upgrade has narrowed the infrastructure gap, and KIXP's role as the primary IXP for the broader East African market gives it a strategic importance beyond what raw traffic numbers suggest. ### KIXP vs IXPN (Nigeria) **IXPN** (Internet Exchange Point of Nigeria) handles approximately 100–120 Gbps peak traffic with 60+ members. Nigeria's larger population gives IXPN potential for massive growth, but infrastructure challenges and a more fragmented telecom market have constrained its development relative to KIXP. ![Nairobi skyline at sunset](/images/nairobi-sunset-wide.webp) *Africa's growing interconnection ecosystem with KIXP positioned as a key East African node* ### Kenya's Competitive Positioning Kenya's advantage lies in its **combination of submarine cable diversity, regulatory environment, and ecosystem maturity**. The Communications Authority of Kenya (CA) has been supportive of IXP development, and the competitive telecom market has driven operators to peer rather than rely solely on transit. The [Kenya data centre market outlook for 2025–2030](/articles/kenya-data-centre-market-outlook-2025-2030) projects that interconnection will be a primary driver of data centre demand, as enterprises and cloud providers seek facilities with the richest network ecosystems. ## The Future of Interconnection in Kenya Several trends will shape Kenya's interconnection landscape in the coming years: 1. **5G and edge computing**: As 5G networks roll out, the demand for local interconnection at the network edge will increase, driving KIXP expansion into additional data centre facilities and potentially edge PoPs. 2. **Cloud-on-ramp proliferation**: All major Kenyan data centres are racing to offer direct cloud connections. The facility with the most cloud on-ramps will attract the most enterprise customers. 3. **AI and GPU workloads**: As [AI computing demands grow in Kenya](/articles/ai-data-centres-east-africa), interconnection between GPU-equipped data centres and cloud AI services will become critical. 4. **Regional integration**: As cross-border fibre capacity increases, KIXP's role as the East African peering hub will strengthen, potentially leading to formal peering agreements with IXPs in Uganda (UIXP), Tanzania (TIX), and Rwanda (RINEX). 5. **Submarine cable expansion**: New submarine cable projects planned for the East African coast will further increase international connectivity capacity, making Kenyan interconnection points even more valuable as traffic aggregation nodes. ## Conclusion Data centre interconnection and peering are not technical niceties, they are fundamental to the value proposition of every data centre in Kenya. KIXP has evolved from a small experimental exchange into a critical piece of national digital infrastructure, handling over 150 Gbps of peak traffic and keeping the majority of locally-sourced content within Kenya's borders. Combined with direct cross-connects within facilities and the growing presence of global cloud providers, Kenya's interconnection ecosystem is a powerful reason why the country is becoming East Africa's undisputed digital hub. For enterprises selecting colocation facilities, for investors evaluating data centre opportunities, and for policymakers shaping the digital economy, understanding interconnection is essential. The facilities and exchanges that invest in connectivity density will be the ones that define Kenya's digital future. ## Frequently Asked Questions ### What is the difference between peering and transit? Peering is a reciprocal arrangement where two networks exchange traffic directly at no cost, typically at an Internet Exchange Point (IXP) like KIXP. Both networks benefit by keeping local traffic local. Transit, by contrast, is a paid service where one network (the transit provider) carries another network's traffic to all destinations on the internet, including international routes. For Kenyan networks, peering at KIXP keeps local traffic within the country, while transit connects them to the global internet via submarine cables like EASSy, Seacom, and TiGS. Most Kenyan ISPs and data centre operators use a combination of both, peering for local content delivery and transit for international reach. The cost differential is significant: peering is typically free or involves a small port fee, while international transit can cost USD 50–200 per Mbps per month. ### How many members does KIXP have and what traffic does it handle? As of October 2026, PeeringDB lists KIXP Nairobi at [144 connected member networks](https://www.peeringdb.com/ix/236) spanning ISPs, mobile network operators, content delivery networks, cloud providers, and enterprise networks, with roughly 2,985 Gbps of connected member port capacity (ISOC Pulse, October 2026). On measured traffic, TESPOK announced a historic peak of 1.3 Tbps on 25 June 2024; the exchange's own statistics site (kixp.or.ke) has been unreachable since at least early October 2026, so more recent traffic levels cannot be independently verified from the operator. KIXP's member base includes Safaricom, Telkom Kenya, Liquid Intelligent Technologies, Jamii Telecommunications, FAIBA (Wananchi Group), and increasingly, global cloud platforms. The exchange operates from multiple points of presence across Nairobi data centres, and its 100G port capabilities have significantly increased capacity for high-bandwidth members. Internet Society Pulse measures that 89% of Kenya's 225 active networks can reach IXP peering either directly or through a member, and 40% of Kenya's 1,000 most-visited websites are served from in-country servers or caches. ### What is a data centre cross-connect and why does it matter? A cross-connect is a physical cable link between two different networks or equipment within the same data centre facility. Unlike connections that route through the public internet, cross-connects provide direct, private, and typically low-latency connectivity between a customer's equipment and their service providers, whether that's an ISP, a cloud provider's on-ramp, a financial trading partner, or another enterprise. In Kenyan [colocation data centres](/articles/colocation-data-centre-kenya), cross-connects are a primary value driver because they enable customers to establish direct connections to multiple networks, cloud platforms, and partners from a single facility. Cross-connects are typically provisioned within 24–72 hours, offer guaranteed bandwidth, and eliminate the 'last mile' costs and latency of accessing networks from separate locations. ### Which global cloud providers are present at KIXP or interconnected in Kenyan data centres? AWS (Amazon Web Services) has established a direct presence in Nairobi, interconnecting with Kenyan networks through local points of presence. Microsoft Azure has expanded its Azure regions and edge nodes closer to East Africa, with interconnection through partner data centres. Google has deployed Google Global Cache nodes at KIXP and within Kenyan data centres, dramatically improving local content delivery. Oracle Cloud Infrastructure (OCI) has also been expanding its footprint in Africa with connectivity options for Kenyan customers. These cloud providers interconnect either directly at KIXP or through dedicated [fibre optic network connections](/articles/fibre-optic-networks-kenya-data-centres) within data centres operated by Africa Data Centres, Liquid Intelligent Technologies, and PAIX. The presence of these providers at Kenyan interconnection points means local enterprises can access cloud services with significantly lower latency than routing through Europe or South Africa. ### How does KIXP compare to NAPAfrica and IXPN? KIXP, NAPAfrica (South Africa), and IXPN (Nigeria) are Africa's three largest Internet Exchange Points, but they differ significantly in scale and maturity. NAPAfrica, operated by INX-ZA in Johannesburg and Cape Town, is Africa's largest IXP by traffic, handling peak volumes exceeding 2 Tbps with over 300 connected members. It benefits from South Africa's larger internet user base and mature telecom market. KIXP is East Africa's leading IXP and the most significant hub for the wider East African region - 144 member networks per PeeringDB (October 2026), roughly 3 Tbps of connected port capacity, and a measured historic peak of 1.3 Tbps (TESPOK, June 2024), impressive for a market of Kenya's size. KIXP has upgraded to 100G ports, narrowing the infrastructure gap with South Africa. IXPN in Nigeria handles roughly 100–120 Gbps peak traffic with 60+ members. KIXP's strategic advantage lies in Kenya's role as the landing point for multiple submarine cables, making it a natural aggregation point for traffic flowing between East Africa, Southern Africa, and the rest of the world. ## Data Centre Physical Security in Kenya: The Layers URL: https://data-centers-254.vercel.app/articles/data-centre-physical-security-kenya Cluster: Security Published: 2026-09-28 Updated: 2026-09-28 Summary: How Kenyan data centres stack physical defences: perimeter, guards, biometrics, mantraps, CCTV retention, rack locks and audit logs. What to verify on a site tour. Ask what makes a data centre secure and most people picture a firewall. In practice, the first attack a Kenyan facility has to survive is physical: someone getting to equipment they should never touch. Every other control, encryption, network segmentation, compliance certificates, assumes one thing, that only authorised people can physically reach the hardware. That is why physical security is Layer 1 in [our security model](/articles/data-centre-security-explained), and why it deserves its own inspection, layer by layer. ## The perimeter: delay and detect before the door The outermost layer begins at the property boundary. For major facilities in Nairobi, that means perimeter walls or fences typically 2.5 to 3 metres high with anti-climb features, CCTV covering all approaches, vehicle barriers such as bollards or crash-rated gates to stop ram-raiding, and guards at the main entrance. The design goal is not one perfect wall. It is to force any approach to consume time, make noise, and leave a record before it reaches the building. Detection matters as much as delay. A fence only helps if someone is watching it, which is why perimeter CCTV, patrol routines, and intrusion sensors on gates and fences are the difference between a boundary and a formality. Roof access points are secured and monitored, and facilities assess adjacent buildings or vantage points that could overlook sensitive areas. ## Building access: badges, biometrics, and the mantrap Inside the perimeter, access becomes a sequence of checkpoints. Reception identifies visitors and verifies them against a pre-approved list before issuing a temporary badge scoped to specific areas and times. Employees and regular contractors authenticate with a combination of proximity cards, biometric readers such as fingerprint or iris scanners, and PINs, so that a stolen badge alone does not open anything. The critical control is the mantrap: a vestibule with two interlocking doors where the first locks behind you and the second opens only after authentication. It defeats tailgating, the single most common physical breach technique, and in facilities serving banks and telcos it is usually paired with a guard's visual check. The GSC-era reality is that buyers ask for this specifically; "badge, retinal scan, mantrap" is now a shorthand tenants use when they describe what a serious facility looks like. ## Inside the white space: suites, racks, and logging Past the mantrap, security becomes granular. In a carrier-neutral facility, access is scoped so that customers cannot physically reach other customers' equipment: suite doors, cage areas, and individually locked racks, with electronic locks replacing mechanical keys where audit granularity matters. Every event, every door, every badge, every biometric acceptance or rejection, lands in an access control system that keeps records typically for 90 days to one year. Logging is what converts hardware into evidence. In a well-run facility you can reconstruct exactly who entered which area at what time, for any date in the retention window. That capability serves incident investigation, customer audits, and, where it comes to it, prosecution under the Computer Misuse and Cybercrimes Act 2018, which requires proof of unauthorised access. Physical security done well does not just stop people; it produces the record that makes stopping them legally meaningful. ## The layer everyone forgets: the loading bay and the vendors Most physical breaches in real facilities do not happen at the front door, they happen where goods and people routinely flow: the loading bay, the generator yard, and the vendor intake process. A delivery van arriving with replacement UPS batteries is a legitimate event; an unverified "technician" arriving with a toolbox and a confident story is a classic intrusion pattern. Kenyan facilities that handle this well run escorted vendor access, verify work orders against a pre-registered contact at the customer, seal and inspect equipment movements, and treat every after-hours delivery as an exception that needs a named approver. ## What Kenya's threat environment changes Kenya's facilities defend against the same attacks as any market, with one local sharpening: the country's history of sophisticated social engineering makes the human layer a primary target rather than an afterthought. A biometric reader is worthless if a guard is talked into propping a door, which is why the [attack patterns that matter in Kenya](/articles/data-centre-security-threats-kenya) pair technical intrusion with social engineering, and why procedures, guard briefing, escort rules, disposal of decommissioned drives, are as much a part of physical security as steel and sensors. ## What the leading Kenyan facilities actually run Kenya's leading operators, iXAfrica, Africa Data Centres, and Safaricom, have invested heavily in this stack. iXAfrica's NBOX1 was designed to Tier III security requirements from the ground up: biometric access, mantraps, 24/7 CCTV surveillance, and on-site security personnel. Africa Data Centres applies group-wide security standards inherited from its pan-African parent. Maturity varies across the wider market, smaller enterprise server rooms and older buildings do not all match this standard, which is exactly why verification, not assumption, is the buyer's job. If you are evaluating a facility, the question is not "is it secure?" but "what exactly does it run, and can you prove it?" Ask which controls are in place at each of the layers above, whether the operator holds [ISO 27001 certification](/articles/iso-27001-data-centre-kenya) and for what scope, how access logs are exported for your audit, and what the [fire and environmental controls](/articles/fire-suppression-systems-kenyan-data-centres) look like next to the physical ones. The [DC254 directory](/directory) lists tracked facilities so you can shortlist and then verify in person. ## Data Centre PUE Explained: Power Efficiency in Kenya URL: https://data-centers-254.vercel.app/articles/data-centre-pue-power-usage-effectiveness Cluster: Energy Efficiency Published: 2026-08-28 Updated: 2026-08-28 Summary: PUE is the key metric for data centre energy efficiency. How it is calculated, what good values look like, and how Kenyan facilities improve. If you run a data centre that consumes 1 megawatt of power for your servers but your total electricity bill is for 1.7 megawatts, you are spending 70% more on electricity than your IT equipment actually needs. The extra 0.7 megawatts is going to cooling, lighting, power conversion losses, and other overhead. This ratio (total power divided by IT power) is the Power Usage Effectiveness (PUE), and it is the single most important metric for understanding how efficiently a data centre uses energy. ![Aerial view of a data centre campus with rooftop solar](/images/dc-environment-sustainability.webp) PUE was developed by The Green Grid, an industry consortium, in 2007. It has become the universal language of data centre energy efficiency, used by operators to track their performance, by customers to compare facilities, by investors to assess operational quality, and by regulators to evaluate environmental impact. Understanding PUE is essential for anyone involved in data centres in Kenya, where power costs are a significant operating expense and where the renewable grid provides both an opportunity and a responsibility to use energy wisely. ## How PUE Is Calculated The formula is simple: **PUE = Total Facility Power / IT Equipment Power** Total Facility Power is everything the facility draws from the grid (or from generators): servers, storage, network equipment, UPS systems, cooling equipment, lighting, security systems, and all other electrical loads. IT Equipment Power is only the power consumed by the computing equipment: servers, storage, and network devices that perform actual work for the facility's customers. A PUE of 1.0 would mean that every watt drawn from the grid goes to IT equipment, zero overhead. This is physically impossible in practice, because some power must always be used for cooling, lighting, and power conversion losses. The theoretical minimum PUE depends on the climate and the cooling technology, but even in the most efficient facilities, PUE rarely drops below 1.05. ## What PUE Values Mean ### PUE 1.0–1.2: Exceptional Facilities in this range are among the most efficient in the world. They typically use advanced cooling technologies (free cooling, liquid cooling, or both), operate in cool climates, and have been purpose-designed for efficiency. Google publishes average annual PUE values of approximately 1.10 for its newest facilities. In Kenya, achieving PUE below 1.2 is challenging but possible for new facilities that leverage Nairobi's climate for free cooling and use hot/cold aisle containment. ### PUE 1.2–1.5: Good ![Rooftop chiller units at a data centre](/images/dc-chillers-roof-2.webp) This is the range that well-designed and well-operated facilities in temperate climates achieve. In Kenya, newer facilities like iXAfrica's NBOX1 target this range. These facilities use a combination of free cooling and mechanical cooling, have good airflow management (hot/cold aisle containment or chimney cabinets), and use efficient power distribution equipment. Most new colocation facilities being built in Kenya should aim for this range. ### PUE 1.5–1.8: Average This is the range where most existing Kenyan facilities operate. It reflects adequate but not optimised cooling, some airflow management inefficiencies, and power distribution equipment that is not the most efficient available. A PUE in this range is not bad (it is average) but it represents an opportunity for improvement. For a 2MW data centre, moving from PUE 1.7 to PUE 1.4 would save approximately KES 50–70 million per year in electricity costs. ### PUE Above 2.0: Inefficient A PUE above 2.0 means that overhead (cooling, lighting, power losses) consumes more power than the IT equipment itself. This is typically found in older facilities, facilities in hot and humid climates without adequate cooling design, or facilities that have been poorly operated or maintained. In Kenya, some older enterprise data centres and poorly designed server rooms may fall into this range. A PUE above 2.0 should be investigated, the causes are usually identifiable and addressable. ## What Drives PUE in Kenyan Data Centres ### Cooling (The Biggest Factor) Cooling is typically the largest single contributor to PUE overhead, accounting for 30–50% of total facility power in a facility without free cooling. In Nairobi's climate (average 18–25°C, moderate humidity), free cooling (using outside air to cool the data centre rather than mechanical refrigeration) can be used for significant portions of the year. Facilities that maximise free cooling achieve lower PUE values. The [cooling systems guide](/articles/data-centre-cooling-systems-explained) covers the technologies in detail. For PUE optimisation, the key strategies are: maximise free cooling hours, implement hot/cold aisle containment to prevent mixing of hot and cold air, use variable-speed drives on cooling fans and pumps, and maintain cooling equipment to ensure peak efficiency. ### Power Distribution Losses Power undergoes multiple conversions between the grid and the server. Each conversion, from the utility's high voltage to the facility's medium voltage, from medium voltage to the UPS input, from the UPS output to the PDU, and from the PDU to the server, involves some energy loss in the form of heat. Modern, high-efficiency transformers and UPS systems can minimise these losses, but they cannot be eliminated entirely. The typical power chain efficiency in a Kenyan data centre is approximately 88–92%. This means that for every 100 watts drawn from the grid, only 88–92 watts reach the IT equipment, with 8–12 watts lost as heat in transformers, UPS systems, and PDUs. Choosing high-efficiency equipment (97%+ efficient transformers, 96%+ efficient UPS systems in eco-mode) can improve this significantly. ### Lighting and Other Loads ![Rooftop exhaust fans and HVAC ducting](/images/dc-chillers-roof-3.webp) Lighting, security systems, office areas, and other miscellaneous loads typically account for 3–5% of total facility power. While individually small, these loads contribute to PUE. LED lighting, occupancy sensors, and efficient building management can minimise these loads. Some facilities have reduced lighting loads to negligible levels by using LED fixtures with occupancy sensors and by minimising the lit area (lighting only aisles where personnel are present). ## Measuring PUE Accurately Accurate PUE measurement requires power metering at two points: the total facility power input (typically at the utility meter or generator output) and the IT equipment power (measured at the output of the facility's power distribution system, typically at the PDU level). The challenge in Kenya is that many facilities lack the metering infrastructure needed for accurate, continuous PUE measurement. Without power meters at the right points, PUE must be estimated, and estimates can be significantly inaccurate. Investing in proper power metering (sub-meters at PDUs, branch circuit monitors, and a centralised monitoring system) is the first step in any PUE improvement programme. The Green Grid recommends measuring PUE continuously and reporting the annual average. Measuring PUE only during favourable conditions (cool nights, mild weather) produces misleadingly good numbers. The annual average captures seasonal variations and provides a fair representation of the facility's efficiency. ## Strategies to Improve PUE in Kenya ### Hot/Cold Aisle Containment Hot/cold aisle containment is the single most cost-effective PUE improvement for most existing facilities. By physically separating the cold air supply from the hot air exhaust (using doors, curtains, or ceiling panels) the cooling system delivers cold air more efficiently and does not waste energy cooling air that is already hot. Containment can reduce cooling energy by 10–25%, typically improving PUE by 0.1–0.3 points. ### Raise the Temperature ![Data centre rooftop chillers and generators](/images/dc-chillers-roof.webp) ASHRAE's recommended environmental range for data centres allows inlet temperatures up to 27°C (with appropriate humidity management). Many older facilities operate at 18–20°C, significantly cooler than necessary. Raising the set point by even 2–3 degrees reduces cooling energy consumption by 5–15%. In Nairobi's climate, operating at 24–26°C with appropriate containment allows more hours of free cooling and reduces mechanical cooling requirements. ### Optimise Free Cooling Nairobi's climate allows free cooling for approximately 4,000–6,000 hours per year (depending on the specific temperature and humidity thresholds). Maximising these hours (through proper controls, appropriate set points, and well-maintained cooling equipment) is the most impactful PUE strategy for Kenyan facilities. Facilities that do not use free cooling at all are leaving significant efficiency gains on the table. ### Variable-Speed Drives Cooling fans and pumps that run at fixed speed consume the same amount of energy regardless of the cooling demand. Variable-speed drives (VSDs) allow fans and pumps to slow down when cooling demand is low, reducing energy consumption proportionally. VSDs are a standard feature on modern cooling equipment but may be absent in older installations. PUE is not just a metric, it is a management tool. By measuring PUE continuously, setting targets, and tracking improvement over time, data centre operators in Kenya can reduce their operating costs, improve their environmental credentials, and demonstrate to customers that they are operating efficiently. In a market where power costs are a significant competitive factor, PUE improvement is a direct path to better margins and stronger market positioning. ## Data Centre Security: Physical, Network, Compliance URL: https://data-centers-254.vercel.app/articles/data-centre-security-explained Cluster: Security Published: 2026-08-28 Updated: 2026-08-28 Summary: Data centre security runs on three layers: physical access, network defences, and compliance frameworks like ISO 27001. How Kenyan facilities compare. When a bank chooses to house its core banking servers in a data centre, it is trusting that facility with the digital equivalent of its vault. When a government agency places citizen data (tax records, health information, biometric data) on servers in a colocation facility, it is trusting that the data will not be accessed, modified, or stolen by anyone who should not have access. This trust is not given freely. It is earned through security systems, policies, and certifications that operate on multiple layers simultaneously. ![Biometric access control and CCTV at a data centre](/images/dc-security.webp) Data centre security is not a single thing you can point to and say "there, that is security." It is a system of systems, physical barriers, electronic access controls, network defences, encryption protocols, compliance frameworks, and human procedures, all designed to work together so that a failure in any one layer does not compromise the whole. Understanding these layers is essential for anyone evaluating a data centre, whether as a potential customer, an employee, or an investor. ## Layer 1: Physical Security Physical security is the most visible and intuitive layer of data centre protection. Its purpose is straightforward: prevent unauthorised people from getting close to the equipment, and ensure that every person who does enter is identified, authenticated, and logged. ### Perimeter Security The outermost layer begins at the property boundary. For major data centre facilities in Nairobi, this includes perimeter walls or fences (typically 2.5–3 metres high, often with anti-climb features), CCTV cameras covering all approaches, vehicle barriers (bollards or crash-rated gates) to prevent ram-raiding, and security guards at the main entrance. The goal is to detect and deter any attempt to approach the facility before an intruder reaches the building itself. Perimeter security also extends below ground and above the building. Anti-tunnelling measures, while less common in Kenya than in high-threat environments, may include ground-penetrating radar or buried sensors. Roof access points are secured and monitored, and adjacent buildings or vantage points that could overlook sensitive areas are assessed for surveillance risk. ### Building Access Control Once inside the perimeter, access to the building itself is controlled through a series of checkpoints. The first is typically the reception area, where visitors are identified, their visit is verified against a pre-approved list, and they are issued a temporary access badge. This badge grants access only to specific areas and only during specific times. For employees and regular contractors, access is managed through a combination of proximity cards (RFID badges), biometric readers (fingerprint or iris scanners), and personal identification numbers (PINs). The most critical access control mechanism in a data centre is the **mantrap**. A mantrap is a small room or corridor with two interlocking doors, you enter through the first door, which locks behind you, and you must be authenticated before the second door opens. This prevents "tailgating," where an unauthorised person follows closely behind an authorised person through a door. Tailgating is one of the most common and difficult-to-detect physical security breaches, and mantraps are the primary defence against it. ### Server Room and Rack-Level Security Inside the server room, security becomes more granular. Access to individual rows or aisles of racks may be further restricted, with some areas accessible only to specific customers (in a colocation facility) or to specific roles (network engineers vs. facility engineers). Individual racks are locked with physical keys or electronic locks, and in carrier-neutral facilities, customers cannot physically access other customers' equipment. Every access event (every door opened, every badge scanned, every biometric authentication) is logged in an access control system. These logs are retained for extended periods (typically 90 days to 1 year) and are used for audit purposes, incident investigation, and compliance verification. In a well-run facility, it should be possible to reconstruct exactly who accessed which area at what time, for any given date within the retention period. For the Kenyan picture of this layer in depth, perimeter standards, mantraps, and what to verify on a site tour, see [physical security in Kenyan data centres](/articles/data-centre-physical-security-kenya). ## Layer 2: Network Security While physical security protects the hardware, network security protects the data flowing through it. A data centre without network security is like a bank vault with the door left open, the servers may be physically secure, but the data they process and store is accessible to anyone who can reach them over the network. ![Comparison of SSH, TCP and VPN roles](/images/diagram-ssh-tcp-vpn.webp) ### Perimeter Network Defence The first line of network defence is the perimeter firewall. A data centre typically deploys enterprise-grade firewalls (from vendors like Palo Alto Networks, Fortinet, Cisco, or Check Point) at the network boundary to control traffic entering and leaving the facility. These firewalls enforce rules about what types of traffic are allowed, block known malicious traffic patterns, and can perform deep packet inspection to detect threats embedded in legitimate-looking traffic. Beyond firewalls, data centres deploy intrusion detection and prevention systems (IDS/IPS) that monitor network traffic for suspicious patterns, unusual connection attempts, data exfiltration, or traffic that matches known attack signatures. These systems can automatically block suspicious traffic and generate alerts for the security operations team. ### DDoS Protection Distributed Denial of Service (DDoS) attacks (where attackers flood a target with traffic from many sources, overwhelming its capacity to respond) are a persistent threat to data centres and their customers. DDoS attacks in East Africa have increased significantly in recent years, driven by both ideological hacktivism and extortion attempts. Data centres mitigate DDoS attacks through a combination of network-level defences (traffic scrubbing, rate limiting, and blackhole routing) and partnerships with DDoS mitigation providers (like Cloudflare, Akamai, or NEUSTAR) that can absorb and filter attack traffic before it reaches the data centre. The largest facilities maintain dedicated DDoS mitigation capacity on-site, while smaller facilities rely on upstream providers or cloud-based mitigation services. ![Surveillance camera overlooking a server hall](/images/dc-security-camera-2.webp) ### Zero Trust Architecture The modern approach to data centre network security is moving toward "zero trust" architecture, which operates on the principle that no user, device, or network segment should be trusted by default, even if it is inside the data centre's network perimeter. Every access request must be authenticated, authorised, and encrypted, regardless of where it originates. This is a significant shift from the traditional model, where traffic inside the perimeter was considered safe and only external traffic was treated with suspicion. Zero trust in a data centre context means implementing micro-segmentation (dividing the network into small, isolated segments with their own access controls), mutual TLS authentication between services, and continuous monitoring of all network traffic for anomalous behaviour. While full zero trust implementation is still evolving, Kenyan data centres serving international customers are increasingly expected to demonstrate progress toward this model. ## Layer 3: Data Security and Encryption The third layer protects the data itself, regardless of whether it is at rest (stored on a server) or in transit (moving across the network). Data security measures include encryption, access controls at the application and database level, and backup and recovery procedures. ### Encryption Encryption converts data into an unreadable format that can only be decrypted with the correct key. Modern data centres encrypt data both at rest (using AES-256 or similar algorithms) and in transit (using TLS 1.3 for network traffic and encrypted connections for storage replication). For colocation customers, encryption is often the customer's responsibility, the data centre provides the physical and network security, while the customer encrypts their own data. However, some facilities offer encryption-as-a-service for customers who want to outsource key management. ### Access Control and Identity Management At the application and data level, access is managed through identity and access management (IAM) systems. These systems define who can access what data, under what conditions, and log all access for audit purposes. In a well-managed data centre environment, access follows the principle of least privilege, users are granted only the minimum access necessary to perform their job, and access is revoked when no longer needed. ## Compliance Frameworks Security is not just about technology, it is about demonstrating to customers, regulators, and partners that specific standards are being met. Compliance frameworks provide the structure for this demonstration. ### ISO/IEC 27001 ISO 27001 is the international standard for information security management systems (ISMS). It specifies a framework of policies and procedures that an organisation must implement to manage information security risks. For data centres, ISO 27001 certification demonstrates that the facility has a systematic approach to identifying, assessing, and treating information security risks, and that this approach is regularly audited by an independent certification body. In Kenya, the leading data centre operators are either ISO 27001 certified or in the process of obtaining certification. For international customers (banks, cloud providers, multinational corporations) ISO 27001 is often a non-negotiable requirement for choosing a colocation facility. Without it, a data centre is effectively excluded from competing for the most valuable customers. Our [ISO 27001 for Kenyan data centres](/articles/iso-27001-data-centre-kenya) guide covers the audit cycle in full, and how certification maps onto the [Data Protection Act](/articles/kenya-data-protection-act-data-centres), the law that makes demonstrable security measures a legal duty. ![Secured server racks with locking cabinets](/images/dc-gpu-cluster-2-wide.webp) ### SOC 2 Type II SOC 2 (Service Organisation Control 2) is an auditing standard developed by the American Institute of Certified Public Accountants. A SOC 2 Type II report evaluates an organisation's systems over a period of time (typically 6–12 months) to verify that they consistently meet security, availability, processing integrity, confidentiality, and privacy criteria. While more common in the North American market, SOC 2 compliance is increasingly expected by international technology companies operating in Kenya. ### PCI-DSS The Payment Card Industry Data Security Standard (PCI-DSS) applies to any entity that stores, processes, or transmits payment card data. Data centres that host payment processing infrastructure (and in Kenya, this includes facilities serving banks, mobile money operators, and payment service providers) must demonstrate PCI-DSS compliance. This includes specific requirements for network segmentation, access control, encryption, and regular security testing. ## Security in Kenyan Data Centres: The Reality Kenya's leading data centre operators (iXAfrica, Africa Data Centres, and Safaricom) have invested significantly in security infrastructure. iXAfrica's NBOX1 facility, for example, was designed from the ground up with Tier III security requirements, including biometric access, mantraps, 24/7 CCTV surveillance, and on-site security personnel. Africa Data Centres, as part of the pan-African Cassava Technologies group, applies group-wide security standards that align with international best practices. ![Security cameras monitoring a colocation facility](/images/dc-security-camera.webp) However, security maturity varies across the Kenyan market. Smaller operator-built facilities and enterprise data centres may not have the same level of investment in security infrastructure, and the skills required to operate and maintain sophisticated security systems (network security engineers, security operations analysts, compliance auditors) are in short supply. As the market matures and customers become more demanding, security differentiation will become an increasingly important competitive factor. For anyone evaluating a Kenyan data centre, the question should not be "is this facility secure?" but rather "what specific security measures does this facility implement, what certifications does it hold, and can I verify these claims through audit reports and site visits?" Security is not a checkbox, it is a continuous process of assessment, improvement, and verification. ## What Can Hit a Kenyan Data Centre, and the Fix URL: https://data-centers-254.vercel.app/articles/data-centre-security-threats-kenya Cluster: Security Published: 2026-09-16 Updated: 2026-09-28 Summary: DDoS, ransomware, data harvesting, insiders and physical intrusion: the attack types that hit Kenyan data infrastructure, real examples, and the defences that work. Ask someone to picture a data centre attack and they usually picture a hoodie and a keyboard. Kenya's documented record tells a broader story. The incidents that have actually hurt Kenyan institutions include a political hacktivist flood, a defaced presidential website with a Bitcoin ransom note, an iris-scanning operation that broke data protection law, and a registry leak touching two million companies. A data centre in Kenya has to defend against all of it, at once, in layers. This article pairs each attack type with its Kenyan example and its working defence. It is the threat-side companion to our [general security explainer](/articles/data-centre-security-explained), and it sticks to documented cases; where an example comes from government or media reporting rather than a data centre specifically, we say so. ![A staff badge being read by a biometric access panel at a secure facility](/images/dc-biometric-access-3.webp) ## Threat one: DDoS, the flood that takes the door offline A distributed denial of service attack does not break in; it crowds the entrance. Thousands of machines, often hijacked, send junk traffic until legitimate users cannot get through. Kenya lived this at national scale in late July 2023, when the group Anonymous Sudan flooded government platforms and degraded the eCitizen portal, the gateway to more than 5,000 public services (BBC, 28 July 2023). People could not buy electricity tokens or reach government payment services during the disruption. **The defence.** You cannot out-muscle a flood at your own door, so the filtering happens upstream: internet service providers and scrubbing services drop the junk before it reaches the facility, content delivery networks absorb attack traffic at the edge, and rate limiting protects whatever survives. Facilities that connect through multiple upstreams and buy DDoS protection as a service recover in minutes; portals that rely on one pipe learn in public. The eCitizen experience is the local case study in why availability architecture and security architecture are the same discipline. ## Threat two: defacement and ransom, the attack meant to be seen Some attackers do not want your data; they want your homepage. On 18 July 2026, hackers breached president.go.ke, replaced its content with an anti-government message, and demanded five Bitcoin, roughly $320,000 (Anadolu Agency, 18 July 2026). The site was restored within about two days and the ransom was not paid, and Kenya opened an investigation into how a flagship web property was compromised at all (The Record, 21 July 2026). The technique behind it is usually mundane: an unpatched content management system, a weak credential, or an exposed admin panel. **The defence.** Patching discipline, a web application firewall in front of public sites, separating public web servers from internal systems, and keeping clean offline backups so restoration is a restore, not a negotiation. The presidency's case ended the way a well-rehearsed one should: offline, cleaned, restored, investigated. The embarrassment was temporary; the backup discipline is what made it temporary. ## Threat three: data harvesting that breaks the law before it breaks anything Not every data disaster is a hack. In August 2023, WorldCoin was collecting Kenyans' iris scans and personal data in exchange for cryptocurrency tokens, a mass collection exercise that ran until the Office of the Data Protection Commissioner investigated it on 2 August and suspended operations on 3 August 2023 (Data Guidance, 1 August 2023). For data centre operators the lesson is uncomfortable: a facility can be perfectly secure and still host a compliance catastrophe, because the tenant controls what the tenant collects. **The defence.** Contractual data processing terms, tenant due diligence, and knowing which facilities hold personal data subject to the Data Protection Act 2019. Since the ODPC issued its first penalty notices in September 2023 and can fine up to five million shillings or one percent of turnover, compliance review is now part of commercial risk, not paperwork. Colocation providers that can show clean data governance increasingly win the regulated workloads, banks and government, that fund the best halls. ## Threat four: ransomware and the quiet encryption of everything Kenya's public record has fewer named ransomware cases than its banks' risk reports imply, but the pattern is African and global at once: a phishing email, one stolen VPN session, days of quiet lateral movement, then encrypted systems and a ransom note. The defences are equally well rehearsed. **The defence.** Multi-factor authentication on every remote access path, network segmentation so a compromised laptop cannot reach the management plane, tested offline backups, and an incident response plan with phone numbers in it. The unglamorous truth is that ransomware resilience is mostly hygiene: the organisations that recover in hours did the boring things months earlier, and those that pay ransoms usually still have to restore from backups anyway. ![A surveillance camera mounted above a data centre corridor](/images/dc-security-camera.webp) ## Threat five: the insider and the human layer The 2025 Business Registration Service breach, which exposed records connected to roughly two million companies including ownership and beneficial owner details (Nation Africa, 1 February 2025), is a reminder that registries and databases are intelligence targets. Not every exposure begins as an outside intrusion: misconfigured access, over-privileged accounts and social engineering move data out the side door, and insiders know where the cameras are not. **The defence.** Least privilege, so accounts hold only the access a role needs; logging that answers who touched what; separation of duties for sensitive operations; and background-checked, trained staff at every level of the facility. Physical insiders matter too, which is why Tier III facilities log every entry, escort visitors, and camera every aisle. Security awareness training sounds soft until you price one leaked database. ## Threat six: physical intrusion, the oldest attack there is Everything above can be undone by one person with a screwdriver and a bad plan. Kenyan facilities built to international standards defend in depth: perimeter fencing and vehicle controls, guards, biometric readers, mantrap doors that admit one person at a time, CCTV coverage with retention, rack-level locks, and disposal rules for decommissioned drives. Kenya's threat environment, including its history of sophisticated social engineering, makes the physical layer a real control rather than a checkbox. **The defence.** Layers, again, because the goal is not one perfect wall but multiple delays and records: delay at the perimeter, detect at the doors, record everywhere, and audit the logs. Physical security done well also produces evidence, which matters under the Computer Misuse and Cybercrimes Act 2018, where a prosecution needs proof of unauthorised access. The full layer-by-layer breakdown lives in our [physical security in Kenyan data centres](/articles/data-centre-physical-security-kenya) guide. ![Diagram comparing encrypted remote access paths including VPN and SSH](/images/diagram-ssh-tcp-vpn.webp) ## Putting the layers together No single control stops six threat classes; the design goal is that every attack path crosses several controls, any one of which can catch it. In practice, a well-run Kenyan facility looks like this: upstream DDoS filtering and a hardened edge for the flood risk; patched, firewalled, segmented systems with multi-factor access for the ransomware and defacement risk; contracts and compliance review for the harvesting risk; least privilege and logging for the insider; and layered physical controls with full audit trails for everything that walks. The KE-CIRT/CC numbers put the scale in perspective: over 840 million detected threat events in a single quarter of 2024. Being attacked constantly is the baseline condition of operating in Kenya, the same as everywhere else. The facilities that survive it are not the ones that bought the most gear; they are the ones that layered the controls, rehearsed the response, and kept records good enough to learn from when, not if, something gets through. ## What Is a Data Centre SLA? Uptime Guarantees URL: https://data-centers-254.vercel.app/articles/data-centre-sla-uptime-guarantees Cluster: Beginner Guides Published: 2026-08-28 Updated: 2026-08-28 Summary: What a data centre SLA is, how 99.9% and 99.999% uptime guarantees translate to real downtime, and what Kenyan operators commit to. ## What Exactly Is a Data Centre SLA? A **Service Level Agreement (SLA)** is a formal contract between a data centre operator and its customer that defines the minimum standards of service the operator promises to deliver. Think of it as a written guarantee: if the facility fails to keep the lights on, keep your servers cool, or respond when you need help, the SLA spells out exactly what consequences follow. ![Data centre power distribution and backup systems](/images/dc-power-systems.webp) For any business placing IT equipment in a [colocation facility in Kenya](/articles/what-is-colocation-kenya), the SLA is arguably the single most important document you will sign. It transforms vague marketing promises like "we never go down" into measurable, enforceable commitments. Without a solid SLA, you have no contractual recourse when things go wrong, and in the data centre world, things eventually do. According to the [Uptime Institute](https://uptimeinstitute.com/tier-standard-topology), the global authority on data centre reliability, SLAs should be directly tied to the facility's design topology and operational practices. A well-crafted SLA protects both parties: the customer gets assurance, and the operator gets clear boundaries around their obligations. ## Why SLAs Matter for Your Business Imagine running a fintech application serving M-Pesa integrations from a Nairobi facility. Every minute of downtime means failed transactions, frustrated customers, and potential regulatory scrutiny from the Central Bank of Kenya. An SLA is not just a technical document, it is a business continuity tool. **The core reasons SLAs matter include:** - **Financial protection:** Service credits provide partial compensation when the operator fails to meet agreed standards. - **Accountability:** Defined metrics give you a clear basis to evaluate whether the provider is delivering value. - **Planning certainty:** Knowing your guaranteed uptime helps you design your own disaster recovery and high-availability architectures. - **Vendor comparison:** Standardised SLA metrics let you objectively compare providers like Africa Data Centres, PAIX, Liquid, and IXAfrica. ## Key SLA Metrics Explained ### Uptime Percentage: The Numbers That Matter Uptime is the headline metric of any data centre SLA. It represents the percentage of time in a given period that the facility's critical systems (power, cooling, and network) are fully operational. Here is what the common tiers actually mean in terms of **annual unplanned downtime**: | Uptime Tier | Downtime Per Year | Downtime Per Month | Typical Use Case | |---|---|---|---| | **99.0%** (Two nines) | 3 days, 15 hours | 7 hours, 18 minutes | Development and testing environments | | **99.9%** (Three nines) | 8 hours, 46 minutes | 43 minutes, 50 seconds | Standard business applications | | **99.99%** (Four nines) | 52 minutes, 36 seconds | 4 minutes, 23 seconds | Mission-critical workloads, fintech, e-commerce | | **99.999%** (Five nines) | 5 minutes, 16 seconds | 26 seconds | Ultra-critical systems, telecoms core, trading platforms | The jump from three nines to four nines is where most serious businesses draw the line. For a Kenyan e-commerce company processing thousands of orders daily, 8 hours and 46 minutes of annual downtime could mean millions of shillings in lost revenue. Four nines (less than an hour per year) is a far more comfortable target. ### Power Usage Effectiveness (PUE) Guarantees Some advanced SLAs now include **PUE commitments**, promising that the facility will operate within a certain energy efficiency range. A PUE of 1.5 means that for every 1 kW of power delivered to your IT equipment, the facility uses 1.5 kW total (including cooling and overhead). Lower PUE values indicate better efficiency, which directly impacts your power costs, a critical factor given Kenya Power's commercial tariff structures. ### Temperature and Humidity Ranges **ASHRAE-recommended** operating ranges are the industry baseline. Most SLAs guarantee that the cold aisle temperature will remain between **18°C and 27°C**, with relative humidity between **20% and 80%**. In Kenya, where ambient temperatures in Nairobi average 18–25°C but Mombasa can exceed 32°C, the cooling SLA is especially important for coastal facilities. ### Response Times: Remote Hands and Smart Hands **Remote hands** refers to basic physical tasks (checking server status lights, rebooting equipment, connecting cables. **Smart hands** involves more skilled work) OS-level troubleshooting, hardware replacement, configuration changes. A typical SLA might guarantee: - **Remote hands response:** 15 minutes or less - **Smart hands response:** 30 minutes to 2 hours depending on complexity For Kenyan operators, response time SLAs are particularly important because many customers manage their infrastructure remotely from other African markets or from overseas. ### Ticket Resolution Times Beyond initial response, SLAs should define **resolution timeframes** for different severity levels: - **Critical (P1):** Complete service outage, target resolution within 1–4 hours - **High (P2):** Major degradation, target resolution within 4–8 hours - **Medium (P3):** Minor issue with workaround available, target resolution within 24 hours - **Low (P4):** General enquiry or cosmetic issue, target resolution within 72 hours ### Network Availability For [understanding data centre tier ratings](/articles/data-centre-tier-ratings-explained) and their associated network SLAs, the guarantee typically covers carrier diversity, latency to major internet exchange points like the **Kenya Internet Exchange Point (KIXP)**, and bandwidth throughput. Most Kenyan colocation providers guarantee **99.9% to 99.99% network availability**. ![Rooftop chiller units at a data centre](/images/dc-chillers-roof-2.webp) ## How SLAs Relate to Data Centre Tier Ratings The [Uptime Institute Tier Standard](https://uptimeinstitute.com/tier-standard-topology) provides a useful framework for understanding what level of SLA a facility can realistically support: ### Tier I, Basic Capacity (No Meaningful SLA) Tier I facilities have no redundancy. A single power feed, a single cooling path, and no backup systems mean that any component failure causes an outage. These facilities typically cannot offer any meaningful uptime SLA. In Kenya, very few commercial colocation providers operate at Tier I. ### Tier II, Redundant Capacity Components Tier II adds redundant components (like an extra UPS or chiller) but still relies on a single distribution path. SLAs here typically hover around **99.5% to 99.9%**. Suitable for non-critical workloads. ### Tier III, Concurrently Maintainable (N+1) Tier III is the sweet spot for most commercial data centres. With N+1 redundancy, any single component can fail or be taken offline for maintenance without affecting operations. These facilities typically guarantee **99.982% to 99.99% uptime**. In Kenya, operators like **Africa Data Centres** (their Nairobi facilities) and **IXAfrica** design to Tier III standards and publish SLAs in this range. ### Tier IV, Fault Tolerant (2N) Tier IV provides fully redundant, independent distribution paths. Even during maintenance, a complete failure of one path will not cause downtime. These facilities can credibly guarantee **99.995% or higher uptime**. Africa Data Centres' premium facilities and some Liquid Intelligent Technologies campuses in South Africa target Tier IV, though genuine Tier IV design in East Africa remains relatively rare. ## SLA Enforcement: Service Credits and Their Limits ### How Service Credits Work Service credits are the primary enforcement mechanism in most data centre SLAs. When the operator fails to meet a committed metric, the customer receives a credit against their monthly invoice. A typical service credit structure looks like this: | SLA Breach | Service Credit | Example (KES 500,000/month) | |---|---|---| | Below 99.9% uptime | 10% of monthly fee | KES 50,000 credit | | Below 99.5% uptime | 25% of monthly fee | KES 125,000 credit | | Below 99.0% uptime | 50% of monthly fee | KES 250,000 credit | | Below 95.0% uptime | 100% of monthly fee | KES 500,000 credit | ### The Critical Limitation of Service Credits Here is the uncomfortable truth that every SLA signer must understand: **service credits almost never cover your actual business losses.** If a four-hour outage costs your fintech platform KES 10 million in failed transactions, a 25% service credit on a KES 500,000 colocation bill gives you just KES 125,000 back. That is a 1.25% recovery of your actual loss. This is not a flaw in the SLA, it is the standard industry model worldwide. Service credits are designed as a **contractual incentive**, not an insurance policy. For true loss protection, you need separate business interruption insurance. ### Dispute Resolution Most SLAs include a dispute resolution clause specifying that disagreements about SLA performance must be raised within **30 to 60 days** of the alleged breach, often requiring the operator's own monitoring data as evidence. This makes independent monitoring on your side essential. ## Kenya-Specific SLA Considerations ### What Kenyan Operators Actually Publish The major Kenyan data centre operators each take a different approach to their published SLAs: - **Africa Data Centres** (East Africa's largest operator, with facilities in Nairobi): Typically guarantees **99.99% uptime** for power and cooling across their Tier III+ facilities. Their SLA also covers remote hands response within 15 minutes. - **PAIX (Pan-African Internet Exchange):** Historically offered **99.9% uptime** guarantees with a focus on network-centric SLAs, given their peering and interconnection heritage. - **Liquid Intelligent Technologies:** Offers SLAs in the **99.9% to 99.99% range** depending on the facility and service tier, leveraging their fibre backbone across East Africa. - **IXAfrica:** Kenya's newest hyperscale-ready carrier-neutral facility, targeting **99.99%+ uptime** with a strong emphasis on PUE efficiency and sustainability metrics in their SLA. ### The Power Reliability Challenge ![Diagram of subsea cable architecture from landing station to POP](/images/submarine-cable-diagram.webp) Kenya's power infrastructure has improved significantly, but [UPS and backup power systems remain critical](/articles/ups-backup-power-kenyan-data-centres) for SLA compliance. Kenya Power's grid reliability in Nairobi has improved with underground cabling projects, but regional variations persist. A data centre in Mombasa may face different grid challenges than one in Nairobi's Industrial Area. The reality is that no Kenyan data centre can meet a 99.99% SLA on grid power alone. Robust battery UPS systems (typically with 10–15 minutes of autonomy) combined with diesel generators (with 24–48 hours of fuel storage) are the minimum for any credible SLA. When evaluating a provider, ask specifically about: - Generator start time and load transfer time - On-site fuel storage capacity - Automatic Transfer Switch (ATS) reliability testing frequency - Historical grid failure events and how the facility responded ### Evaluating a Kenyan Colocation SLA When choosing a provider, go beyond the headline uptime number. A practical evaluation checklist includes: 1. **Read the exclusions.** Does the SLA exclude scheduled maintenance? Force majeure events? Third-party carrier failures? 2. **Verify the metrics.** Is uptime measured at the facility level or at your cabinet/rack level? 3. **Check reporting.** Does the provider offer real-time monitoring dashboards and monthly SLA performance reports? 4. **Review the credit structure.** Are the credit percentages meaningful, or are they capped at levels that make claims impractical? 5. **Ask for history.** Request 12 months of actual SLA performance data, not just the contractual targets. ## Marketing Claims vs. Contractual Obligations One of the most important distinctions in the SLA world is the gap between what a provider **markets** and what they **contractually commit to**. A data centre's website might boldly state "99.999% uptime" in large type, but the actual SLA document buried in the contract appendix may guarantee only 99.9%. This is not unique to Kenya, it is a global industry practice. However, it is especially important to watch for in a rapidly growing market like East Africa, where competition is driving aggressive marketing. Here is how to protect yourself: - **Always request the full SLA document before signing.** Do not rely on sales collateral or website claims. - **Identify the legally binding metrics.** The only numbers that matter are the ones in the signed contract with financial penalties attached. - **Look for weasel words.** Phrases like "target," "aim," or "design intent" are not guarantees. - **Cross-reference with the ITIL framework.** The [ITIL service management best practices](https://www.axelos.com/certifications/itil-service-management) provide a useful benchmark for evaluating whether an SLA is comprehensive and professionally structured. ![Data centre rooftop chillers and generators](/images/dc-chillers-roof.webp) ## What to Negotiate in Your SLA If you are a serious colocation customer committing multiple racks or a private cage, you should negotiate several SLA elements: ### Custom Uptime Targets Standard SLAs are designed for the average customer. If your workloads demand higher availability, negotiate for a four-nines or five-nines commitment, but expect to pay a premium for it. ### Reporting Requirements Insist on **monthly SLA performance reports** with detailed metrics covering power availability, cooling performance, network uptime, and ticket response times. Some Kenyan operators provide real-time dashboards through customer portals; negotiate for API access if you want to integrate this data into your own monitoring systems. ### Response Time Guarantees If your team is not physically present at the facility, remote hands and smart hands response times are critical. Negotiate for guaranteed response times with financial penalties if the provider misses them. ### Exit Clauses Ensure your SLA or master services agreement includes a clear **exit clause** that allows you to terminate the contract without penalty if the provider repeatedly fails to meet SLA commitments. A common threshold is three SLA breaches within any 12-month period. ### Maintenance Windows Scheduled maintenance is a common SLA exclusion. Negotiate for maintenance windows that fall outside your peak business hours (for example, 02:00–06:00 EAT on Sundays) and ensure the provider gives adequate advance notice (typically 7 to 14 days). ## SLA Tier Comparison: At a Glance | Feature | Basic (99.9%) | Standard (99.99%) | Premium (99.999%) | |---|---|---|---| | **Annual downtime allowance** | ~8 hrs 46 min | ~52 minutes | ~5 minutes | | **Typical Tier rating** | Tier II | Tier III (N+1) | Tier IV (2N) | | **Power redundancy** | N+1 components | N+1 distribution | 2N independent paths | | **Cooling redundancy** | Single path | Concurrently maintainable | Fault tolerant | | **Service credit (per breach)** | 10% of monthly fee | 10–25% of monthly fee | 25–100% of monthly fee | | **Remote hands response** | 30 minutes | 15 minutes | 10 minutes or less | | **SLA reporting** | Quarterly | Monthly | Real-time + monthly | | **Typical Kenyan providers** | Smaller facilities | Africa Data Centres, IXAfrica, Liquid | Rare, premium suites only | ## Final Thoughts A data centre SLA is your primary contractual safeguard when entrusting your IT infrastructure to a third-party facility. In the Kenyan market, where the data centre industry is maturing rapidly and operators are competing intensely, SLAs are becoming more generous, but also more complex. The key is to look beyond the headline uptime percentage and understand the full picture: what metrics are covered, what exclusions apply, how enforcement works, and whether the contract reflects reality or marketing aspiration. Before signing any colocation agreement in Kenya, invest time in reviewing the SLA thoroughly. Request historical performance data, negotiate where possible, and ensure you have independent monitoring in place. The strength of your SLA today determines the resilience of your business tomorrow. --- ## Frequently Asked Questions ### What does 99.99% uptime actually mean in practice? 99.99% uptime means your service can experience no more than 52.6 minutes of unplanned downtime in an entire year. For most Kenyan businesses running e-commerce or SaaS platforms, this is generally considered an excellent standard. However, it is important to check whether the SLA covers scheduled maintenance windows, which some operators exclude from their calculations. ### Can I negotiate a better SLA with a Kenyan data centre provider? Yes, especially if you are committing significant rack space or a multi-year contract. Larger customers regularly negotiate custom SLA terms around response times, reporting frequency, and service credit percentages. Start by asking for their standard SLA document, then identify the gaps between their offering and your business requirements. ### What happens if a data centre breaches its SLA? Typically, you are entitled to service credits, a percentage refund on your monthly colocation fee for the month in which the breach occurred. Most providers offer 10% to 100% credit depending on the severity and duration of the outage. However, service credits rarely compensate for your actual business losses such as lost revenue, reputational damage, or regulatory penalties. ### Do Kenyan data centres actually meet their published uptime SLAs? The major operators (Africa Data Centres, PAIX, Liquid Intelligent Technologies, and IXAfrica) generally meet or exceed their published SLAs, particularly for power and cooling. However, Kenya's national grid (Kenya Power) still experiences occasional outages, which is why robust on-site power infrastructure is critical. Always ask prospective providers for their historical SLA performance reports before signing. ### Is a 99.999% uptime SLA realistic for a Kenyan data centre? Genuine 99.999% ("five nines") uptime is extremely demanding, it allows only 5.26 minutes of downtime per year. Very few facilities globally achieve this, and it typically requires a Tier IV design with 2N redundancy. In Kenya, some operators may market "five nines" but the contractual SLA may be lower. Always distinguish between marketing language and the legally binding commitment in your contract. ## The Software Data Centre Technicians Actually Use URL: https://data-centers-254.vercel.app/articles/data-centre-software-technicians-use Cluster: Careers Published: 2026-09-16 Updated: 2026-09-16 Summary: DCIM, BMS, monitoring, ticketing, NetBox and more: the software behind every Kenyan data centre job, what each tool does, and free ways to practise before you are hired. Ask a Kenyan data centre technician what they actually did last week and the honest answer is rarely "installed servers". It is closer to: closed forty tickets, checked a hundred dashboards, updated the rack map twice, chased one aircon fault through the building management system, and documented all of it. Data centre work looks physical from the outside, but inside it runs on software. Facilities are managed, monitored, documented and billed through tools, and the people who master those tools get promoted by them. This article is the software stack, category by category: what each tool does, which products dominate, and, because we write for people trying to break in, which free versions you can install tonight. It pairs with our guides to [certifications](/articles/data-centre-certifications-kenya) and [getting hired](/articles/how-to-get-job-kenyan-data-centre). ![An engineer working on a laptop beside server racks](/images/dc-woman-engineer-laptop.webp) ## DCIM: the living map of the facility Data centre infrastructure management software, DCIM for short, is the category that defines modern operations. A DCIM platform holds a digital model of the facility: every rack position, every device, every power feed and network port, every cable run, together with the capacity each one has left. Raise a new cabinet request and the DCIM shows which rack has the power, the space and the cooling headroom to take it, before anyone walks the floor. The big names Kenyan facilities deploy are Sunbird, Nlyte and Schneider Electric's EcoStruxure IT, often chosen because the facility's electrical and cooling vendor is Schneider already (Sunbird; Schneider Electric). What DCIM replaces is the honest enemy of every operations team: the spreadsheet. A rack inventory in a spreadsheet is out of date the first time someone moves a cable and forgets to write it down; a DCIM that is properly maintained becomes the source everyone trusts, and auditors love it because it proves discipline. **Practise it free:** NetBox is the open source standard for exactly this kind of infrastructure documentation, racks, devices, cables, IP addresses, and it runs on a modest machine (NetBox Labs). It is not full DCIM, no live power telemetry, but the mental model it teaches, that a facility is a database first and a room second, is the one that transfers. ## The building management system: where the facility talks A data centre's electrical and mechanical plant, generators, UPS systems, chillers, air handlers, pumps, fire panels, reports through a building management system, the BMS. Technicians use it to watch temperatures and humidity per aisle, see power readings from the switchgear, receive alarms before humans feel them, and build the trends that justify maintenance. When a chiller starts cycling oddly, the BMS history is the evidence that says when it began. In most facilities the BMS is a vendor platform such as Schneider or Siemens, sitting closer to the mechanical team than the IT team, and the boundary between the two is where good technicians add value. A network engineer who can read a BMS trend and say "your intake temperature rose two degrees after the cabling change" is worth more than either discipline alone. **Practise it free:** You cannot run a real chiller at home, but the concepts, sensors, thresholds, alarms, trend logs, are the same ones Zabbix and Grafana teach, which is why the monitoring practice below doubles as BMS preparation. ![A technician studying technical material in a training room](/images/dc-careers-tech.webp) ## Monitoring: the dashboards that never sleep Monitoring software watches everything with an IP address and everything without one: is the switch reachable, is the UPS on bypass, is the humidity inside spec, is the website answering. Kenyan facilities typically run a stack rather than one product. A network monitoring platform such as Zabbix or Nagios polls devices and fires alarms; Grafana with Prometheus collects and graphs time-series data for humans to reason about; and log systems keep the receipts of what happened when. The skill that employers actually test is not clicking around a dashboard, it is interpreting one. Why did this graph blip? What does a UPS transfer event look like, and how long is it acceptable? A candidate who can read a latency spike and reason aloud about upstream versus internal causes sounds like a hire, whatever tools they learned on. **Practise it free:** Zabbix is open source and runs on a single Linux virtual machine (Zabbix); point it at a router, a Raspberry Pi, anything with SNMP, and you have real alerts and real graphs. Add Grafana in front of it and your home lab starts looking professionally boring, which is the goal. ## Ticketing and asset management: the paperwork that is the job Every task in a disciplined facility, every rack move, every filter change, every customer request, exists as a ticket. Ticketing platforms, ServiceNow, Jira Service Management, Freshservice, or a homegrown CMMS for the maintenance side, carry the request, the approval, the change window, the execution record and the closure note. Asset management systems track every physical thing: which serial numbers the facility owns, where they are, their warranty and their lifecycle. Young technicians often experience ticketing as bureaucracy. Senior people know it is the memory of the organisation: the ticket history is how a facility proves what changed when, spots repeat offenders among its equipment, and answers the auditor's question with evidence instead of memory. A well-written ticket, problem, action, result, is a professional document, and hiring managers read them that way. **Practise it free:** Snipe-IT is a solid open source asset tracker you can run for a home lab's inventory, and your own change log in a wiki or Git repository teaches the same discipline ticketing enforces. ## The engineer's own bench: operating systems and networks Underneath everything sits the stack the tenants actually run, and facility engineers touch it constantly: Linux on most servers and appliances, Windows in enterprise corners, virtualisation platforms that turn one server into many, VMware ESXi commercially and Proxmox VE in the open source world, and the network tooling of the trade. Wireshark for packet-level answers, iperf for throughput tests, SSH for everything, and increasingly Ansible-style automation so a change applied to forty switches is a template, not forty mistakes. ![An engineer checking cabling at the back of server racks](/images/dc-woman-engineer-racks.webp) **Practise it free:** Proxmox VE is free and turns any half-decent computer into a virtualisation host; install it, break it, fix it. Wireshark and Ansible are free outright. Linux fundamentals cost nothing but evenings and are the single highest-return investment a Kenyan candidate can make, because every monitoring tool, DCIM appliance and BMS server in a real facility is a Linux box wearing a costume. ## How the stack fits a Kenyan career Map the categories to the roles and the strategy becomes obvious. Facility technicians live in the BMS, ticketing and DCIM; NOC engineers live in monitoring and the network bench; systems and cloud engineers live in the virtualisation and automation layers; and the managers who advance furthest are fluent in all of it, because capacity planning is just the DCIM database with a spreadsheet's intention. The Kenyan route in is the same as the global one, with a local twist: the free tools above give you genuine practice without a job or a lab budget, the certifications from our [Kenya guide](/articles/data-centre-certifications-kenya) get you past the HR filter, and the interview differentiator is the thing this article is really about, being able to say, specifically, which software you have run, what broke, and how the tools helped you fix it. Start with Linux and Zabbix this week. Add NetBox the next. By the time a facility calls you in, you will be explaining dashboards from experience, not from YouTube. ## Data Centre Tier Ratings: Tier I to IV Explained URL: https://data-centers-254.vercel.app/articles/data-centre-tier-ratings-explained Cluster: Data Centres Published: 2026-08-26 Updated: 2026-09-16 Summary: Tier I to IV ratings define data centre reliability from basic to fault-tolerant. Most Kenyan facilities target Tier III. What each tier means. When you hear that a data centre is "Tier III certified," what does that actually mean for the servers inside it? The tier rating system, developed and administered by the Uptime Institute, is the global standard for classifying data centre reliability. It defines four levels of infrastructure redundancy, each guaranteeing a specific amount of annual uptime. Understanding these tiers is essential for anyone evaluating where to host their digital infrastructure, and for anyone considering [a career in data centres](/careers). The tier system answers one fundamental question: **if something breaks, what happens to the servers?** The answer ranges from "everything goes down" at Tier I to "nothing happens" at Tier IV. The difference between these extremes is measured in redundant power supplies, backup cooling systems, and the physical paths that connect them. ![Data centre power distribution and backup systems](/images/dc-power-systems.webp) ## The Four Tiers: What They Guarantee ### Tier I, Basic Capacity (99.671% uptime) A Tier I data centre has a single path for power and cooling, with no redundant components. If the power fails, the servers go down. If the cooling fails, the servers overheat and shut down. There is no backup for anything. Tier I guarantees 99.671% availability, which translates to approximately 29 hours of downtime per year. That sounds high, but most of this "availability" assumes the facility operates perfectly. In practice, any planned maintenance, component failure, or power blip causes complete downtime. Tier I facilities are suitable for development environments, testing, or non-critical workloads. They are entirely inadequate for production systems that need to be available 24/7. You will find very few, if any, commercial Tier I data centres in Kenya, the market demands higher reliability. ### Tier II, Redundant Components (99.741% uptime) Tier II adds redundant components (typically N+1) for power and cooling. This means there is at least one backup for every critical component. If a UPS unit fails, a second one takes over. If a cooling unit fails, a backup unit activates. However, Tier II still has a **single distribution path**. All power and cooling travels through a single set of cables, switches, and panels. This means that while the components have backups, the infrastructure connecting them does not. If a distribution panel fails, or if maintenance requires shutting down the power path, the facility goes down. Tier II guarantees 99.741% availability, approximately 22.6 hours of downtime per year. This is an improvement over Tier I, but the single distribution path remains a significant vulnerability. Planned maintenance still causes downtime because there is no alternative path to keep servers running while work is performed. ### Tier III, Concurrently Maintainable (99.982% uptime) Tier III is the standard that most commercial data centres target globally, and it is the standard that most Nairobi facilities aspire to. The key innovation at Tier III is **multiple, independent distribution paths**. Power and cooling are delivered through two completely separate systems, so any single component (or even an entire distribution path) can be taken offline for maintenance without affecting the servers. In a Tier III facility, you can replace a UPS unit, swap out a cooling module, or perform maintenance on a power distribution panel while the servers keep running normally. This "concurrent maintainability" is what enterprises require: the ability to maintain infrastructure without scheduled downtime. Tier III guarantees 99.982% availability, no more than 1.6 hours of downtime per year. This is the minimum requirement for most enterprise contracts, banking regulations, and cloud service level agreements. In Kenya, iXAfrica NBOX1, Africa Data Centres' Nairobi facilities, and several other commercial operators design to this standard. The certification history starts earlier than most people think: East Africa Data Centre, now operated by Africa Data Centres, became the first facility in central and eastern Africa to earn Uptime Tier III Design certification in May 2017, and Raxio applies Uptime Tier III certification across its facilities as a group-wide policy. ![Structured cabling connecting server racks](/images/server-rack-patch-cabling.webp) ### Tier IV, Fault Tolerant (99.995% uptime) Tier IV is the highest standard. It adds **fault tolerance** on top of concurrent maintainability. This means the facility can withstand any single failure (component, distribution path, or system) without any impact whatever on the servers. Not even a momentary blip. Achieving Tier IV requires 2N or greater redundancy: two completely independent, fully-sized systems for power and cooling, with automatic failover that operates within milliseconds. If the primary power path fails, the secondary path takes over so quickly that the servers never notice. Tier IV guarantees 99.995% availability, no more than 26 minutes of downtime per year. This level of reliability is required for the most critical applications: financial trading platforms, military systems, healthcare life-support systems, and similar workloads where any downtime has immediate, serious consequences. Very few facilities in Africa have achieved Tier IV certification. The cost premium over Tier III is substantial, typically 25-40% higher construction cost and 15-25% higher operating cost. For most Kenyan use cases, Tier III provides more than adequate reliability. ## Does Kenya Have a Tier IV Data Centre? As of September 2026, no data centre in Kenya has announced Uptime Institute Tier IV certification, and DC254 has found no Kenyan facility at Tier IV on the Uptime Institute's issued-awards register. The strongest independently verified reliability claims in the Kenyan market sit at Tier III: East Africa Data Centre (now operated by Africa Data Centres) earned the first Tier III Design certification in central and eastern Africa in May 2017, Raxio applies Uptime Tier III certification across its facilities, and iXAfrica's NBOX1 campus is designed to Tier III standards. That is not a gap Kenya needs to apologise for. Tier IV exists for workloads where a momentary blip is unacceptable, such as high-frequency trading platforms or hospital life-support systems, and it carries the cost premium described above. Kenya's rational benchmark is Tier III reliability paired with the country's green, low-cost geothermal grid, which is exactly the combination most local operators sell. If your workload genuinely needs fault tolerance, the practical routes are software-level redundancy spread across two independent Tier III facilities (the approach hyperscalers use globally), or waiting for a verified Tier IV announcement. Whenever any operator claims a tier, ask for the Uptime certificate number and check it against the [official register](https://uptimeinstitute.com/resources/research-and-reports/tier-certification-list) rather than the brochure, and see [our Nairobi colocation buyer's guide](/articles/colocation-data-centres-nairobi-buyers-guide) for the questions that actually matter in a lease. ## N+1, 2N, and 2(N+1): The Redundancy Math The redundancy notation describes how many backup components a system has: - **N** is the number of components needed to operate at full capacity. If your servers need 4 UPS units, N = 4. - **N+1** means N components plus one extra backup. If one of the 4 UPS units fails, the 5th takes over. Any single failure is covered. - **2N** means two completely independent sets of N components. You install 8 UPS units in two independent groups of 4. Either group alone can support the full load. This is the minimum for Tier IV. - **2(N+1)** goes even further: two independent sets, each with its own backup. This provides redundancy against multiple simultaneous failures. For power, the redundancy applies to every stage: utility feeds, generators, UPS systems, power distribution units, and the cables connecting them all. For cooling, it applies to chillers, pumps, air handling units, and the piping that carries the coolant. ![Data centre cooling and power distribution equipment](/images/dc-cooling-crac-2.webp) ## Beyond the Tier: What the Rating Does Not Cover The Uptime Institute tier rating is specifically about infrastructure reliability, power, cooling, and the physical paths that connect them. It does not cover: **Security.** A Tier IV facility can have terrible physical security, and a Tier I facility can have excellent security. Physical security (biometrics, CCTV, mantraps, guards) is evaluated separately. **Network connectivity.** The tier rating says nothing about the quality or diversity of internet connectivity. A Tier III facility with a single fibre connection from one ISP is less reliable in practice than a Tier II facility with connections from three different providers. **Software and operations.** A perfectly built Tier IV facility run by an incompetent team will have more outages than a Tier II facility with excellent operations and monitoring. **Sustainability.** The tier system does not measure energy efficiency, renewable energy use, or carbon footprint. A Tier IV facility running on diesel generators has a larger environmental impact than a Tier II facility on geothermal power. This matters for Kenya because the country's competitive advantage in data centres is not just about reliability, it is about the combination of reliability with renewable, low-cost geothermal energy. A facility that achieves Tier III reliability while maintaining a PUE of 1.3 on geothermal power is more attractive to many customers than a Tier IV facility running on diesel. ![Data centre rooftop chillers and generators](/images/dc-chillers-roof.webp) ## Certification vs. Design: The Trust Gap Any operator can claim their facility is "Tier III" or "designed to Tier III standards," but only facilities that have been formally certified by the Uptime Institute have been independently verified. The certification process involves a detailed audit of the facility's design documents, followed by an on-site inspection of the physical infrastructure. Certification is expensive, typically USD 50,000 to 200,000 depending on the facility size and the level of certification sought. For operators in emerging markets like Kenya, this cost can be a significant barrier. As a result, many facilities are "designed to Tier III" meaning they have the right architecture on paper, but they have not been formally certified. For enterprises evaluating data centre providers, the question to ask is not just "what tier are you?" but "are you Uptime Institute certified?" The difference between a claim and a certification is the difference between marketing and engineering. ![Data centre security systems](/images/dc-biometric-access-2.webp) ## What This Means for Kenya's Data Centre Industry Kenya's data centre market is at a stage where reliability standards matter more than ever. As international cloud providers, AI operators, and enterprise customers evaluate Kenya as a potential regional hub, they will look for facilities that meet or exceed Tier III standards, preferably with formal certification. The [Kenya data centre directory](/directory) shows which facilities publicly state their tier ratings. As the market matures, expect tier certification to become a standard requirement in enterprise and government procurement, much as it has in South Africa, Nigeria, and Egypt. The operators that invest in certification will have a significant competitive advantage over those that do not. ## Digital Realty's NBO2: The 'Gateway' Claim, Tested URL: https://data-centers-254.vercel.app/articles/digital-realty-nbo2-launch-nairobi Cluster: Operators Published: 2026-09-08 Updated: 2026-09-28 Summary: Digital Realty formally launched NBO2 in Nairobi on 7 September 2026. What the launch delivers: verified capacity, interconnection, ramping reality. On 7 September 2026, Digital Realty formally launched NBO2, the second facility of the iColo campus it inherited when it acquired the Kenyan carrier-neutral operator. The event drew the state's digital establishment: ICT and Digital Economy Principal Secretary John Tanui framed the launch as part of the government's push to make Kenya the region's digital hub, and Digital Realty's own announcement carried the year's most borrowed phrase, Nairobi as East Africa's "digital gateway". The launch matters. Our directory has carried NBO2 since our September re-verification, and it fills a genuine gap: Kenya's carrier-neutral capacity has been concentrated in a handful of buildings, and a second large neutral site in Karen adds real supply to a market growing at roughly 30% a year. But a launch press release is a starting point for verification, not the end of it, so here is what the evidence supports today, and what the "gateway" language still has to earn. ![Officials mark the formal launch of NBO2 under the Digital Realty arch in Nairobi](/images/nbo2-launch-ribbon-cutting.webp) ## What actually launched The physical facts first. NBO2 sits on Bogani East Road in Karen, near its sibling NBO1, the facility iColo launched in September 2019 as "the first truly carrier-neutral data centre in Nairobi". The building is specified at 3,600 m² with 2N power, N+2 cooling, ISO 27001 and PCI-DSS claims per the Digital Realty listing. Developing Telecoms reported in August 2024 that NBO2 carries a **6.5 MW design IT load**, expected online in Q3 2025, within a master plan of three data centres plus a captive substation consuming over 20 MW at full build. That master plan (three buildings, 18,000 m² on the Karen side, 13 MW and 1,800 racks at the Mombasa campus) is the strategic story Digital Realty is buying into: iColo already runs the largest facility count in Kenya, with four buildings across Nairobi and Mombasa. ![Digital Realty company logo](/images/digital-realty-logo.webp) What was *not* published at launch, and what we flagged on [NBO2's directory entry](/directory/icolo-nbo2): the commissioned (as opposed to design) IT load, the number of halls fitted out, and the commercial live date for general colocation. None of this is criticism, it is the standard gap between a ribbon-cutting and a ramp. The discipline is to price the facility on what is commissioned, not on the master plan. Days after the ribbon-cutting, follow-on coverage filled in the government's side of the ledger. Powers of Africa reports that Principal Secretary Tanui used the inauguration to encourage Digital Realty to keep expanding toward a **20 MW** campus (a target that lines up with the master plan's captive substation, already sized for more than 20 MW at full build) and pledged that "work will continue on the policy front to position Kenya as an attractive investment destination." The launch coverage also spells out the campus-level pitch: access to more than 100 networks, two internet exchange points and a satellite teleport, with customers able to distribute workloads between NBO1 and NBO2 as a built-in business-continuity pair. Read that framing carefully, though, those connectivity figures describe what the wider campus and its ecosystem offer, not NBO2's own building-level register, which is the number that has to climb for the gateway claim to cash out. ## The gateway claim, and the one number that decides it "Digital gateway" is a claim about traffic, not buildings. The single best public proxy for whether a building is becoming a gateway is its **registered interconnection** (the number of networks present) and this is where NBO2's evidence is thinnest. The PeeringDB record for NBO2 (last updated September 2025) showed **two networks and a single internet exchange**, against NBO1's 62 networks and four exchanges. Our own [data methodology](/methodology) treats PeeringDB as a cross-check rather than ground truth, and the register is community-maintained, new buildings are often under-registered. But the direction is unambiguous: NBO2 is a real, formally launched, still-ramping building. For the gateway claim to cash out, three things have to happen, and each is observable: 1. **Networks move in.** NBO1 took years to accumulate its density, and it is [interconnection and peering](/articles/data-centre-interconnection-peering-kenya), not floor space, that turns a building into a gateway. Watch whether NBO2's registered count climbs from two toward double digits within its first year, and whether the exchanges on site grow beyond one. 2. **Content and cloud land.** The 27 January 2026 announcement that iXAfrica will host Oracle Cloud Infrastructure's Nairobi region shows what a gateway transaction looks like: a hyperscaler committing to local capacity. If Digital Realty lands a comparable cloud on-ramp at NBO2, the claim strengthens materially. 3. **Cable economics show up in pricing.** A gateway is supposed to lower the cost of reaching the world. If cross-connect and IP transit pricing at Karen converges toward (or beats) the Mombasa Road corridor, that is the market voting. ![Server racks inside the new NBO2 facility](/images/dc-servers-racks.webp) ## Why Digital Realty bought iColo in the first place The launch is also a window into the consolidation logic reshaping African digital infrastructure. Digital Realty (one of the world's largest data centre companies) did not need another building; it needed **the carrier-neutral franchise in East Africa's best-connected market**, and iColo was exactly that: four facilities, the densest interconnection point on the Kenyan coast (MBA1 in Miritini, 94 networks), and a decade of neutrality credentials. NBO1 and NBO2 gave it the Nairobi footprint; MBA1 and MBA2 gave it the cable-landing play. For Kenyan buyers, the consolidation cuts both ways. The upside: capital depth, global standards playbooks, and the kind of master-plan patience (captive substations, three-building campuses) that local capital struggles to fund. The caveat: every Kenyan carrier-neutral market now has a Digital Realty price-maker in it. Buyers negotiating colocation in Karen should do what our [buyer's guide](/articles/colocation-data-centre-kenya) always recommends, benchmark against ADC NBO1, iXAfrica and PAIX, and ask for as-built capacity in the specific hall, not the campus brochure figure. ## What it means for the market this quarter Three practical takeaways. First, Kenya's operating map now has 27 verified entries in [our directory](/directory), but only eight genuinely carrier-neutral buildings, so every new neutral rack is absorbed quickly; NBO2's ramp is worth watching monthly. Second, the competitive response is already in the market: iXAfrica's 22.5 MW Mombasa Road campus expansion, the Tilisi land commitment, and Africa Data Centres' Sameer expansion are all racing the same demand curve, which means pricing power is shifting to buyers in the near term, whatever the press releases imply. Third, for the government's hub ambitions, the binding constraint remains power delivery and interconnection density, not concrete. [Our market-in-numbers reference](/articles/kenya-data-centre-market-numbers) tracks all three, and the [NBO2 entry](/directory/icolo-nbo2) will be re-verified the moment its interconnection record moves. ## DigitalOcean's AI Cloud and Kenya's Hosting Market URL: https://data-centers-254.vercel.app/articles/digitalocean-ai-cloud-kenya-hosting Cluster: AI & Infrastructure Published: 2026-09-08 Updated: 2026-09-26 Summary: DigitalOcean's AI bet shows in its numbers: $894M in reservations, AI ARR up 212%, 155 MW committed. What the megawatt race means for Kenyan hosts. A line in DigitalOcean's second-quarter report deserves more attention in Nairobi than it will probably get there: the company **locked in another 20 megawatts of data centre capacity for 2027 and 2028, bringing total committed capacity to about 155 MW**. DigitalOcean is not a hyperscaler, it is the cloud African startups actually start on, the $6-a-month droplet provider, the company whose brand equity in Lagos, Nairobi and Kampala was built on being small and simple. When a company like that starts transacting in megawatts, the unit of account for the whole hosting industry has changed. The quarter's numbers (reported 4 August, analysed by Yahoo Finance on 7 September) tell a clean story. Revenue climbed 29% year over year to $281 million. Remaining performance obligations (contracted revenue not yet recognised) swelled from $71 million a year ago to **$894 million**, which is the financial signature of customers signing multi-year deals instead of monthly ones. AI customer ARR grew **212% to $234 million**, and 85% of that AI revenue now comes from inference and core cloud workloads rather than raw bare metal. The company added a record $93 million of incremental ARR, signed its first nine-figure annual commitments, and stretched its weighted average contract life from 1.6 years to more than 3. ![GPU servers powering the AI cloud build-out](/images/ai-gpu-servers.webp) ## The inference shift, explained plainly The strategically interesting number is not the 212%, it is the **85% from inference and core cloud**. Training an AI model is a concentrated industrial event: a few giant sites with grid-scale power, mostly in the US and China. Inference is the opposite: every user query, every API call, every embedded AI feature in an app needs a server close enough to answer quickly. Inference demand therefore spreads, across regions, across cities, across whatever power-secure real estate exists. That is why the megawatt commitments matter. DigitalOcean is contracting 155 MW of future capacity precisely because inference customers sign three-year commitments and expect the capacity to exist when they arrive. Multiply this pattern across every mid-size cloud (and the large ones are committing ten times more) and you get the infrastructure reality underneath the AI hype: **the world's AI build-out is, physically, a global race for power-secured data centre capacity**. Kenya is part of that race whether or not the hyperscalers have noticed yet; iXAfrica's Oracle Cloud Infrastructure Nairobi region announcement in January 2026 was the first local transaction of this exact type. ## What it means for Kenya, in three layers **For developers and startups**, the direct read: pricing and capacity on the global clouds stays competitive for now. DigitalOcean's margin pressure (operating income fell 18% even as revenue accelerated) is the market's way of saying AI capacity is being bought ahead of demand. That is good for buyers this year. The risk sits further out: if AI-driven demand keeps compounding, the entry prices that made DO the startup default will creep upward, and the free-credit economics that seeded the African developer ecosystem will get tighter. **For Kenyan hosting businesses**, the harder truth: none of DigitalOcean's 155 MW is coming to Nairobi on current evidence. Its capacity hunt plays out in US and European markets with proven grid supply. Kenya's cloud-on-ramp story is being written by others, iXAfrica with OCI, the PAIX and ADC interconnection layers, Digital Realty's NBO2 launch. The local opportunity is not to out-cloud DigitalOcean; it is to be the **inference edge** the global clouds need: cache the models, serve the users, keep latency local and shillings-denominated. That is a colocation and interconnection business, which is why our [directory](/directory) tracks carrier neutrality and network counts more closely than marketing claims. **For policymakers**, the uncomfortable arithmetic. A single mid-size AI cloud commits more megawatts in one quarter (20 MW) than Kenya's entire verified operating data centre capacity ([~28 MW across 27 verified facilities](/articles/kenya-data-centre-market-numbers)). The global AI build-out is a power-allocation contest, and the Microsoft–G42 project's well-documented stall on grid delivery shows what happens when a 100 MW-class request meets a 3,192 MW grid serving a 2,316 MW peak. If Kenya wants a seat in the inference economy, the binding policy question is not "how do we attract AI companies", it is how fast dispatchable generation and transmission can be built behind the geothermal belt. ![Power infrastructure behind cloud capacity](/images/dc-power-systems.webp) ## The watch-list from here Three things worth tracking this year. One: whether DigitalOcean's AI ARR growth survives the margin squeeze, the bulls' case is that nine-figure inference contracts mature into profitability; the bears' case is that it is renting GPUs at negative margin to buy growth. Two: whether inference capacity starts being committed *on the continent*, any announcement of a hyperscaler or mid-size cloud region in East Africa is a signal our [AI infrastructure coverage](/articles/ai-data-centres-east-africa) will chase. Three: the price of entry-level cloud. The moment African startups' default droplet gets more expensive without getting better, the local-hosting value proposition (data residency, M-Pesa billing, shilling pricing, local support) becomes the strongest it has ever been. The megawatt race is global. The customer is local. ## Disaster Recovery for Kenyan Businesses: A Playbook URL: https://data-centers-254.vercel.app/articles/disaster-recovery-business-continuity-kenya Cluster: Infrastructure Published: 2026-09-16 Updated: 2026-09-16 Summary: Cable cuts, grid trips and floods are normal Kenyan events. A practical disaster recovery and business continuity playbook built for Kenyan conditions, not textbook ones. Every Kenyan business has a disaster recovery story it did not choose. The May 2024 undersea cable cuts slowed or severed East Africa's connectivity for weeks and forced businesses to activate continuity plans as a routine rather than an exception (Citizen Digital, 14 May 2024; Internet Society, 23 July 2024). The July 2023 DDoS attacks took government portals offline at national scale. Fibre cuts, grid trips and floods arrive with the seasons. None of this is unusual, and that is precisely the point: in Kenya, disruption is not a tail risk, it is a scheduled guest, and the businesses that survive it treat continuity as engineering rather than paperwork. This playbook is built for those conditions. It covers the two numbers that decide everything, the failure modes Kenya actually produces, the design patterns that answer them, and the testing discipline that separates a plan from a wish. ![UPS units and switchgear in a data centre power room](/images/ups-switchgear-power-room.webp) ## Start with two numbers, not a product Everything in DR pricing and design flows from two questions the business must answer honestly before talking to any vendor. **RTO: how long can you be down?** The recovery time objective is the clock. A market payment platform may have an RTO measured in minutes; an internal HR system may tolerate days. Write the number down per system, because "all systems are critical" is the answer that makes every solution unaffordable. **RPO: how much data can you lose?** The recovery point objective is measured backwards from failure. An RPO of zero means synchronous replication, expensive and distance-limited. An RPO of 24 hours means nightly backups, cheap and fragile. Most Kenyan SMEs discover they have an implicit RPO of "whenever the backup last ran", which is a hope, not a number. The cost curve is brutal and clarifying: recovering in minutes costs orders of magnitude more than recovering in hours. The professional move is tiering: payment systems get near-zero RTO and RPO, customer databases get an hour, email gets a day. Nobody can afford uniform perfection, and nobody needs it. ## The Kenyan failure catalogue Design for the failures that actually happen here, and they cluster into four families. **Connectivity.** The 2024 cable cuts were the national lesson: multiple undersea systems were damaged and East African connectivity degraded for weeks (BBC, 14 May 2024). On land, fibre cuts from roadworks, theft and weather are chronic. The design answer is diversity that is genuinely diverse: two providers over different physical routes, ideally with a non-terrestrial backup such as satellite for management traffic, because two providers sharing one fibre duct share one backhoe. **Power.** Grid trips and load management are a Kenyan normal. Facilities answer with the classic chain, UPS ride-through, generators, fuel contracts, and businesses answer by hosting in facilities whose redundancy they can audit rather than running their own server rooms. The grid is not the enemy; single-path dependence is. **Facility and environment.** Fire, flood, cooling failure, building access loss. This is what Tier ratings and SLAs price in: concurrent maintainability means the facility can lose a component and keep running while it is fixed. Our [SLA article](/articles/data-centre-sla-uptime-guarantees) decodes what the nines actually buy. **Cyber.** The July 2023 national attacks proved that a flood of junk traffic can degrade services that millions depend on, and ransomware remains the scenario that turns backups into the whole company. Continuity planning and security are the same discipline wearing different badges; the DR site that was never tested is exactly the one ransomware will find unprepared. ![Map of submarine cables serving the East African coast](/images/submarine-cables-map.webp) ## The four patterns that work in Kenya **Pattern one: backup, done honestly.** Nightly copies, offsite, immutable where possible, restored monthly on a schedule. This covers RPOs of a day and RTOs of days. It is the floor, not the plan, but it is also the pattern that saved every business whose ransomware story had a happy ending. **Pattern two: warm DR in a colocation facility.** Replicate the critical systems to a second site, in a facility with Tier III power and a different network footprint, and keep them warm: data current, capacity reserved, failover rehearsed. For a Kenyan enterprise this is the sweet spot: Nairobi colocation with a DR footprint in Mombasa, or two providers in separate power and network domains, converts the cable-cut and grid-trip scenarios from crisis to inconvenience. **Pattern three: cloud as the disaster site.** Replicate to a cloud region (today, that means outside Kenya for most providers) and accept the latency and data-residency conversation that follows. The cloud pattern shines for elastic recovery: if the DR load only runs during failures, you pay for it only when it runs. Our [cloud comparison](/articles/cloud-services-kenya-compared) covers the trade-offs. **Pattern four: multi-site production.** The bank pattern: run actively in two locations, each able to carry the other, with data synchronously or near-synchronously replicated. Maximum cost, minimum drama. It is the pattern the CBK's continuity expectations effectively push financial institutions toward, and the reason Nairobi-Mombasa and Nairobi-Kampala pairs appear in regional architecture diagrams. ![Wind turbines and transmission infrastructure in Kenya](/images/kenya-wind-turbines-4.webp) ## The testing discipline An untested DR plan is a document that lies to you. The professional cadence in Kenyan practice looks like this. **Monthly:** restore something from backup, rotating systems, and measure how long it took against the RTO. **Quarterly:** a tabletop walkthrough of one scenario, cable cut, ransomware, building loss, with the actual people who would respond. **Annually:** a live failover of a real system to the DR site, with users on it, during business hours, accepting the risk that the test itself may cause a small outage, because that risk is the cheapest tuition you will ever pay. **After every real incident:** a written review that updates the plan, because every disaster is a free audit if you let it be. Add the human layer that no diagram shows: printed call trees that work when the phone network is degraded, out-of-band access to the DR site that does not ride the production network, and wallets with fuel money, because continuity is sometimes a person with a diesel receipt. ## The compliance tail Two Kenyan legal hooks attach to all of this. Financial institutions carry the Central Bank of Kenya's business continuity expectations, in place since the January 2008 prudential guideline and extended through ICT risk guidance. And the Data Protection Act follows you into disasters: a flood that soaks a server room, or a fire that scatters records, can be a notifiable personal data breach, with the same duties as a cyber intrusion (Bowmans, 23 January 2024). The DR plan should therefore include a notification step, because the regulator's clock starts at discovery, not at recovery. The final word belongs to the numbers. Kenya's internet will be cut again, the grid will trip again, and the cables will break again in somebody's anchorage. The businesses that meet those events calmly will be the ones that picked two honest numbers, built to them, and rehearsed until failover was boring. Boring, in disaster recovery, is the whole prize. ## East Africa's Data Centres Compared: The 2026 Guide URL: https://data-centers-254.vercel.app/articles/east-africa-data-centres-compared Cluster: Market Analysis Published: 2026-09-16 Updated: 2026-09-16 Summary: Kenya, Uganda, Tanzania, Rwanda and Ethiopia compared: facilities, operators, power, cables and policy. Where East Africa's data capacity actually sits today. Ask where East Africa's internet actually lives and the answer is a map with one heavy dot. Kenya hosts the region's deepest data centre market, its cable landings and its biggest exchange point. But the story of 2026 is the other four capitals: Kampala, Dar es Salaam, Kigali and Addis Ababa, each getting their first carrier-neutral facilities, each deciding how much of their digital future to host at home versus transit through Nairobi. This article compares the five East African markets on the things that decide where workloads land: facilities and operators, power, connectivity, policy and demand. It is the infrastructure-side companion to our [regulation comparison](/articles/africa-data-centre-regulation-compared), and it uses named, sourced facts rather than market-report averages. ![Map of Africa showing data centre capacity by country](/images/africa-dc-map.webp) ## Kenya: the hub with the receipts Kenya's lead is structural, and it rests on three legs. First, connectivity: seven live submarine cable systems land on the Kenyan coast, with more in the pipeline, making Mombasa one of Africa's great cable towns and Nairobi the natural aggregation point for the region's traffic. Second, power: about 45 percent of Kenyan electricity is geothermal, renewable baseload that neither drought nor fuel prices interrupt, the resource that anchors every hyperscale conversation in the country. Third, ecosystem depth: the Kenya Internet Exchange Point, two decades of carrier competition, and an enterprise economy that actually buys cloud, colocation and managed services. The facility landscape reflects that depth. Nairobi hosts Africa Data Centres' NBO1 and the newer NBO2 campus launched with Digital Realty, Raxio's Nairobi facility, Wingu's campus, ixafrica, carrier-owned capacity at Safaricom and Liquid, and a corridor of enterprise server rooms that anchor the market. Mombasa adds coastal capacity led by iColo's MBA1, sitting where the cables come ashore. Our [Kenya directory](/directory) tracks 27 facilities, the region's largest single-country list, and our [market numbers](/articles/kenya-data-centre-market-numbers) article carries the detail. The honest weaknesses: grid capacity is finite, as the suspended Microsoft and G42 project proved in public, and landlocked regional customers still route much of their traffic through Nairobi because their home markets cannot yet serve them. ## Uganda: first mover among the neighbours Kampala earned the region's first outside-Nairobi Tier III badge early. Raxio Data Centre Uganda launched in 2020 as the country's first certified, carrier-neutral Tier III facility, with room for up to 400 racks (Built in Africa). Wingu operates in Kampala as well, and the country's enterprise market, banks, telecoms, government systems, gives the facilities a real customer base rather than speculative demand. Uganda's constraint is power economics. Hydro generation is cheap when the lakes are full and strained when they are not, and industrial tariffs and reliability sit below Kenya's geothermal baseload in investor conversations. Uganda's internet path also leans on Kenyan transit: the undersea capacity its traffic rides lands in Mombasa, which keeps Kampala's costs one hop removed from Kenya's ecosystem. For regional architects, Kampala is now a legitimate site for in-country hosting and DR, especially for regulated Ugandan workloads that prefer or require local residency. ## Tanzania: the big market finally getting carrier-neutral capacity Dar es Salaam has long been East Africa's second-biggest economy with the least independent data infrastructure, its traffic historically riding through Kenya or South Africa. That is changing on two fronts. Raxio Tanzania, a 6 megawatt facility in Dar es Salaam, is set to launch in 2026 as the country's first carrier-neutral Tier III data centre (Raxio Group), the kind of facility that lets Tanzanian banks and telcos interconnect without leaving the country. Wingu, which already runs a Dar facility as part of its regional network (Datacenters.com), adds competition and choice. Tanzania's advantages: gas-fired generation gives its grid a fuel-based stability, its own submarine cable landings mean it is not dependent on Kenyan routes for international bandwidth, and a large domestic market with data localisation instincts. Its frictions: historically restrictive policy instincts and a slower enterprise cloud adoption curve. Watch the Raxio launch; carrier-neutral capacity is the catalyst that turned Nairobi's market competitive, and Dar is running the same playbook five years later. ![A submarine cable landing station building on the Kenyan coast](/images/mombasa-cable-landing-2.webp) ## Rwanda: small market, deliberate strategy Kigali is the region's policy showcase. The government made digital infrastructure a national priority early, and Raxio Data Centre Rwanda, opened in late 2022 as the country's first carrier-neutral Tier III facility, gave the strategy a physical home. Rwanda's pitch is not market size, it is governance: predictable regulation, light corruption friction, and a government that buys locally, which together make it a credible site for regional and diplomatic workloads. The physics remains stubborn. Rwanda is landlocked and land-powered: its international bandwidth transits neighbours, and its grid is modest, hydro-heavy and exposed to drought. Kigali will keep the workloads a proud capital and its policy environment attract, but regional scale flows to the coasts and the geothermal fields. Rwanda's role in the regional map is niche, high-trust hosting, and there is nothing wrong with that. ## Ethiopia: the sleeping giant waking up Addis Ababa is the market where the starting line is furthest back and the potential is largest. Ethiopia's state telecom monopoly kept enterprise infrastructure thin for decades; liberalisation and the arrival of Safaricom Ethiopia changed the incentive structure. Raxio's Addis Ababa facility, about 3 megawatts of capacity at roughly 21,500 square feet (ocolo), and Wingu's presence at the Ethio ICT Park (Datacenters.com) make the two pan-African operators the private-sector vanguard in the country. Ethiopia's resources are real: hydro generation at enormous scale, a population above 120 million, and a government that wants data at home. Its constraints are equally real: currency and repatriation risk that gives foreign investors pause, internet controls that shape what workloads are welcome, and an enterprise IT culture still building cloud appetite. For regional planners, Addis is a market to enter carefully and early, precisely because the incumbency you build now is cheap. ![Wind turbines generating power in rural Kenya](/images/kenya-wind-turbines-5.webp) ## The comparison in one table | Market | Flagship facilities | Power story | Connectivity angle | Watch for | | --- | --- | --- | --- | --- | | Kenya | ADC NBO1/NBO2, Raxio Nairobi, Wingu, ixafrica, iColo MBA1 (Mombasa) | About 45 percent geothermal baseload | 7 live cable systems, KIXP peering | Grid capacity limits shaping hyperscale deals | | Uganda | Raxio Kampala (2020, first certified Tier III, up to 400 racks), Wingu Kampala | Hydro, seasonal exposure | Transit via Kenyan routes | In-country hosting mandates pulling workloads home | | Tanzania | Raxio Dar (6 MW, launching 2026), Wingu Dar | Gas-fired stability | Own cable landings | Carrier-neutral launch as the market catalyst | | Rwanda | Raxio Kigali (2022, first carrier-neutral Tier III) | Modest hydro grid | Landlocked, transits neighbours | High-trust niche and government workloads | | Ethiopia | Raxio Addis (about 3 MW), Wingu at Ethio ICT Park | Massive hydro potential | Largest population, thinnest base | Liberalisation pace and currency risk | ## What the comparison means Three conclusions survive the detail. First, Kenya's hub position is not sentiment, it is cables, geothermal and ecosystem depth, and it will outlast any single project's delay. Second, the neighbours are not trying to beat Nairobi; they are trying to stop depending on it for workloads their regulators and latency requirements want at home, and carrier-neutral facilities in Kampala, Dar, Kigali and Addis are exactly that strategy in building form. Third, the region's growth is real and fast: Africa's data centre market is projected to nearly double by 2030 (Wingu guide, 17 December 2025), and the continent's active capacity, 360 megawatts with another 894 megawatts in construction or planned (ADCA, February 2026), will increasingly include East African megawatts. For a Kenyan reader, the practical takeaway is opportunity, not threat. Regional operators build shared playbooks, shared talent and shared careers, and the engineers Kenya trains are the ones the region hires. The five markets are becoming one network, and Nairobi is its switching centre. ## Africa's Corridor Era and Kenya's Digital Backbone URL: https://data-centers-254.vercel.app/articles/economic-corridors-kenya-digital-backbone Cluster: Policy Published: 2026-10-20 Updated: 2026-10-20 Summary: PIDA counts 42 priority corridors and AfDB has funded 25. What Africa's corridor era means for Kenya's cables, fibre, border posts and data centres. The continent has quietly standardised on a unit of infrastructure planning, and Kenya is one of its best case studies. PIDA, the Programme for Infrastructure Development in Africa, currently identifies **42 priority corridors**; the African Development Bank has helped fund 25 of them and deployed more than $55 billion into African infrastructure over the last decade; and the Africa Center for Strategic Studies counts **79 ongoing and planned regional economic corridors** across the continent. A spotlight published by the Africa Center, "Transforming Africa's Economic Corridors for a New Era", argues these corridors are misunderstood when they are treated as roads and railways. Done right, they are economic ecosystems: production centres, industrial clusters, energy, one-stop border posts and markets moving together. The line worth keeping: *"Infrastructure becomes economically transformative when complementary investments, institutions, markets, financing, and governance operate as a single system."* Kenya's data centre conversation almost never uses the word corridor. It should, because the country's digital infrastructure is already corridor-shaped, and the corridor era now being planned will decide where the next layer of it lands. ## Corridors as ecosystems, not concrete The Africa Center's central warning is about the extract-and-export trap. A corridor that only digs up raw materials and ships them out, the historical pattern, generates transit revenue without transformation. The alternative reading, corridors as integrated economic systems, connects production centres to markets, builds industrial clusters for processing and manufacturing along the route, and deepens intra-continental trade, with AfCFTA (in force since 2019) and the AU's Agenda 2063 providing the continental frame. The financing numbers the spotlight assembles explain why governance matters as much as concrete: Africa needs roughly **$130-170 billion a year** for infrastructure but generates only about **$68 billion** domestically; over 40% of active infrastructure commitments are already financed by African governments themselves; and debt distress has made sovereign borrowing expensive, pushing the playbook toward pooled African financing, multilateral lenders and PPPs. The AfDB's 2024 structure for the Central Corridor Electric Railway, a $696 million credit guarantee releasing $3.9 billion, is the model the piece praises. For our readers the translation is direct: the same financing realpolitik that shapes a railway shapes a data centre. Facilities are capital-intensive, long-lived, and only as good as the power and connectivity that reach the site. Corridor planning is where those inputs are decided. ## Kenya is a corridor economy Look at Kenya through the Africa Center's lens and the country reads as a textbook corridor state. The **Northern Corridor**, anchored on the port of Mombasa, carries most of the freight for Kenya and its landlocked neighbours, and our own pull of Kenya's records from the African Infrastructure Database (AUDA-NEPAD's registry behind PIDA, fetched 8 September 2026) shows how much of the country's built infrastructure is corridor infrastructure: the **$5 billion Mombasa-Nairobi SGR section** sits in the registry's completed column alongside the Nairobi Southern Bypass, the Mombasa Port container terminal and the Malaba one-stop border post, while the wider Mombasa-Nairobi-Malaba-Kampala-Kigali standard-gauge programme is the registry's continental flagship at **$19.2 billion**. ![Port of Mombasa](/images/mombasa-port-wide.webp) The second corridor is **LAPSSET**, the Lamu Port-South Sudan-Ethiopia Transport corridor, Kenya's bet on a second gateway. Its registry entries are pipeline rather than concrete: a roughly **$3 billion** crude oil pipeline and a **$13.5 billion** railway, at preparation and structuring stages. And the Africa Center spotlight is refreshingly honest about the risks such projects carry: it notes the **2022 al Shabaab attacks targeting infrastructure and workers along the LAPSSET corridor in Kenya**, a reminder that corridor security is infrastructure security, and it documents the counter-example of Kenya's own SGR, financed largely through Chinese loans with lower-than-expected freight volumes and revenue that raised concerns about returns and added fiscal burden. The lesson the spotlight draws is one worth quoting precisely: anchor investment to *"phased, demand-driven development based on cargo flows, revenue potential, and secured financing."* Corridors that outrun their demand become debts. That lesson transfers to the digital layer with almost no translation. ## The digital layer was built on these corridors all along Here is the part of the story Kenya's data centre coverage usually misses: every layer of the country's digital infrastructure sits on the physical corridors' bones. **The cables landed where the port is.** Mombasa's cable station geography exists because the port existed first, and seven live submarine cable systems now surface within reach of the freight gateway. Container ships and fibre pairs chose the same beach for the same reasons. **The fibre followed the freight.** The national backbone runs along the Mombasa-Nairobi transport spine, and the SGR's civil works carried conduit the same way highway projects carry ducts. When we map [why data centres cluster in Nairobi and Mombasa](/articles/why-data-centres-cluster-nairobi-mombasa), the answer is corridor logic: power, fibre and demand concentrate along the route that freight built. **The nodes are corridor nodes.** Nairobi sits at the hinge where the Northern Corridor splits toward Uganda, Rwanda and South Sudan, and the country's largest data centre cluster, IXP and cloud on-ramps sit exactly there. Konza, Kenya's smart-city play, is positioned on the same Mombasa-Nairobi axis. The registry's flagship ICT entry, the **$395 million regional smart hub with data centre capacity at Mombasa** (PIDA I.02.30.05, being advanced as a PPP under PIDA PAP 2 for six countries), is the corridor idea applied to digital infrastructure explicitly: stack regional demand onto the node where cables, port and border already meet. Meanwhile the **$45 million Juba-Nairobi fibre link**, whose Kenyan segment is operational, is best understood as corridor plumbing for landlocked demand: every kilometre completed turns South Sudanese traffic into Kenyan colocation and transit revenue. ![Nairobi Expressway](/images/nairobi-expressway-wide.webp) What is missing is intentionality. Edge computing in East Africa is still waiting for its corridor moment: the one-stop border posts that transformed cargo transit, Malaba, Busia, Namanga, are the natural hosts for the region's next layer of edge sites, caching and content closer to cross-border trade. No one-stop data post exists yet. The corridor era is the window in which that changes, if digital infrastructure is written into the corridor plans from the start rather than bolted on after the ribbon cuttings. ## The financing question reaches the digital layer too The Africa Center's governance recommendations read as if they were written for Kenya's digital build-out. Diversify financing so no single partner holds the infrastructure hostage. Agree priorities, financing responsibilities and measurable outcomes up front, the registry is full of projects that failed this test. And think beyond the extract-and-export model: for a century Kenya's corridors exported tea and coffee; the value-addition frontier is exporting computation, turning geothermal steam, human capital and cable landing rights into AI services processed locally rather than raw capacity sold cheaply abroad. That is the economic logic behind the national compute hub ambition, and it is also why the corridor framing matters: compute hubs are industrial clusters, and industrial clusters are what corridors are supposed to produce. Two honest caveats before anyone invests on the strength of a registry entry. First, registry records are project preparations, not built facilities, and "Active" means in the pipeline, a discipline we explain in [our PIDA registry read-through](/articles/kenya-pida-infrastructure-pipeline), which is why our [infrastructure map](/infrastructure/map) keeps the PIDA layer visually and numerically separate from the verified DC Directory. Second, registry numbers are the registrant's own, published as registered and never independently verified, so we carry them dated and attributed, and you should read them the same way. ![Transmission corridor](/images/kenya-transmission-pylons-2.webp) ## What to watch Three things will tell you whether Kenya's digital infrastructure actually joins the corridor era. Watch the **Northern Corridor Transit and Transport Coordination Authority** and whether digital infrastructure, fibre ducts, border data rooms, edge siting, starts appearing in its planning documents the way one-stop border posts did a decade ago. Watch the financing structures for the **Mombasa regional smart hub**, the $395 million PPP: if it reaches financial close with pooled African and multilateral capital behind it, the corridor financing playbook has officially crossed from railways into data centres. And watch the **PIDA layer on our own infrastructure map**, which we refresh from the African Infrastructure Database's public API with each quarterly directory pass, because the gap between registry ambition and verified reality is exactly the information this platform exists to publish. The corridor era is not coming to Kenya. It already built the country's digital backbone, cable stations at the port, fibre along the freight spine, data centres at the hinge. What the next decade decides is whether the digital layer gets a seat at the corridor planning table while the concrete is still wet, or keeps discovering its geography by accident after the fact. ## Edge Computing in East Africa: Bringing Compute Closer URL: https://data-centers-254.vercel.app/articles/edge-computing-east-africa Cluster: Infrastructure Published: 2026-08-28 Updated: 2026-08-28 Summary: Edge computing places smaller facilities closer to end users, cutting latency for real-time apps. How the edge is emerging in East Africa and Kenya. Every millisecond matters more than it used to. A decade ago, a 200-millisecond delay between a user in Mombasa and a server in Nairobi was acceptable for almost every application. Today, applications like autonomous vehicles, real-time video analytics, augmented reality, remote surgery, and industrial automation require responses in under 10 milliseconds, delays that are physically impossible to achieve over a 500-kilometre round trip to a Nairobi data centre. This is the problem that edge computing solves, and it is a problem that is becoming increasingly relevant in East Africa. ![Westlands commercial district in Nairobi](/images/nairobi-westlands.webp) Edge computing is the practice of placing data processing capabilities closer to where data is generated and consumed, rather than routing all data to centralised facilities. The concept is not new, content delivery networks (CDNs) have been doing a form of edge computing for two decades by caching web content in servers distributed around the world. What is new is the range of applications that now require edge capabilities, the sophistication of the computing hardware that can be deployed at the edge, and the economic case for building edge infrastructure in emerging markets like East Africa. ## Why Edge Computing Matters Now Three technological shifts are driving edge computing from a niche concept to a mainstream infrastructure requirement. ### 5G Networks 5G networks, being deployed by Safaricom and Airtel across Kenya, are designed with edge computing built in. The 5G standard (specifically the Multi-Access Edge Computing or MEC framework) allows computing resources to be placed at or near the base station, enabling ultra-low-latency applications that would not be possible if traffic had to travel back to a central data centre. Every 5G base station is, in effect, a potential edge computing node. For Safaricom, which is Kenya's largest 5G operator, this creates both an opportunity and a requirement. The operator is already deploying edge computing resources at selected 5G sites to support enterprise applications, and this capability will expand as 5G coverage grows. For data centre operators, 5G edge computing represents a new type of customer (the telecom operators themselves) and a new model for distributed infrastructure. ### Internet of Things (IoT) IoT devices (sensors, cameras, industrial controllers, connected vehicles) generate enormous volumes of data that often needs to be processed in real time. A smart traffic management system in Nairobi cannot wait 200 milliseconds for camera footage to travel to a data centre and back before deciding whether to change a traffic light. An industrial automation system at a manufacturing plant in Thika cannot tolerate latency when a safety system needs to trigger an emergency stop. ![Nairobi skyline lit up at night](/images/nairobi-skyline-night.webp) IoT drives edge computing because it creates data at the periphery of the network, in locations that may be far from traditional data centres, and it often requires real-time processing of that data. Edge computing places the processing where the data is, reducing latency, reducing bandwidth costs (by processing data locally and only sending summaries or alerts to the central facility), and improving reliability (by continuing to function even if the connection to the central data centre is interrupted). ### AI at the Edge Artificial intelligence is moving to the edge. Rather than sending all data to a central facility for AI processing, increasingly capable AI models are being deployed on edge devices, cameras that can detect anomalies locally, sensors that can predict failures without cloud connectivity, and smartphones that run language models on-device. This trend reduces latency, improves privacy (data does not leave the device), and reduces bandwidth costs. In the Kenyan context, edge AI has particularly compelling applications. Agricultural sensors can run crop disease detection models locally without requiring internet connectivity. Security cameras at remote facilities can perform person detection and alert generation without sending video feeds to Nairobi. Mobile money systems can perform real-time fraud detection at the edge, reducing the round-trip to central fraud detection systems. ## Edge Computing in East Africa: Current State Edge computing in East Africa is in its early stages, with most deployments driven by specific use cases rather than general-purpose infrastructure. ### Content Delivery Edge The most established form of edge computing in East Africa is content delivery. CDN providers like Cloudflare, Akamai, and Google have caching nodes in Nairobi and Mombasa (and sometimes in other East African cities) that store copies of popular web content, video, and software updates. When a user in Dar es Salaam requests a YouTube video, the video is served from a local cache rather than from Google's servers in South Africa or Europe, dramatically reducing latency and improving the user experience. These CDN nodes are technically edge computing infrastructure, they process and serve data at the edge of the network. However, they are specialised for content delivery and cannot run general-purpose computing workloads. They represent the first wave of edge infrastructure, and they demonstrate that the economic model for distributed computing in East Africa works. ### Telecom Edge Safaricom's 5G deployment is driving the second wave of edge computing. 5G base stations in Nairobi, Mombasa, and other major towns are being equipped with computing resources that can handle latency-sensitive applications. These resources are primarily used for Safaricom's own network functions (mobile core processing, video optimization) and for enterprise customers who contract for edge computing services. ![Corridor inside a data centre hall](/images/hero-server-hall.webp) ### Enterprise Edge The third wave (enterprise-owned or enterprise-leased edge computing facilities) is still emerging. A bank might place a small computing facility in Mombasa to process coastal branch transactions locally. A logistics company might place edge computing at the port of Mombasa to process container tracking and customs documentation in real time. A manufacturing company might place edge computing at its factory in Athi River to run industrial automation systems. These enterprise edge deployments are typically small, a rack or two of equipment in a dedicated room, a shelter, or even a containerised data centre. They connect to the organisation's central systems in Nairobi (or to cloud services) via fibre or microwave links, sending processed data and receiving updates and configuration changes. ## Edge Computing Form Factors Edge computing facilities come in various form factors, each suited to different deployment scenarios. ![Diagram of AAU and RRU equipment on a telecom pole](/images/diagram-tower-installation-rru-aau.webp) ### Micro Data Centres A micro data centre is a self-contained computing unit, typically a rack or half-rack of equipment housed in a secure, weatherproof enclosure with integrated power, cooling, and connectivity. Micro data centres can be deployed outdoors, in locations where a traditional data centre room is not available or practical. They are used for edge computing at cell towers, industrial sites, and remote offices. ### Containerised Data Centres Shipping containers converted into data centres have been used by the military and oil and gas industry for years. In the East African context, containerised data centres offer a way to deploy computing capacity quickly in locations where building a traditional facility would take years. A 40-foot container can house 10–20 racks of equipment with integrated power and cooling, and can be deployed on a concrete pad in a matter of weeks. ### Colocation Edge Some colocation providers are developing smaller facilities specifically designed for edge computing, facilities of 100–500kW in secondary cities that serve as regional hubs for distributed computing. In Kenya, this could mean small colocation facilities in Mombasa, Kisumu, Nakuru, or Eldoret that provide professional data centre infrastructure (redundant power, cooling, security, connectivity) at a scale appropriate for edge workloads. ## Challenges Specific to East Africa Edge computing in East Africa faces challenges that differ from those in mature markets. ### Power Reliability ![Nairobi Expressway stretching across the city](/images/nairobi-expressway-wide.webp) The most significant challenge is power reliability. Edge sites are often in locations where grid power is less reliable than in Nairobi's industrial zones. A micro data centre at a cell tower in a rural area may depend on a single power source, with battery and generator backup adding cost and complexity. Power reliability is the primary factor that determines whether an edge deployment is practical at a given location. ### Physical Security Small computing facilities in remote or less secure locations are vulnerable to theft, vandalism, and unauthorised access. Securing a containerised data centre at a remote site is more difficult and more expensive per rack than securing a large, staffed facility in Nairobi. Innovative approaches (solar-powered surveillance cameras, remote monitoring, and rapid-response security partnerships with local police) are being developed to address this challenge. ### Connectivity Edge sites need reliable, high-bandwidth connectivity to central data centres and to each other. While fibre connectivity is expanding across Kenya through NOFBI and operator investments, many potential edge locations still depend on microwave links, which have lower bandwidth, higher latency, and are more vulnerable to weather disruption than fibre. ### Skills and Operations Who maintains an edge computing site in Kisumu or Mombasa? The skills required (network engineering, server management, power systems maintenance) are concentrated in Nairobi. Edge computing requires either decentralising these skills (hiring and training local technicians) or developing remote management capabilities that allow centralised teams to monitor and manage distributed edge sites. ## The Outlook Edge computing in East Africa will grow, driven by 5G deployment, IoT adoption, and the increasing need for real-time processing. The timeline is measured in years, not months, the infrastructure, skills, and business models are still developing. But the direction is clear: the future of computing in East Africa is not just centralised data centres in Nairobi, but a distributed architecture where processing happens where it is needed, at the edge of the network, close to users and data sources. For Kenya, edge computing extends the country's digital infrastructure influence beyond Nairobi. If Kenyan operators build edge infrastructure in Mombasa, Kisumu, and other towns (and if these edge nodes connect back to Nairobi's central data centres via the expanding fibre network) Kenya's data centre industry becomes a national infrastructure platform rather than a Nairobi-specific one. This is the long-term vision, and edge computing is how it becomes reality. Latency is the reason edge matters, and our guide to [reducing cloud latency for users in Kenya](/articles/reduce-cloud-latency-kenya) shows how CDNs, the AWS Local Zone in Nairobi and local hosting shorten the round trip in practice. ## Ethiopia Power Exports and Kenya Data Centres URL: https://data-centers-254.vercel.app/articles/ethiopia-power-exports-kenya-data-centres Cluster: Energy Published: 2026-09-21 Updated: 2026-09-26 Summary: Ethiopia plans to double electricity exports to Kenya. What 400 MW of regional power could mean for data-centre capacity, cost and reliability. Ethiopia is preparing to increase electricity exports to Kenya from 200 megawatts to 400 megawatts from December 2026, according to Ethiopian Electric Power. The change remains subject to technical preparation and testing of the interconnected grids (Ethiopian Electric Power, 16 September 2026). For Kenya's growing data-centre industry, the announcement matters. Not because Ethiopia is about to supply dedicated power to server halls in Nairobi. It is not. The additional electricity will enter Kenya's wider grid, alongside domestic generation and other regional imports. The more useful question is this: > **Could a larger regional electricity market improve the power environment in which Kenya's data centres operate?** Possibly. But the answer depends on far more than an extra 200 MW. It depends on transmission capacity, local connections, the delivered cost of power, grid stability and the backup systems inside every facility. ## The connection is a grid link, not a private data-centre feed Kenya and Ethiopia are connected by the Ethiopia-Kenya Electricity Highway, a 1,045-kilometre, 500-kilovolt high-voltage direct-current interconnector. The link has a transfer capacity of up to 2,000 MW, far above the 200 MW currently supplied and the planned 400 MW peak arrangement (African Development Bank, project information). ![Map of the 500 kV transmission line route from southern Ethiopia through Moyale to Suswa in Kenya](/images/ethiopia-kenya-500kv-route-map.webp) Electricity does not travel from an Ethiopian power station directly into a Kenyan data centre. It enters the national transmission system and is balanced with generation from Kenya's geothermal, hydro, wind, solar and thermal plants, as well as other imports. It must then reach the relevant substation, feeder and facility connection. That distinction is essential. The planned extra 200 MW is **not** 200 MW reserved for data centres. Homes, manufacturers, telecoms networks, public services and businesses all draw on the same system. Still, larger supply can matter to a sector that cannot tolerate interruption. Data centres run servers, cooling, network equipment, security systems and control systems continuously. Their UPS batteries and generators protect customers from grid disturbances, but those systems sit on top of the grid. They do not make grid capacity, transmission constraints or electricity prices irrelevant. ![Diagram showing the power chain from electricity grid to GPU rack](/images/diagram-power-chain-grid-to-gpu.webp) Kenya has been importing Ethiopian electricity for some time. The African Development Bank reported that the imports accounted for more than 12% of Kenya's national energy purchases in the final six months of 2024 and displaced about 635 gigawatt-hours of fossil-fuel generation over that period (African Development Bank, 2024). This is an expansion of an existing energy-trade relationship, not the start of one. Under the reported plan, peak-period supply would rise to 400 MW from 200 MW, while off-peak supply would rise to 150 MW from 65 MW. Those figures should be read as planned operating volumes, not as a guarantee of delivered capacity until the testing and operational arrangements are complete. ## What another 200 MW could change ### 1. It could improve system adequacy at the margin More imported electricity can give Kenya additional supply during high-demand periods. That may reduce pressure on the system and improve the context in which large consumers plan future load. For a data-centre developer, this is relevant to one practical question: can the grid make a credible, timely connection available at the planned site? More generation or imports improve the national picture, but they do not automatically create capacity at every Nairobi, Konza, Mombasa or secondary-city location. ![Server racks inside a Nairobi data centre hall](/images/atlancis-nairobi-datacentre-hall.webp) Transmission and distribution are the constraints that turn national megawatts into usable site power. A facility can face a long wait for a new feeder, substation upgrade or high-voltage connection even when the country has sufficient aggregate generation. That is why a developer should ask for a site-specific connection study, not rely on a national capacity headline. ![Schematic of a grid connection from a high-voltage substation through switchgear and transformers to a data centre](/images/diagram-grid-connection-anatomy.webp) ### 2. It could affect the cost conversation, but not automatically lower bills Electricity is a material operating cost for data centres. Regional trade can improve system economics when lower-cost generation is available and can be delivered through the grid. The African Development Bank said Ethiopia-Kenya imports displaced fossil-fuel generation and reduced pass-through costs during its 2024 assessment. That does **not** mean a Kenyan data centre will automatically receive cheaper electricity. The question for an operator or investor is not, "Is Ethiopian hydropower cheap?" It is, "What is the delivered, reliable cost of electricity at this specific site?" The answer includes the power-purchase arrangement, transmission and distribution charges, system losses, retail tariff structure, connection costs, backup generation, fuel contracts and the facility's target redundancy. A low wholesale import price can be useful to the system while leaving the delivered cost at a particular facility largely unchanged. ### 3. It makes renewable-power provenance more important Ethiopia's electricity system is dominated by hydropower, while Kenya has substantial geothermal generation alongside hydro, wind and solar. In a more integrated regional market, the relevant story is not a simplistic chain from one dam to one data centre. It is a wider system that can exchange electricity generated from different resources. ![The Grand Ethiopian Renaissance Dam on the Blue Nile photographed at dusk](/images/gerd-dam-dusk.webp) That may become commercially relevant as enterprise customers and cloud buyers ask more searching questions about renewable electricity, hourly matching, power-purchase arrangements and emissions. Operators should be precise: importing regional power can support a lower-carbon grid mix, but a claim about a particular facility's electricity must be backed by the applicable contract, metering and accounting method. For more context on the domestic generation advantage, read our guide to [why geothermal energy matters to Kenyan data centres](/articles/geothermal-energy-kenya-data-centres). ## Reliability is the more important story More supply is valuable. Reliable supply is decisive. Ethiopian Electric Power says a joint technical committee is overseeing testing ahead of the planned increase. That is a meaningful part of the announcement. Interconnectors allow electricity to cross borders, but they also require the connected systems to operate safely and stably. ![Capacitor stacks and valve equipment at the Suswa HVDC converter station in Kenya](/images/suswa-hvdc-converter-station.webp) A more connected East African grid can diversify supply. It can also introduce new dependencies: a transmission fault, generation shortfall, commercial dispute or operational problem outside Kenya can affect the power available to Kenya. The right conclusion is not that interconnection is risky or undesirable. It is that resilience has to be designed at several levels. At the national level, Kenya needs adequate reserves, stable frequency and voltage, robust transmission corridors and clear operating arrangements with neighbouring systems. At the facility level, data-centre operators need independent utility paths where available, UPS systems, generators, fuel autonomy, tested transfer procedures and credible maintenance discipline. ![Electrical switchgear cabinets inside a data centre power room](/images/dc-switchgear-2.webp) A customer buying colocation should therefore judge the facility, not only the national grid. Ask what happens when a utility feed fails, how long the UPS bridge lasts, when generators were last tested under load, how long on-site fuel lasts, and whether the two power paths are genuinely independent. Our explainer on [how data-centre UPS backup power works](/articles/ups-backup-power-kenyan-data-centres) covers the handover between a grid disturbance and generator power. ## The regional market is still being built The Kenya-Ethiopia link is part of a broader East African integration story. KETRACO identifies interconnections with Ethiopia, Tanzania and Uganda, including the Ethiopia-Kenya connection and the planned Kenya-Tanzania link. The infrastructure is intended to make regional electricity trading more practical, but physical links alone do not complete a market. ![A corridor of high-voltage transmission towers running through hills toward the Ethiopian border](/images/ethiopia-kenya-power-line-corridor.webp) Ethiopia has also been testing exports to Tanzania. Reporting on 19 September 2026 said the six-month pilot had ended without a final long-term commercial agreement, with volumes, tariffs, operating arrangements and transmission constraints still under discussion (The Reporter, 19 September 2026). That is a useful reminder that regional power trade progresses through engineering, regulation and commercial negotiation at the same time. ![Engineers operating switchgear inside the GERD powerhouse beneath a large Ethiopian flag](/images/gerd-powerhouse-ethiopian-flag.webp) The World Bank approved a $1.6 billion, 10-year programme in June 2026 to support electricity transmission, regional trading and system resilience in Eastern Africa. The programme includes support for the Eastern Africa Power Pool and a planned day-ahead electricity market (World Bank, June 2026). This is the larger development to watch. East Africa is gradually creating the infrastructure and market rules for electricity to become more regional. For digital infrastructure, that could widen the pool of available supply and strengthen resilience over time. It could also make power planning more complex because facility operators will need to understand cross-border supply exposure as well as local grid conditions. ## Where This Sits Among Kenya's Power Options Imports are one of three supply models now converging on Kenya's data centre conversation, and they are best understood side by side rather than added together. Domestic geothermal near the wells is the Olkaria corridor model, covered in [our Olkaria analysis](/articles/olkaria-geothermal-data-centre-corridor): putting compute where the baseload is generated instead of stretching the grid to it. Regional hydro imports are this article's model: energy generated in Ethiopia entering Kenya's bulk grid through the interconnector. The newest arrival is announced coastal fossil baseload: on 26 September 2026, Aliko Dangote said his proposed Lamu refinery would include an LNG-fired power plant of about 1,000 megawatts, with roughly 500 megawatts potentially sold to the Kenyan government, covered in [our Dangote Lamu analysis](/articles/dangote-lamu-refinery-power-plant-data-centres). The status labels differ, and the difference matters more than the headline sizes. The 400 megawatts planned from Ethiopia is a trade volume awaiting technical testing and operational arrangements. The Dangote plant is an announcement attached to an unbuilt refinery, with no power purchase agreement or financing close public, which is the same distinction this site applies to the stalled Microsoft and G42 campus announcement. Neither figure is reserved for data centres, and none of the three models' numbers should be summed into a single national total, because they describe different grid positions at different stages of certainty. If a PPA or financing close lands at Lamu, that is the trigger to re-grade the third model; until then, the December 2026 test of the higher export volumes is the milestone this article watches. ## What investors, operators and policymakers should watch It is too early to claim that the planned increase will transform Kenya's data-centre market. The evidence does not support that conclusion. It does create a useful watchlist: - **Connection capacity:** Can new facilities secure firm power at the exact site and timetable they need? - **Delivered cost:** Do lower-cost imports improve the fully delivered cost after tariffs, network charges and resilience costs? - **Grid stability:** Can Kenya integrate larger import volumes while maintaining frequency, voltage and reserve margins suitable for critical loads? - **Transmission:** Are high-voltage corridors, substations and local feeders keeping up with demand in data-centre locations? - **Renewable procurement:** Can operators substantiate renewable-energy claims with contracts, metering and a clear accounting method? - **Regional resilience:** Does cross-border trading diversify supply effectively, and how are disruption and settlement risks managed? These are not abstract policy questions. They determine whether a data-centre project can obtain power, what it costs to run, how much backup it must build and how credibly it can promise availability to customers. ## The infrastructure beneath the infrastructure Data centres are often discussed as buildings full of servers. In reality, they sit on a chain of physical systems: ```text Electricity generation ↓ Transmission ↓ Distribution ↓ Data-centre power systems ↓ Servers and cooling ↓ Networks and fibre ↓ Cloud, AI and digital services ``` A weakness in any layer can affect the layers above it. That is why the Ethiopia export announcement belongs in a conversation about Kenya's data-centre future. ![Schematic of a data centre microgrid combining grid supply, on-site generation and battery storage](/images/diagram-datacentre-power-microgrid.webp) The immediate test comes in December 2026, when the planned higher export volumes are expected to begin if technical preparations are completed. The key evidence to watch will be delivered volumes, system performance, commercial terms and whether Kenyan grid operators can translate additional imports into dependable service where new digital infrastructure is being built. The servers may sit in Kenya. Increasingly, the power system behind them is regional. ## Frequently asked questions ### Will Ethiopia's additional electricity supply power Kenya's data centres directly? No. The planned additional supply enters Kenya's interconnected grid and serves the wider electricity system. It is not a dedicated allocation for data centres. Any benefit to a facility depends on local connection capacity, transmission, tariffs, contract terms and the facility's own redundancy design. ### Does importing 400 MW make Kenya's grid reliable enough for data centres? Not by itself. More supply can help system adequacy, but data-centre reliability also depends on grid stability, the strength of transmission and distribution at the site, dual utility feeds, UPS systems, generators, fuel logistics and tested operating procedures. ### Could Ethiopian imports reduce data-centre electricity prices in Kenya? They could improve system economics if lower-cost generation is delivered efficiently, but there is no automatic pass-through to a specific data centre. Delivered cost depends on the power contract, transmission and distribution charges, losses, tariffs and the cost of resilience at the chosen site. ## Europe's Digital Playbook for Africa: Kenya's Part URL: https://data-centers-254.vercel.app/articles/eu-africa-digital-cooperation-kenya Cluster: Policy Published: 2026-09-08 Updated: 2026-09-15 Summary: A new EU security institute brief maps how Europe wants to partner on Africa's digital infrastructure and AI, with Ghana and Kenya as test cases. Africa's digital infrastructure has become an arena of great-power competition, and Europe has finally published its game plan. A new brief from the European Union Institute for Security Studies, *Africa's digital edge: Building EU-Africa digital cooperation*, by Rossella Marangio and Clotilde Bômont (7 September 2026), lays out how the EU intends to compete in a market where China builds, America sells, and African governments are, refreshingly, the ones holding the selection committee. For those of us tracking Kenya's data centre market, the document is more than diplomacy: it describes the financing and policy machinery that will shape who builds what, where, over the next decade. ![Policy and regulation shaping Africa's digital infrastructure](/images/dc-policy-regulation.webp) ## The numbers behind the contest The brief grounds its argument in the market's momentum. Between 2019 and 2024, the share of Africans using the internet rose by 37%. Africa's digital transformation market (valued at **USD 30.24 billion in 2025**) is projected to reach **USD 72.23 billion by 2031**. Those are the stakes the external partners are bidding for. The brief's framing is candid: China, the US, South Korea and India are all actively competing to shape African digital markets, and African governments are responding by pursuing **strategic autonomy, supplier diversification and local value creation**, which is exactly where Europe believes its offer fits. That offer converges on six priorities: **trusted connectivity, data governance, AI, cybersecurity, digital public infrastructure, and skills**. The word doing the heavy lifting is *trusted*, Europe's differentiation pitch is security, standards and governance-aligned infrastructure, against rivals whose offers are faster, cheaper, or both. ## Kenya and Ghana: the two test cases The brief's most useful section for our readers is its pairing of country models. **Ghana** illustrates the market-led partnership: a stable, competitive telecoms and cloud environment where European engagement works through commercial frameworks. **Kenya** illustrates the comprehensive model, and here the brief could have been written from our own research notes. Kenya brings East Africa's densest data centre cluster ([27 verified facilities](/directory), the region's deepest interconnection), an existing **EU-Kenya Digital Partnership** with dedicated funding, a regulator with a track record, and a data protection regime that aligns naturally with European standards. The machinery behind the Kenya relationship is already visible on the ground. The EU's **Global Gateway** framework has channelled significant digital funding toward Kenya, including the EUR 102 million EU-Kenya Digital Partnership package announced in June 2026 and the AfricaConnect research-and-education network boost. The **Digital Investment Facility** (implemented with GIZ and HAUS under the Data Governance in Africa initiative) commissions the Xalam Analytics country data centre briefings we regularly cite ([the free Kenya brief](https://d4dhub.eu/news/dif-country-market-briefs) is the best public market document in this space), alongside green data centre e-learning and cybersecurity guidance. And the **D4D Hub** provides the political umbrella, with Team Europe actively courting African counterparts at events from GITEX Africa to the New Space Africa conference. ![ICT training for Kenya's digital workforce](/images/classroom-ict-training-kenya.webp) ## The honest reading: frameworks are cheap, financing is not The brief is admirably clear-eyed about its own risks, and three of its caveats deserve to be quoted in every Kenyan boardroom that touches European infrastructure money. First, **success depends on financing, implementation and local ownership**, cooperation frameworks are abundant in African digital policy; shovels and switchgear are not. Second, the EU must deliver **credible alternatives** to other external partners, which means matching their speed and price points, not just their values. Third, everything must be **African-led**, the brief explicitly frames the future as a coherent, African-led integration of security, infrastructure, markets and governance, not a European checklist. For Kenya's data centre sector specifically, the practical translation is this: European money is most likely to arrive as **enablers rather than landlords**. Expect EU-aligned financing for power infrastructure (the binding constraint on every hyperscale announcement, as the Microsoft–G42 stall demonstrated), for skills programmes that feed the [careers pipeline](/careers), and for governance frameworks that make Kenyan-hosted data trustworthy across borders. Do not expect the EU to build data centres directly, that is private capital's job, with the EU de-risking what private capital cannot price. ## What to watch in the Kenyan market Three indicators will show whether the playbook is working. One: whether Global Gateway and DIF financing visibly moves Kenya's power constraint, new geothermal and transmission capacity is the precondition for every 50 MW-class data centre ambition in the country. Two: whether the EU-Kenya Digital Partnership produces *interoperable* data governance, actual mechanisms that let Kenyan-hosted data serve European customers under adequacy-style arrangements, which would make local colocation dramatically more valuable to multinationals. Three: whether African institutions, not Brussels pressrooms, own the narrative, the brief's own test. Our [regulation coverage](/articles/kenya-data-centre-licensing-framework) and [market-in-numbers reference](/articles/kenya-data-centre-market-numbers) will keep tracking all three, because the party that builds Kenya's digital edge will be decided less in strategy documents than in power purchase agreements, cable landings and the unglamorous work of verification. ## Fibre Networks and Kenya's Data Centre Last Mile URL: https://data-centers-254.vercel.app/articles/fibre-optic-networks-kenya-data-centres Cluster: Connectivity Published: 2026-08-28 Updated: 2026-09-16 Summary: Kenya's 15,000+ km of fibre connects its data centres to the world. Operators, pricing, redundancy, and the last-mile challenges that decide speed. A data centre without fibre connectivity is a warehouse full of expensive computers that nobody can reach. Every transaction on M-Pesa, every query to a bank's application, every page load on a Kenyan government website, all of it travels over fibre optic cables that connect data centres to users, to each other, and to the global internet. Kenya's fibre optic network is the circulatory system of its digital economy, and understanding how it works, who operates it, and where the weaknesses are, is essential for anyone investing in or depending on digital infrastructure. ![Fibre optic termination equipment in a data centre](/images/dc-fibre-optics.webp) Kenya has built one of the most extensive fibre networks in sub-Saharan Africa over the past two decades. The combination of submarine cable landing stations in Mombasa, a competitive operator landscape, and government investment through the National Optic Fibre Backbone Infrastructure has created a network that reaches all 47 counties and connects to the global internet through multiple redundant paths. But coverage is uneven, pricing is opaque, and the "last mile" (the final connection from the fibre backbone to the actual data centre or office building) remains the most expensive and problematic part of the chain. ## The Major Fibre Operators in Kenya Kenya's fibre market is dominated by four operators, each with distinct strengths, network footprints, and strategic positions. Understanding who they are and what they own is the first step to understanding data centre connectivity. ### Liquid Intelligent Technologies Liquid Intelligent Technologies, formerly Liquid Telecom, is the largest fibre operator in Kenya by route kilometres. The company, now part of the Cassava Technologies group (which also owns Africa Data Centres), has built a pan-African fibre network spanning over 100,000 kilometres across more than 20 countries. In Kenya, Liquid's network connects Nairobi, Mombasa, Kisumu, Nakuru, Eldoret, and most major towns, with multiple redundant rings around Nairobi for metro connectivity. Liquid's strategic advantage for data centre connectivity is vertical integration. Because the same parent company owns Africa Data Centres, Liquid can offer bundled connectivity and colocation packages that are difficult for competitors to match on price. A customer placing servers in an Africa Data Centres facility can get connectivity, colocation, and managed services from a single vendor, with SLAs that span the entire stack. This integration also means Liquid's fibre routes are optimised for data centre-to-data-centre traffic, not just for connecting enterprise branch offices. ### Safaricom Safaricom is Kenya's largest telecom operator by revenue and subscriber base, and its fibre network is extensive, particularly in Nairobi, Mombasa, and the corridors connecting major towns. Safaricom's fibre strategy has historically focused on serving its own needs (connecting base stations, enterprise clients, and its own data centres) but the company has increasingly offered wholesale fibre to third-party data centres and enterprises. Safaricom's unique advantage is its mobile network. The company operates the largest 4G and 5G network in Kenya, and many data centre customers want connectivity that spans both fixed fibre and mobile networks. Safaricom can offer integrated solutions that include dedicated fibre links, mobile backhaul, and even direct interconnection with the M-Pesa platform, a compelling proposition for financial services companies. ![Shore-based cable works on the Kenyan coast with a vessel offshore](/images/mombasa-cable-landing.webp) ### Telkom Kenya Telkom Kenya, majority-owned by the Helios Investment Partners-led consortium (now rebranded as Jamhuri Holdings), operates a significant fibre network that includes the former Kenya Telkom backbone inherited from the privatisation era. Telkom's network is particularly strong in Nairobi's central business district and along the Mombasa Road corridor where most data centres are concentrated. Telkom's competitive position is built on pricing and its role as an alternative to Safaricom and Liquid. For data centre operators, having Telkom as a second or third fibre provider is important for redundancy, it ensures that traffic can reroute if the primary provider has an outage, and it provides negotiating leverage on pricing. ### Jamii Telecommunications (Wananchi Group) Jamii Telecommunications, part of the Wananchi Group, operates fibre networks primarily in Nairobi's residential and commercial areas. The company's Waba fibre-to-the-home service has made it a significant player in the last-mile market, and it also provides enterprise and data centre connectivity. Jamii's advantage is its metro footprint in areas like Westlands, Kilimani, and Karen, neighbourhoods with high concentrations of businesses that need data centre connectivity. ## How Fibre Reaches a Data Centre ![Fibre splicing and termination panel with test light](/images/fibre-splicing-termination.webp) The path from a user's phone or laptop to a server inside a data centre involves multiple segments of fibre, each managed by different entities, each with different reliability characteristics and cost structures. Understanding this path is critical for understanding data centre connectivity. ### Submarine Cables to Landing Stations The journey begins (or ends) at the submarine cable landing stations in Mombasa. Kenya has seven live submarine cable systems as of September 2026: SEACOM, TEAMS, EASSy, LION2, DARE1, PEACE, and 2Africa, with Africa-1 landed at Mombasa in 2024 pending full service and Daraja in development. Each cable lands at dedicated stations along the Mombasa coast and connects to terrestrial fibre networks that carry traffic inland to Nairobi, where the data centres are. The [submarine cables themselves](/articles/submarine-cables-landing-mombasa) provide Kenya's connection to the global internet, to Europe, Asia, the Middle East, and the rest of Africa. ### Long-Haul Backbone: Mombasa to Nairobi From the landing stations, traffic travels over long-haul fibre to Nairobi. This 500-kilometre route is one of the most critical infrastructure segments in Kenya's digital economy. Multiple operators have built parallel fibre routes between Mombasa and Nairobi, and most follow the Mombasa-Nairobi highway corridor. The diversity of routes is important, a single fibre cut along this corridor could potentially degrade connectivity for the entire country if all operators' cables followed the same physical path, which is why regulators encourage route diversity. ### Metro Fibre: Nairobi Distribution Once traffic reaches Nairobi, it is distributed over metropolitan fibre networks to individual data centres, offices, and base stations. Nairobi's metro fibre network is the densest in East Africa, with multiple concentric rings providing redundancy. Major data centre clusters along Mombasa Road, in Westlands, and in Sameer Business Park are served by multiple fibre providers, each entering the buildings through different ducts and manholes to ensure physical route diversity. ### In-Building Fibre and Cross-Connects The final segment is inside the data centre itself. Fibre from external providers terminates in a meet-me room (MMR) or carrier room, where patch panels allow connections to be made between providers and customers. A "cross-connect" is a physical cable (often a short fibre patch cord) that connects a customer's rack or cage to a specific provider's network. Cross-connects are how customers choose which networks they connect to, and they are typically provisioned within 24–48 hours in a well-managed facility. ![Aerial view of the Port of Mombasa and its container terminal](/images/mombasa-port-wide.webp) ## Redundancy: The Most Critical Requirement For data centre operators and their customers, fibre redundancy is not optional, it is the single most important connectivity requirement. A data centre with a single fibre connection is a single point of failure, regardless of how reliable that connection is on a day-to-day basis. The standard requirement for a carrier-neutral colocation facility is at least two independent fibre connections from different providers, entering the building through different physical routes. This means the cables should not share the same trench, the same manhole, or the same conduit, if a construction crew digs up one road and cuts a fibre, the other connection should be unaffected. Beyond dual entry, larger facilities aim for three or more diverse fibre paths. This provides additional resilience against multiple simultaneous failures and also enables traffic engineering, routing different types of traffic (e.g., latency-sensitive financial transactions vs. bulk data transfers) over different paths based on their characteristics. ## The National Optic Fibre Backbone Infrastructure (NOFBI) ![Cable drums staged for urban fibre works](/images/metro-fibre-civil-works.webp) The Kenyan government has invested heavily in fibre through the National Optic Fibre Backbone Infrastructure programme. NOFBI was designed to connect all 47 county headquarters and major government institutions, providing a backbone that government agencies, schools, and hospitals could use for connectivity. NOFBI's relationship with commercial data centres is indirect but relevant. The network provides an alternative path for traffic, particularly in areas where commercial operators have not invested. In some cases, data centres in smaller towns (like Kisumu or Nakuru) may access NOFBI through peering arrangements, though the government network is not designed or priced for commercial data centre use. There have been periodic discussions about making NOFFI available to commercial operators on a wholesale basis, which could significantly reduce connectivity costs in underserved areas. ## Pricing and Cost Structures Fibre connectivity pricing in Kenya is complex and often opaque. The market lacks the transparent pricing that characterises more mature markets like South Africa or Europe, and pricing varies significantly based on the customer's size, contract length, and negotiating leverage. For data centre connectivity, the key pricing models are dedicated internet access (DIA), where the customer gets a committed bandwidth with symmetrical upload and download speeds; point-to-point links, which connect two specific locations (e.g., a bank's headquarters to its disaster recovery site in a data centre); and peering/transit, where the data centre connects to internet exchange points and upstream providers for global reach. A 1Gbps dedicated internet connection in Nairobi typically costs between KES 80,000 and KES 200,000 per month. 10Gbps connections range from KES 400,000 to KES 1.2 million per month, with significant volume discounts available for larger commitments. These prices have fallen substantially over the past five years (a 1Gbps connection that cost KES 300,000 in 2020 can now be had for under KES 100,000) but they remain higher than equivalent connections in South Africa or Egypt, reflecting Kenya's smaller market and the costs of serving a geographically dispersed customer base. ## Challenges and Gaps Despite the progress, Kenya's fibre infrastructure faces significant challenges. Last-mile connectivity (the final segment from the backbone to the customer's premises) remains the most expensive part of the chain, often accounting for 40–60% of the total cost of a connectivity service. Getting fibre into older buildings in Nairobi's CBD, where building managements may charge access fees or where physical duct space is limited, can be prohibitively expensive. Right-of-way challenges persist. Despite regulations requiring utility companies to coordinate and share duct space, fibre deployments are frequently delayed by county government permits, road construction, and disputes with other utility operators. These delays increase costs and slow the expansion of connectivity to new areas. ![Labelled diagram of a fibre patch panel with SC and FC connectors](/images/dc-networking.webp) Rural and peri-urban areas remain underserved. While NOFBI has connected county headquarters, the last-mile from county towns to individual facilities, schools, and businesses is often still missing. This matters for data centres because edge computing (placing smaller data processing facilities closer to end users) requires fibre connectivity in locations that currently lack it. ## The Future: 5G Backhaul, Edge, and New Submarine Cables Looking ahead, several trends will shape Kenya's fibre infrastructure. 5G deployment by Safaricom, Airtel, and others will require massive fibre backhaul, every 5G base station needs a fibre connection, and the density of 5G cells means far more fibre endpoints than 4G required. This will drive further investment in metro fibre networks. ![Diagram of a long-distance microwave link between two sites](/images/diagram-microwave-data-link.webp) Edge computing, where smaller data processing facilities are placed in towns and neighbourhoods rather than centralised in Nairobi, will require fibre connectivity in new locations. As applications like autonomous vehicles, remote surgery, and real-time AI inference demand lower latency than a 500-kilometre round trip to Nairobi can provide, edge facilities in Mombasa, Kisumu, and other towns will become necessary. New submarine cable systems, including 2Africa (Meta-backed) and Equiano (Google-backed, landing in South Africa but with branching units that could serve East Africa), will add further capacity and resilience. More capacity means lower per-megabit costs, which will eventually flow through to data centre connectivity pricing. Kenya's fibre network is the foundation on which its entire digital economy is built. For data centre operators, the quality, diversity, and cost of fibre connectivity is a primary differentiator, and for customers choosing between facilities, it should be one of the first questions asked. ## Fire Suppression Systems in Kenyan Data Centres URL: https://data-centers-254.vercel.app/articles/fire-suppression-systems-kenyan-data-centres Cluster: Infrastructure Published: 2026-08-28 Updated: 2026-08-28 Summary: An in-depth guide to fire detection and suppression in Kenyan data centres: clean agents, VESDA, regulatory requirements, real installations. ![Server racks in a modern data centre hall](/images/whats-inside-ai-data-center-wide.webp) Fire protection in a data centre is fundamentally different from fire protection in an ordinary commercial building. The very systems designed to suppress fires in offices and warehouses (primarily water-based) can cause more damage than the fire itself when deployed among servers, storage arrays, and network switches. For Kenyan data centre operators, designing and maintaining an appropriate fire suppression system is both a technical imperative and a regulatory and insurance requirement. This guide covers the full spectrum of **fire suppression systems in Kenyan data centres**, from detection technologies to clean agent chemistries and the regulatory landscape. ## Why Water Sprinklers Are Problematic in Data Centres Traditional wet-pipe sprinkler systems are the default fire protection method in most Kenyan commercial buildings under the Kenya Fire Brigade Act and county by-laws. However, in a data centre environment, water is arguably as dangerous as fire: - **Equipment destruction:** Even a brief sprinkler discharge can destroy servers, SANs, and network switches. Water conducts electricity, so energised equipment hit by sprinklers can short-circuit catastrophically - **Corrosion and contamination:** Water from sprinkler systems often carries rust, pipe scale, and chemical additives that accelerate corrosion of electronic contacts - **Collateral damage:** A single sprinkler head typically discharges 15–25 litres per minute. In a data hall with hundreds of millions of shillings in customer equipment, the financial exposure is immense - **Extended downtime:** Water-damaged equipment requires thorough drying, cleaning, and often complete replacement. Recovery timelines can stretch from weeks to months For these reasons, Kenyan data centres use **pre-action dry pipe sprinklers** (which only fill with water after detection confirms an actual fire) in non-critical areas like corridors, offices, and electrical rooms, and **clean agent suppression systems** in the white space and critical infrastructure areas. ## Clean Agent Fire Suppression Systems ### FM-200 (HFC-227ea) **FM-200** has been the most widely deployed clean agent in Kenyan data centres over the past two decades. It works by absorbing heat from the fire, interrupting the combustion chain reaction. Key characteristics: - **Concentration:** Typically designed at 7–8.6% by volume for Class A (ordinary combustibles) and Class C (electrical) fires - **Discharge time:** 10 seconds or less to reach design concentration - **Hold time:** Minimum 10 minutes to prevent re-ignition - **Safety:** NOAEL (No Observed Adverse Effect Level) is 9%, providing a reasonable safety margin at design concentrations - **Environmental concern:** FM-200 has a **GWP of approximately 3,220** and an atmospheric lifetime of about 34 years. While it has zero Ozone Depletion Potential (ODP), its high GWP places it under phase-down schedules Africa Data Centres' Nairobi facility and several banking data centres in the city initially deployed FM-200 systems. However, the phase-down trajectory under the **Kigali Amendment** to the Montreal Protocol is driving a transition to newer alternatives. ### Novec 1230 (FK-5-1-12) **Novec 1230**, manufactured by 3M (now part of Solvay), is increasingly the agent of choice for new Kenyan data centre builds: - **Concentration:** Designed at 4.2–5.9% for typical data centre hazards - **GWP:** Less than 1, essentially equivalent to carbon dioxide - **Atmospheric lifetime:** Approximately 5 days, it breaks down rapidly in the lower atmosphere - **Safety:** NOAEL is 10%, well above typical design concentrations, making it the safest clean agent available - **Storage:** Liquid at room temperature, requiring smaller storage cylinders compared to gaseous agents **IXAfrica**, Nairobi's newer carrier-neutral data centre, has specified Novec 1230 for its white space fire protection, positioning itself as environmentally responsible while meeting the stringent fire protection requirements of international colocation customers. ### Inergen (IG-541) **Inergen** is an inert gas blend of 52% nitrogen, 40% argon, and 8% carbon dioxide. It suppresses fire by reducing oxygen concentration to approximately 12.5%, below the level that supports combustion but above the level that sustains human life (the CO2 component helps the body absorb the reduced oxygen): - **Concentration:** Typically 34–42% by total room volume, significantly higher than chemical agents - **Environmental impact:** Zero ODP, zero GWP, entirely composed of atmospheric gases - **Storage:** Requires high-pressure cylinders (200 or 300 bar), demanding significant structural considerations for cylinder storage rooms - **Cost:** High cylinder count and pressure requirements make Inergen systems more expensive to install than FM-200 or Novec 1230 Inergen is less commonly deployed in Kenyan data centres due to the high-pressure storage requirements, but it remains an option for facilities that prioritise zero environmental impact above all other considerations. ![Data centre rooftop chillers and generators](/images/dc-chillers-roof.webp) ## Environmental Impact and the Kigali Amendment The **Kigali Amendment to the Montreal Protocol**, which Kenya ratified in 2017, mandates the phasedown of hydrofluorocarbons (HFCs), the chemical family that includes FM-200. Under this agreement, developing countries (including Kenya) must begin reducing HFC consumption from 2024, with a target of 80% reduction by 2045. For Kenyan data centre operators, this has practical implications: - **FM-200 supply constraints:** As HFC production is curtailed globally, the cost and availability of FM-200 for refills and new installations will become increasingly problematic - **Transition planning:** Facilities currently using FM-200 need to plan for system replacement or conversion to Novec 1230 or inert gas alternatives within the next 10–15 years - **Environmental credentials:** Multinational tenants and sustainability-conscious customers increasingly prefer facilities using low-GWP agents, making Novec 1230 a competitive differentiator Kenya's National Environment Management Authority (**NEMA**) oversees the implementation of the Kigali Amendment domestically and requires facilities using ozone-depleting or high-GWP substances to maintain detailed inventory records and submit annual reports. ## Fire Detection Systems ### VESDA (Very Early Smoke Detection Apparatus) **VESDA** by Xtralis (now part of Honeywell) is the industry standard for early warning smoke detection in data centres. The system works by: 1. Drawing air continuously through a network of **PVC or ABS sampling pipes** installed in the ceiling void, under the raised floor, and at cabinet level 2. Passing air samples through a **laser-based detection chamber** that identifies microscopic smoke particles (as small as 0.0015 microns) 3. Analysing particle density and reporting alarm levels: **Alert**, **Action**, and **Fire** The value of VESDA lies in its **aspirating detection** capability, it can detect fires at the smouldering stage, often minutes or hours before conventional point detectors would trigger. This early warning allows operators to investigate potential issues (such as an overheating UPS battery or a failing power supply) and take corrective action **before** the fire suppression system discharges. In Kenyan facilities, VESDA is deployed in the white space, electrical rooms housing UPS systems and switchgear, and generator rooms. Africa Data Centres' Nairobi campus uses VESDA with multiple detection zones, each independently addressable to the Building Management System (BMS). ### Beam Detectors **Infrared beam detectors** project a beam of infrared light across the data hall from a transmitter to a receiver. Smoke particles that enter the beam path scatter or absorb the infrared light, triggering an alarm. Beam detectors are useful in large, open data halls where installing point detectors on the ceiling would require extensive wiring and could obstruct airflow. ### Multi-Criteria Detectors **Multi-criteria detectors** combine multiple sensing technologies (typically smoke, heat, and sometimes carbon monoxide) in a single device. They are programmed to require confirmation from more than one sensor before triggering an alarm, significantly reducing false alarm rates. These are increasingly used in Kenyan data centre support areas (offices, meeting rooms, corridors). ## Regulatory Framework for Fire Protection in Kenya ### Kenya Fire Brigade Act (Cap 211) The **Kenya Fire Brigade Act** is the primary national legislation governing fire safety. It establishes the powers of fire authorities and requires building owners to maintain fire safety measures. While the Act predates modern data centres, its provisions apply to all buildings, including technology facilities. ### County Fire By-Laws **Nairobi City County** and **Mombasa County** each enforce their own fire safety by-laws that supplement national legislation. Key requirements for data centres include: - **Fire-rated compartmentalisation:** Minimum 2-hour fire-rated barriers between the data hall and adjacent spaces - **Fire detection and alarm systems:** Addressable fire alarm systems with central monitoring - **Emergency exits:** Minimum two independent means of egress from the data hall - **Fire extinguishers:** CO2 and clean agent portable extinguishers at designated points (water and foam extinguishers must NOT be used on electrical equipment) - **Fire safety officer:** Facilities above a certain size must employ or retain a qualified fire safety officer ### NEMA Requirements The **National Environment Management Authority** requires data centres to: - Register any ozone-depleting substances or high-GWP substances used in fire suppression systems - Submit annual inventory reports - Develop and maintain an **Environmental Management Plan** that includes fire emergency response procedures - Ensure that fire suppression agents are handled and disposed of in accordance with the Montreal Protocol and its amendments ### Insurance Company Requirements Kenyan insurers underwriting data centre risks, including **Kenya Reinsurance Corporation (Kenya Re)**, **Jubilee Insurance**, **UAP Old Mutual**, and **CIC Insurance**, typically impose fire protection requirements that often exceed minimum regulatory standards. Common insurance stipulations include: - Clean agent fire suppression in all areas housing electronic equipment - VESDA or equivalent early warning detection - Annual system testing and certification by a licensed fire protection contractor - Fire-rated construction meeting minimum hourly ratings - Documented evacuation procedures and fire drill records - Compliance with NFPA 75 or equivalent international standards Failure to maintain these systems can result in **policy voidance**, premium surcharges of 25–100%, or denial of claims following a fire incident. ## System Design Considerations ### Zoning Effective fire suppression design divides the data centre into **protection zones**, each with independent detection and suppression capability. Typical zoning includes: - **White space zones:** Individual data halls or sections of halls, each treated as a separate suppression zone - **Electrical rooms:** UPS rooms, switchgear rooms, and battery rooms each have dedicated systems - **Mechanical rooms:** Chiller plants, cooling towers, and HVAC equipment rooms - **Support areas:** Corridors, offices, and meeting rooms typically use conventional fire detection with pre-action sprinklers ### Total Flooding vs Local Application **Total flooding** (the standard approach in data centre white spaces) involves filling the entire protected zone with the fire suppression agent to the design concentration. This requires the room to be reasonably well-sealed to prevent agent escape. **Local application** directs agent only at the specific hazard area and is generally used for smaller, contained risks. Kenyan Tier III data centres use **total flooding** for the white space. Sealing integrity is verified during commissioning using a **door fan test** (also called a blower door test) that measures the room's leakage rate and calculates the required agent quantity to compensate for leakage during the hold time. ### Hold Times and Evacuation Clean agent systems are designed to maintain the design concentration for a minimum **hold time of 10 minutes**, sufficient to ensure the fire is fully suppressed and will not re-ignite when ventilation is restored. During this period: - **Pre-discharge alarms** (typically 30 seconds) warn personnel to evacuate the protected zone - **Abort switches** at all exits allow authorised personnel to cancel a discharge if investigation reveals a false alarm - **Access control integration** ensures fire doors unlock automatically during a fire event Kenyan facilities are required to conduct **quarterly fire drills** and maintain evacuation plans that have been reviewed and approved by a qualified fire safety officer. ### Integration with Building Management Systems Modern Kenyan data centres integrate fire detection and suppression into their **Building Management System (BMS)**, which provides: - Centralised monitoring of all detection zones - Automated sequences: detection → alarm → pre-discharge warning → agent discharge → HVAC shutdown → notification - Event logging for compliance auditing and insurance documentation - Integration with [data centre security explained](/articles/data-centre-security-explained) systems for coordinated emergency response ![Blue-lit server racks in a data hall](/images/hero-server-hall.webp) ## Cost Implications | **System Type** | **Approx. Cost (KES)** | **Pros** | **Cons** | |-----------------------------|------------------------|----------------------------------------------|-----------------------------------------------| | Wet-pipe sprinklers | 3M – 8M | Low cost, simple, well understood | Destroys equipment, not suitable for white space | | Pre-action dry pipe | 8M – 15M | Water damage limited to confirmed fires | Still uses water, slower response | | FM-200 clean agent | 40M – 100M | Fast, effective, no residue | High GWP, phase-down underway | | Novec 1230 clean agent | 50M – 150M | Near-zero GWP, excellent safety profile | Highest upfront cost | | Inergen (IG-541) | 60M – 180M | Zero environmental impact, inert gases | High-pressure storage, many cylinders needed | These costs include agent, cylinders, piping, nozzles, detection systems, control panels, and installation. **Annual maintenance** typically adds 5–10% of the initial system cost. ## Maintenance and Inspection Schedules Fire suppression systems require rigorous ongoing maintenance to remain effective: ### Monthly - Visual inspection of all detection devices and control panels - Verification that abort switches and manual release stations are unobstructed - Check that all zone indicator lights and alarms are functional ### Quarterly - Test a representative sample of detection devices (typically 25% per quarter, rotating to achieve 100% annually) - Inspect cylinder pressures and weights - Verify door seals and room integrity for total flooding zones - Conduct fire drills and review evacuation procedures ### Annually - Full system test by a **licensed fire protection contractor** registered with the Kenya Fire Brigade - Cylinder hydrostatic testing (typically every 5–10 years depending on agent type) - Recalibration of VESDA detectors - Review and update fire safety documentation - Submit inspection certificates to insurers and NEMA as required ### Special Inspections - After any system discharge, the entire system must be inspected, recharged, and re-certified before returning to service - After any structural modifications to the protected space (changes to walls, ceilings, or penetrations) that could affect room integrity - After changes to the [data centre design and construction](/articles/data-centre-design-build-kenya) layout that alter the protected volume or hazard classification ## Comparison to International Standards ### NFPA 75 and NFPA 76 The **National Fire Protection Association (NFPA)** publishes two standards directly relevant to data centre fire protection: - **NFPA 75: Standard for the Fire Protection of Information Technology Equipment**, covers fire protection for IT equipment areas within buildings - **NFPA 76: Standard for the Fire Protection of Telecommunications Facilities**, covers telecommunications equipment and infrastructure While NFPA standards are American, they are widely referenced in Kenyan data centre specifications, particularly by multinational operators and their insurers. Key NFPA 75 requirements that Kenyan facilities adopt include: - Automatic fire detection throughout the IT equipment area - Fire suppression appropriate for the protected hazard (clean agents for electronic equipment) - Compartmentalisation with fire-resistant barriers - Emergency power-off capability as a last-resort fire control measure - Documentation of all fire protection systems and maintenance records ### EN 54 The **EN 54** family of European standards covers fire detection and fire alarm systems. EN 54-compliant equipment is increasingly available in the Kenyan market through European distributors. Facilities with European investment or serving European clients may specify EN 54 compliance for detection systems. ## Real Installations in Kenyan Facilities ### Africa Data Centres, Nairobi Africa Data Centres' flagship Nairobi campus deploys a **zoned clean agent system** (FM-200, with plans to transition to Novec 1230) protecting the white space, complemented by VESDA aspirating detection throughout. The facility maintains dedicated fire suppression cylinder rooms with structural reinforcement and climate control. Annual inspections are conducted by a licensed contractor with certificates provided to all colocation tenants and insurers. ### IXAfrica, Nairobi As a newer build, **IXAfrica** specified Novec 1230 from the outset, aligning with international best practices for environmental sustainability. The facility uses a multi-zone VESDA system with addressable detection integrated into a centralised BMS. Fire-rated compartmentalisation between the white space and support areas exceeds 2-hour ratings. ### PAIX Nairobi The **Pan African Internet Exchange** facility, serving as a critical interconnection hub, employs clean agent suppression in its equipment halls and pre-action sprinklers in support areas. Given its role as a peering and interconnection point, even brief fire-related outages would have wide-ranging connectivity impacts across East Africa. ## Frequently Asked Questions ### Why don't Kenyan data centres use water sprinklers in the server hall? Traditional water sprinklers cause catastrophic damage to electronic equipment even when they successfully suppress a fire. A sprinkler discharge can destroy servers, storage arrays, and networking gear far beyond the area of the original fire, resulting in data loss and extended downtime. Clean agent systems suppress fires without leaving residue or causing water damage. Most Kenyan Tier III facilities use clean agents in the white space and may have pre-action dry pipe sprinklers in non-critical areas like corridors and offices as a secondary measure. ### What is the difference between FM-200 and Novec 1230? FM-200 (HFC-227ea) and Novec 1230 (FK-5-1-12) are both clean agent fire suppressants, but they differ significantly. FM-200 has a Global Warming Potential (GWP) of approximately 3,220 and is being phased down under the Kigali Amendment to the Montreal Protocol. Novec 1230 has a GWP of less than 1, an atmospheric lifetime of just 5 days, and is not subject to phase-down schedules. Novec 1230 also has a wider safety margin, it can be used at higher concentrations relative to its NOAEL (No Observed Adverse Effect Level). For new Kenyan data centre builds, Novec 1230 is increasingly the preferred choice despite its higher upfront cost. ### How does VESDA smoke detection work and why is it preferred? VESDA (Very Early Smoke Detection Apparatus) is an aspirating smoke detection system that continuously draws air samples through a network of piping installed in the ceiling void and under the raised floor. A central laser-based detector analyses these samples for microscopic smoke particles, detecting fires at the incipient stage, often before visible smoke or flame appears. In a data centre context, VESDA provides critical early warning that allows operators to investigate and potentially resolve issues (such as an overheating power supply) before a fire suppression discharge is triggered. This early detection capability is why VESDA is standard in Kenyan Tier III and above facilities. ### What do Kenyan insurers require for data centre fire protection? Kenyan insurance companies underwriting data centre risks typically require: certified fire suppression systems with annual inspection certificates, VESDA or equivalent early warning detection, maintained fire extinguishers at designated points, fire-rated compartmentalisation between the white space and other building areas, emergency evacuation plans approved by a fire safety officer, and compliance with the Kenya Fire Brigade Act and relevant county by-laws. Insurers may also mandate NFPA 75 or NFPA 76 compliance as a policy condition. Failure to maintain these systems can result in policy voidance or significantly higher premiums. ### How much does a clean agent fire suppression system cost in Kenya? A complete clean agent fire suppression system for a Kenyan data centre typically costs between KES 40 million and KES 150 million or more, depending on the size of the protected area, the agent selected (Novec 1230 is approximately 30–50% more expensive than FM-200), and the complexity of the detection and control systems. This is 5 to 10 times more expensive than a conventional water sprinkler system. However, when weighed against the potential loss of equipment valued at hundreds of millions of shillings (and the reputational damage from a major outage) clean agent systems are considered essential for any facility housing customer infrastructure. *For authoritative guidance on data centre fire protection standards, refer to the [NFPA 75: Standard for the Fire Protection of Information Technology Equipment](https://www.nfpa.org/75) and the [Kigali Amendment to the Montreal Protocol](https://www.unep.org/montreal-protocol/kigali-amendment) for environmental phase-down schedules affecting HFC-based suppression agents.* ## Geothermal Energy in Kenya: Power for Data Centres URL: https://data-centers-254.vercel.app/articles/geothermal-energy-kenya-data-centres Cluster: Renewable Energy Published: 2026-08-28 Updated: 2026-09-28 Summary: Kenya draws about 45% of its electricity from geothermal energy, giving data centres some of the world's lowest-carbon power. How the Rift Valley does it. Deep beneath the floor of the Great Rift Valley, water seeping into fractures in the Earth's crust encounters rock heated by magma chambers lying only a few kilometres below the surface. This water becomes superheated (reaching temperatures of 200–350 degrees Celsius) and rises to the surface as steam through geological faults. For millions of years, this process occurred unseen and unutilised. Today, it powers approximately 45% of Kenya's electricity and, through the national grid, powers every data centre in Nairobi. ![Geothermal pipelines and steam wellheads](/images/kenya-geothermal-plant-3.webp) Geothermal energy is Kenya's most important energy asset for data centres, and it is an advantage that no amount of solar panels, wind turbines, or policy incentives can fully replicate in other markets. Understanding how geothermal power works, why it is so well-suited to data centre loads, and what its limitations are, is essential for anyone evaluating Kenya as a data centre location or investment destination. ## Kenya's Geothermal Resources Kenya sits astride the East African Rift System, a tectonic plate boundary that extends from the Afar Triangle in the northeast through the Rift Valley to Mozambique in the south. This geological feature, which is slowly splitting the African continent apart, creates the conditions for geothermal energy by bringing heat close to the Earth's surface and providing pathways for water to circulate through hot rock. Kenya's geothermal potential is estimated at 7,000 to 10,000 megawatts, enough to power the entire country several times over. Only a fraction of this potential has been developed. The country's current installed geothermal capacity is approximately 950 megawatts, generated primarily from the Olkaria geothermal complex in Naivasha, about 120 kilometres northwest of Nairobi. Additional development is ongoing at Menengai (near Nakuru), Suswa, and other Rift Valley sites. The Olkaria complex is the largest geothermal installation in Africa. It includes multiple power stations (Olkaria I, II, III, IV, and V) operated by the Kenya Electricity Generating Company (KenGen), with a combined capacity of approximately 800 megawatts. The Geothermal Development Company (GDC), a state-owned enterprise, was established to develop geothermal resources and provide steam to independent power producers, broadening the investment base beyond KenGen. ## How Geothermal Power Reaches Data Centres Geothermal power does not flow directly from Olkaria to data centres. The journey involves several steps. ### Generation At the geothermal field, production wells (typically drilled to depths of 1,500–3,000 metres) tap into the geothermal reservoir. Steam and hot water are brought to the surface through these wells and directed to a power plant. In a typical geothermal power plant, the steam drives a turbine connected to a generator, producing electricity. After passing through the turbine, the steam is condensed back into water and reinjected into the geothermal reservoir through injection wells, maintaining the reservoir's pressure and sustainability. ### Transmission ![Transmission lines carrying power toward Nairobi](/images/kenya-transmission-pylons-3.webp) The electricity generated at Olkaria enters the national grid through high-voltage transmission lines operated by Kenya Electricity Transmission Company (KETRACO). The primary transmission corridor from Olkaria to Nairobi runs through Naivasha, Limuru, and into the city, where it connects to the distribution network operated by Kenya Power. This transmission infrastructure was originally built to serve general national demand, but the growth of data centre loads along Nairobi's Mombasa Road corridor has created concentrated demand nodes that benefit from this geothermal backbone. ### Distribution Kenya Power distributes the electricity from the transmission network to end customers through its distribution network. Data centres, as large industrial consumers, typically receive power at high voltage (11kV or 33kV) through dedicated feeders, which provide better power quality and reliability than the shared low-voltage feeders serving residential and small commercial customers. ## Why Geothermal Is Ideal for Data Centres ### Baseload Reliability The most important characteristic of geothermal energy for data centres is its reliability as a baseload power source. Unlike solar panels that generate electricity only when the sun shines, or wind turbines that generate only when the wind blows, geothermal power plants generate electricity 24 hours a day, 365 days a year. Geothermal plants in Kenya achieve capacity factors of 90–95%, meaning they produce electricity at or near their rated capacity almost all the time. This baseload reliability aligns perfectly with data centre loads, which are constant and unrelenting. A 1-megawatt data centre draws approximately 1 megawatt of IT power (plus cooling overhead) every hour of every day. It does not reduce its power consumption at night or on cloudy days. A power source that matches this constant demand profile (without the need for battery storage or backup generation to bridge gaps) is inherently more valuable for data centres than intermittent sources. ### Low Carbon Intensity ![Steam stacks at a Kenyan geothermal plant](/images/kenya-geothermal-plant-5.webp) Geothermal energy produces approximately 38 grams of CO2 per kilowatt-hour of electricity generated. This is among the lowest carbon intensities of any power source, including other renewables (solar PV produces approximately 40–50 g/kWh over its lifecycle, wind produces 10–15 g/kWh). For comparison, coal-fired power produces 900–1,100 g/kWh, and natural gas produces 400–500 g/kWh. Kenya's overall grid carbon intensity, dominated by geothermal and hydro, is approximately 50–100 g/kWh. This means that a data centre running on Kenya's grid is already operating with a carbon footprint that is 5–10 times lower than a comparable facility in a coal-dependent market like South Africa or India. This is not a trivial advantage, it is a genuine, measurable, and marketable sustainability credential that Kenyan data centres can leverage. ### Cost Stability Geothermal power has another advantage that is often overlooked: cost stability. Once a geothermal plant is built, its ongoing fuel cost is essentially zero, the "fuel" is heat from the Earth's interior, which is freely and perpetually available. This means that the cost of geothermal electricity is driven primarily by capital costs (drilling wells, building power plants) and operating costs (maintenance, staffing), which are relatively predictable and stable over time. This stability contrasts sharply with fossil fuel-dependent power systems, where electricity costs fluctuate with global oil, gas, and coal prices. A data centre operator in Kenya can forecast its power costs with more confidence than one in Nigeria (where diesel costs vary with global oil prices) or South Africa (where Eskom's tariff increases are unpredictable). ## The Geothermal-to-Data-Centre Direct Connection One of the most interesting concepts in Kenyan data centre energy is the possibility of building data centres directly at or near geothermal fields, bypassing the transmission and distribution network and accessing power at or near generation cost. This idea has been discussed in industry circles and has genuine technical and economic merit. A data centre at the Olkaria geothermal complex would have several advantages. Power costs could be 30–40% lower than in Nairobi (bypassing transmission losses and distribution margins). The facility would have direct access to a renewable baseload source. And the cooler climate of the Rift Valley floor (Olkaria sits at approximately 2,000 metres altitude) could provide some natural cooling benefit. The challenges are equally significant. Olkaria is 120 kilometres from Nairobi, the primary market for data centre services. Connectivity would require building or extending high-capacity fibre links to a location that currently does not have the dense metro fibre network available in Nairobi. The skilled workforce required for 24/7 data centre operations is concentrated in Nairobi. And the data centre would need to attract customers willing to accept the latency and accessibility trade-offs of a remote location. For certain use cases, particularly AI training clusters that need massive amounts of cheap power but do not need low-latency connectivity to end users, a Rift Valley data centre could be compelling. For general-purpose colocation serving Nairobi-based enterprises, the location is less practical. The concept remains an opportunity for the future, potentially as part of the Konza Technopolis development or as a specialised facility for compute-intensive workloads. ## Geothermal Development Pipeline ![Map of Kenya's Rift Valley geothermal sites](/images/kenya-geothermal-map.webp) Kenya's geothermal development is ongoing, with several projects that will increase capacity and reinforce the power supply available to data centres. The Menengai geothermal project, located near Nakuru, is being developed by GDC and is expected to add approximately 160 megawatts of capacity. Independent power producers (IPPs) are being invited to build power plants at the Menengai field using steam provided by GDC, creating a model where the state develops the resource and private companies build and operate the power plants. Additional prospects at Suswa, Baringo, Silali, and other Rift Valley sites are at various stages of exploration and development. The Kenyan government's target is to increase geothermal capacity to approximately 1,600 megawatts by 2030, which would further strengthen the renewable baseload that data centres depend on. ## The Competitive Advantage Kenya's geothermal resource is not just an energy supply, it is a competitive advantage in the global data centre market. As environmental sustainability becomes a criterion for data centre site selection, as ESG reporting requirements drive companies to reduce their carbon footprint, and as customers (particularly international enterprises) demand green infrastructure, Kenya's geothermal grid becomes a marketable asset. No amount of on-site solar panels at a data centre in South Africa can match the carbon intensity of running on Kenya's geothermal-dominated grid. No amount of corporate renewable energy certificates can substitute for the measurable, verifiable reality of drawing power from a grid that is 90%+ renewable. This is a structural advantage that will become more valuable over time, and Kenyan data centre operators should be communicating it aggressively to domestic and international customers. National policy has caught up with this argument. The Principal Secretary for ICT and the Digital Economy, writing in September 2026, made additional geothermal capacity for green computing an explicit national ask: dedicated renewable megawatts along the Naivasha, Nakuru and Baringo corridor, built new for computing rather than reallocated from households and existing industry. His essay, fact-checked in [Kenya's data centre and compute hub plan, explained](/articles/kenya-data-centre-compute-hub-plan-explained), also cites the geothermal corridor as the anchor of the Microsoft and G42 discussions near Naivasha, evidence that green baseload power is now the opening card in Kenya's hyperscale pitches. The Rift Valley's heat, rising through fractures in the Earth's crust to turn turbines and generate electricity, is an unlikely hero in Kenya's digital economy story. But it is the foundation on which the country's data centre industry is built, and it is the reason that Kenya can credibly position itself as home to Africa's greenest data centres. ## GPU Cloud Infrastructure in Kenya: A Simple Guide URL: https://data-centers-254.vercel.app/articles/gpu-cloud-infrastructure-kenya Cluster: AI & Infrastructure Published: 2026-09-16 Updated: 2026-09-16 Summary: Why Kenya needs local GPU power for AI, what the law says about keeping data home, and which facilities are actually built for it, in plain English. Ask what powers artificial intelligence and most people picture code, or maybe robots. The honest answer is more industrial: AI is powered by electricity pushed through a very particular kind of chip, in buildings engineered to keep those chips cool, in quantities Kenya has only recently begun to host at home. This guide explains what GPU cloud infrastructure is, why it is suddenly a Kenyan topic (the law and the market both moved), and what actually exists on the ground as of September 2026. We are writing this for the newcomer. If you have ever wondered what people mean when they say a facility is AI-ready, or why anyone would train a model in Nairobi instead of on some American cloud, the next few sections are for you. ![A cluster of GPU servers in a data centre rack](/images/dc-gpu-cluster.webp) ## First, the chip: why GPUs and not ordinary servers A normal computer processor, a CPU, is like a brilliant chef: superb at cooking one complicated dish at a time, in strict order. A GPU is more like a stadium kitchen with a thousand cooks: not each one brilliant, but all working at once, which is exactly the right shape for AI's maths. Training an AI model (teaching it patterns from examples) boils down to billions of small calculations that can all run in parallel, and GPUs, chips originally built to draw video game graphics, happen to be built for parallel work. This is the core of our earlier explainer on [what GPU computing is and why AI needs it](/articles/gpu-computing-kenya-ai-needs-data-centres), and the phrase to remember is GPU cloud infrastructure: computing capacity rented from facilities that can feed, cool and connect these machines. A single GPU machine is unremarkable. The infrastructure part matters when you need tens or hundreds of them together, because then the building itself, power, cooling, and the wiring between machines, becomes the hard engineering problem. ![AI GPU server systems with dense cabling](/images/ai-gpu-servers.webp) ## Why train in Kenya at all? The law moved first The strongest argument for local GPU infrastructure in Kenya is not speed or pride. It is written down. Kenya's Data Protection Act came into force on 25 November 2019 and remains the primary law on data protection (DLA Piper's Kenya briefing, updated March 2026). The regulator, the Office of the Data Protection Commissioner (ODPC), publishes practical guidance, and its handbook says something every Kenyan AI builder should memorise: organisations can meet localisation requirements by processing personal data through a data centre located in Kenya, or by storing a serving copy of the data in Kenya (ODPC handbook, fetched 16 September 2026). In plain terms, if your AI project touches Kenyan personal data (medical records, mobile money histories, student results, ID numbers), the law expects that data to have a Kenyan home. Sending it abroad for training on a foreign cloud is exactly the pattern the rules push against, and registration with the ODPC is required before processing personal data at all (Manwa Advocates, 3 July 2026). The government's own cloud policy pushes the same direction for public-sector data. This is the concept engineers call data gravity: data is heavy, in a legal and practical sense, and the computing that touches it gets pulled to wherever the data must live. Kenyan personal data now has strong gravity. The pull points at Kenyan facilities. ## What actually exists on the ground So where would a Kenyan team actually rent GPU power today? The honest map, as of September 2026, is short but real, and it starts with one name. **iXAfrica NBOX1**, in Nairobi, is the flagship. The operator announced it as East Africa's first hyperscale, carrier-neutral, AI-ready data centre (February 2025), and trade coverage highlights its high-density GPU hosting capability (Marcopolis, July 2025). The campus's overall design capacity, what it is built to reach, is 22.5MW per the operator (fetched 16 September 2026), with a second campus, NBOX2, planned at Tilisi. Design capacity is a plan, not a switch-on date, so buyers should always ask what is live today. We unpack what hyperscale means in a separate guide. **The full stack is already running, not just the shells.** In November 2025, Atlancis Technologies, Everse Technology and iXAfrica jointly unveiled what the partners describe as East and Central Africa's first GPU-powered AI infrastructure, hosted at iXAfrica's Nairobi campus and built on NVIDIA GPUs (announcement dated 10 November 2025). The distinction matters: a facility that could host GPUs is a shell, while a facility where GPU clusters are installed, powered and cooled is a working AI factory, and Nairobi now has one. Kenyan teams can engage local providers for GPU capacity instead of importing every training run. Beyond iXAfrica, the pattern to know is this: Kenya's operational colocation market (NBO1 on Mombasa Road, the new NBO2, iColo, PAIX, all covered in our [Nairobi colocation buyer's guide](/articles/colocation-data-centres-nairobi-buyers-guide)) provides the shells, and AI-ready capability is arriving as those shells upgrade their power and cooling. The loudest sign of where this is heading is the [Amaco HERCULES proposal](/articles/amaco-hercules-mombasa-ai-data-centre): a $1.5 billion AI data centre plan for Mombasa that would generate its own electricity on a floating power barge, precisely because AI at scale is a power story before it is a software story. ![Close-up of NVIDIA GPU chips mounted on a circuit board](/images/worldbank-nvidia-gpu-board.webp) ## The power problem, in one paragraph Every AI training run is an electricity bill. An AI rack draws many kilowatts, several times what an ordinary server rack needs, and the heat that comes out has to be fought with chillers and airflow that consume power themselves. Kenya's grid is remarkable by world standards (about a gigawatt of geothermal, one of the greenest electricity mixes anywhere), but total national peak demand is around 2,439MW, which is why a single large AI campus can strain a national conversation about power. Our guide to [Kenya's power infrastructure and the data centre constraint](/articles/kenya-power-infrastructure-data-centres) covers that squeeze in detail. The practical takeaway for buyers: when you evaluate any AI-ready claim, ask for the kilowatts per rack the facility can actually deliver today. ## What Kenya would do with local GPUs The point of all this infrastructure is not prestige; it is the work. Three families of use cases come up again and again in the Kenyan market, and our deeper piece on [AI model training use cases in Kenya](/articles/ai-model-training-kenya-use-cases) profiles them properly. - ✔ **Finance:** credit scoring models trained on Kenyan mobile money patterns. They understand the local reality of irregular incomes better than models trained on Western banking data. - ✔ **Agriculture:** computer vision (AI that looks at images) checking crops for disease using photos of Kenyan fields, deployed at the edge close to the farms. - ✔ **Language:** models that speak Kiswahili, Sheng and Kenya's local languages, trained on local text so digital services stop being foreign by default. Each of these runs on Kenyan data. And under the rules above, each one wants Kenyan compute. ![Diagram of the power chain from grid to GPU rack](/images/diagram-power-chain-grid-to-gpu.webp) ## A practical starting path For a Kenyan organisation reading this with a real project in mind, the path is simpler than the hype suggests. If you are experimenting and your data is not sensitive, a global cloud GPU by the hour is fine for learning, if you can pay for it. The moment real Kenyan personal data enters the project, bring it home: talk to the AI-ready facilities about reserved or dedicated capacity, size your need in kilowatts and GPUs honestly, and let the legal requirement do part of your negotiating. Between those two poles, hybrid setups (development abroad, data-bearing training at home) are common and sensible. Kenya spent two decades exporting its digital problems: data abroad, models built elsewhere, value added elsewhere. GPU cloud infrastructure is how that reverses. The chips are arriving, the law is pulling, and the buildings, finally, are standing. ## Frequently asked questions ### What is GPU cloud infrastructure? GPU cloud infrastructure is computing capacity built around GPUs (graphics processing units, chips designed to run thousands of small calculations at once) instead of ordinary processors. AI workloads, like training a model to recognise crop diseases or answer in Swahili, are mostly thousands of small maths problems happening together, which is exactly what GPUs are good at. GPU cloud infrastructure means renting that kind of computing power from facilities that can power and cool these machines. ### Why does Kenya need local GPU hosting? Three reasons. Legally, Kenyan data protection rules expect personal data to be processed inside Kenya or to keep a serving copy in Kenya, which pushes AI work on Kenyan personal data toward local facilities. Technically, a GPU server in Nairobi answers Kenyan users faster than one in Europe because the round trip is shorter. Practically, paying in shillings and building on local infrastructure keeps skills, jobs and value in the country instead of exporting them with every training run. ### Does Kenyan law require data to stay in Kenya? Kenya's Data Protection Act, in force since 25 November 2019, is the primary law, and the regulator's own handbook states that organisations can meet localisation requirements by processing personal data through a data centre located in Kenya, or by storing a serving copy of the data in Kenya. Every data controller and processor must also register with the Office of the Data Protection Commissioner before processing personal data. The exact duty depends on your sector and data type, so get legal advice for your case, but the direction is clear: Kenyan personal data is expected to have a Kenyan home. ### Can I rent GPUs by the hour in Kenya today? The by-the-hour model familiar from global clouds is still mostly abroad. Inside Kenya, AI compute today comes mainly through AI-ready facilities: iXAfrica's NBOX1 campus in Nairobi is explicitly built and marketed for high-density GPU hosting with a 22.5MW design capacity, and since November 2025 the campus hosts a GPU-powered AI infrastructure service unveiled by Atlancis Technologies, Everse Technology and iXAfrica, built on NVIDIA GPUs. Similar capacity is emerging as new facilities launch. For most Kenyan teams that means engaging a local provider for dedicated or reserved capacity rather than swiping a card for one hour of training. Global clouds remain the option for small experiments, if you have the payment method and accept the data leaving the country. ### What makes a data centre AI-ready? Three upgrades over a normal facility. Power density: AI racks draw many kilowatts each, several times an ordinary server rack, so floors, breakers and cooling must be engineered for the heat. Networking: training clusters pass enormous amounts of data between machines, so the cabling between racks matters as much as the internet connection. Space and weight: GPU machines are heavier and deeper than standard servers. A facility advertising AI-ready today, like iXAfrica's Nairobi campus, is claiming all three. ## GPU Computing in Kenya: Why AI Needs Data Centres URL: https://data-centers-254.vercel.app/articles/gpu-computing-kenya-ai-needs-data-centres Cluster: AI & Infrastructure Published: 2026-08-28 Updated: 2026-09-18 Summary: AI workloads demand GPU-dense data centres at 20-40kW per rack. Kenya's emerging GPU infrastructure and the use cases driving demand. Artificial intelligence does not exist in the cloud. It exists on physical servers, in physical data centres, consuming physical electricity and generating physical heat. Every time you use ChatGPT, every time a bank's fraud detection system flags a suspicious M-Pesa transaction, every time an AI model translates a document from Swahili to English, a GPU somewhere is performing billions of calculations per second. The gap between the promise of AI and the reality of running it is bridged by data centre infrastructure, and in Kenya, that bridge is still being built. ![Server racks in a Nairobi data centre hall](/images/dc-gpu-cluster-2.webp) GPU computing represents a fundamental shift in what data centres must be designed to handle. Traditional data centres were built for CPU workloads (web servers, databases, virtual machines) where a typical rack consumes 5–10 kilowatts of power. GPU computing pushes rack power densities to 20, 40, or even 100 kilowatts. This changes everything: the power infrastructure, the cooling systems, the cabling, the floor loading, and the economics of running the facility. Kenya's data centre industry, still in its early growth phase, must reckon with this shift or risk being left behind as AI becomes the primary driver of new data centre demand globally. ## Why AI Needs GPUs To understand the infrastructure challenge, it helps to understand why AI needs GPUs in the first place. AI models, the large language models behind ChatGPT, the computer vision models behind autonomous vehicles, the recommendation systems behind every streaming service, are trained on massive datasets using a technique called deep learning. Deep learning involves performing billions or trillions of simple mathematical operations (primarily matrix multiplications) in parallel. Central processing units (CPUs), the general-purpose processors in traditional servers, are designed to handle a wide variety of tasks sequentially. They are good at doing one complex thing at a time. GPUs, originally designed for rendering graphics (where millions of pixels must be calculated simultaneously), are designed to do thousands of simpler things at the same time. For deep learning, this parallelism makes GPUs 10 to 100 times faster than equivalent CPUs. ![NVIDIA GPU accelerator cards mounted on a server motherboard, server racks behind](/images/nvidia-gpu-cards-board.webp) The scale of modern AI is staggering. GPT-4, for example, was trained on an estimated 13 trillion tokens of text using approximately 25,000 GPUs over several months. The training run consumed an estimated 50 gigawatt-hours of electricity, enough to power 5,000 Kenyan homes for a year. Even inference (running a trained model to generate responses) requires significant GPU resources, as millions of users make simultaneous requests. ![Illustrative render of an NVIDIA H100 accelerator chip on a circuit board](/images/nvidia-h100-chip-render.webp) ## What GPU Computing Demands from Data Centres ### Power Density The most obvious impact of GPU computing on data centres is power density. A traditional server rack with 20–40 CPU servers might consume 5–10 kilowatts. A rack with 4–8 NVIDIA H100 GPU servers can consume 30–60 kilowatts. The latest NVIDIA Blackwell B200 GPUs, released in 2024, push single-rack power beyond 100 kilowatts when fully populated. ![CRAC cooling units behind server racks](/images/dc-cooling-crac.webp) This power density has cascading effects on every aspect of data centre design. Standard power distribution units (PDUs) rated for 20–30kW per rack must be replaced with higher-capacity units. Electrical cable sizes must increase to carry more current without excessive voltage drop. Floor loading (the weight per square metre that the raised floor can support) must be designed for heavier transformers and switchgear. And the total power demand of the facility increases, requiring larger transformer connections from Kenya Power and larger generator capacity for backup. ![UPS units and switchgear in a data centre power room](/images/ups-switchgear-power-room.webp) ### Cooling Cooling is where GPU computing creates the most significant engineering challenge. Traditional data centre cooling uses cold air blown from computer room air conditioning (CRAC) units through the raised floor and into the server racks. This air-cooling approach works well at 5–10kW per rack but becomes increasingly inefficient and eventually impractical at higher densities. ![Data centre cooling and power distribution equipment](/images/dc-cooling-crac-2.webp) At 20kW per rack, air cooling requires very high airflow volumes and very cold supply air, which increases energy consumption and reduces cooling efficiency. At 40kW per rack, air cooling is at the practical limit of what is achievable. Beyond 40kW, liquid cooling becomes necessary. Liquid cooling for GPU data centres takes several forms. Direct-to-chip cooling circulates cold liquid through cold plates mounted directly on the GPU processors, removing heat at the source with much higher efficiency than air. Immersion cooling submerges entire servers in a dielectric fluid that absorbs heat directly from all components. Both approaches can handle 50–100kW per rack and dramatically reduce the overall cooling energy required. For Kenyan data centres, liquid cooling represents both a challenge and an opportunity. The challenge is that it requires different facility design, different skills, and higher capital investment than air-cooled facilities. The opportunity is that liquid cooling is more energy-efficient, which reduces operating costs and aligns with the sustainability positioning that Kenyan data centres are building. ### Connectivity GPU computing also has specific networking requirements. AI training clusters require high-bandwidth, low-latency interconnects between GPU servers, because training large models requires distributing the computation across many GPUs that must communicate frequently. NVIDIA's InfiniBand and NVLink technologies provide these high-speed interconnects, with bandwidths of 400–800 Gbps between servers. ![NVIDIA logo on the company headquarters building](/images/nvidia-hq-building.webp) This networking requirement affects data centre design in several ways. The cabling between GPU servers must support these high-speed interconnects, using specialised optical cables and switches. The network topology within a GPU cluster is different from a traditional data centre network, with spine-leaf or fat-tree topologies optimised for east-west (server-to-server) traffic rather than north-south (server-to-internet) traffic. ## Kenya's Emerging GPU Landscape Kenya does not yet have a purpose-built GPU data centre, but the building blocks are being put in place. ![Consumer GeForce graphics card](/images/dc-gpu-cluster-5.webp) ### Research and Academic Institutions The University of Nairobi, Strathmore University, and the Kenya Medical Research Institute (KEMRI) operate small GPU clusters for research purposes. These are typically a handful of GPU servers, often housed in general-purpose IT facilities rather than dedicated data centres. They serve important research functions (training AI models for healthcare diagnostics, agricultural analysis, and natural language processing) but they are not designed for commercial AI services. ### The Microsoft Africa Development Centre Microsoft's Africa Development Centre (ADC) in Nairobi employs hundreds of engineers working on global Microsoft products, including Azure AI services. While the ADC does not publicly disclose its local GPU infrastructure, its work on AI for African languages and African markets likely involves local GPU resources, whether on-premises or in Azure regions. ### AI Startups Kenya's startup ecosystem includes several AI-focused companies working on natural language processing (building models that understand Swahili, Sheng, and other Kenyan languages), computer vision (agricultural monitoring, security), and fintech AI (credit scoring, fraud detection). These startups typically use cloud GPU services (AWS, Azure, Google Cloud) rather than local infrastructure, paying per-hour rates for GPU instances. As these startups grow and their GPU needs become more sustained and predictable, the economic case for local GPU hosting in Kenyan data centres strengthens. ### Data Centre Operators The most significant development is the planning and design of GPU-capable zones within Kenyan colocation facilities. iXAfrica, in the design of its NBOX1.1 expansion, has included provisions for high-density zones that can support GPU rack power densities of 20–40kW per rack, with provisions for future liquid cooling upgrades. Africa Data Centres, with its pan-African scale, has been deploying GPU-ready infrastructure in South Africa and can bring that expertise to Kenya as demand materialises. ![Secured server racks with locking cabinets](/images/dc-gpu-cluster-2-wide.webp) ## The Economic Opportunity The economic opportunity for GPU computing in Kenya is driven by three factors. First, Africa has 1.4 billion people and 2,000+ languages, most of which are underserved by current AI models that are primarily trained on English and European language data. Building AI for African languages and African use cases requires GPU infrastructure in Africa, not just access to overseas clouds. Second, data sovereignty requirements and latency constraints mean that certain AI workloads (particularly those involving government data, financial transactions, or real-time applications) must run within Kenya's borders. This creates a floor of domestic demand for GPU infrastructure. Third, Kenya's competitive advantages in renewable energy (geothermal power at $0.07–0.09/kWh), connectivity (four submarine cables), and strategic location (serving East Africa's 300 million people) make it a natural location for AI infrastructure serving the region. The challenge is that GPU data centres are expensive to build and require specialised expertise that is still developing in Kenya. A dedicated GPU zone with liquid cooling, high-density power, and InfiniBand networking might cost 30–50% more per rack to build than a traditional air-cooled zone. The question is whether demand will materialise fast enough to justify this investment. ## What This Means for East Africa GPU computing is not just a Kenya story, it is an East African opportunity. Kenya's data centres serve the entire East African Community and beyond. If Kenya builds GPU-capable infrastructure, it becomes the AI processing hub for a region of 300+ million people. Developers in Tanzania, Uganda, Rwanda, and Ethiopia could access GPU resources in Nairobi with sub-100ms latency, far better than connecting to South Africa or Europe. This positions Kenya as the AI infrastructure capital of East Africa, in the same way that it has become the data centre capital. The countries that invest in AI infrastructure early will attract the AI talent, the AI startups, and the AI investment that will define the next decade of digital economic growth. For Kenya, the opportunity is real, the timing is right, and the question is not whether to build GPU infrastructure, but how fast. For where that infrastructure stands right now, our [GPU cloud infrastructure in Kenya guide](/articles/gpu-cloud-infrastructure-kenya) maps the AI-ready facilities, the data protection rules pulling training work onshore, and a practical starting path for Kenyan teams. ## How to Get a Job in a Kenyan Data Centre URL: https://data-centers-254.vercel.app/articles/how-to-get-job-kenyan-data-centre Cluster: Career Guide Published: 2026-08-28 Updated: 2026-08-28 Summary: The certifications, skills, salary ranges, and hiring pathways for data centre careers in Kenya, from network engineer to facility manager. Kenya's data centre industry is growing fast, and it needs people. Not just any people, it needs network engineers who can configure BGP on a router at 3 AM during an outage, facility managers who understand both HVAC systems and power distribution, cabling technicians who can terminate 96 fibre strands in a day, and security operators who can manage access control for a facility housing billions of shillings worth of equipment. The demand for these skills is real, it is growing, and the supply is not keeping up. This guide is for anyone in Kenya who wants to work in a data centre, whether you are a fresh graduate trying to figure out where to start, an IT professional looking to transition from enterprise IT into infrastructure, or an engineer in a related field (power, cooling, telecommunications) who sees the opportunity. We will cover the roles that exist, the certifications employers actually look for, realistic salary expectations, and practical steps to get your foot in the door. ## The Roles: What Jobs Actually Exist Data centres employ a surprisingly diverse range of professionals. The common perception is that data centre jobs are all about servers and networking, but the reality is that a modern facility requires expertise across multiple disciplines. Here are the primary roles you will find in Kenyan data centres, grouped by function. ### Network and Systems Engineering Network engineers are the backbone of data centre operations. They design, configure, and maintain the network infrastructure that connects servers to each other, to the internet, and to customers. In a Kenyan data centre, a network engineer's daily work might involve configuring BGP peering with Safaricom and Liquid Telecom, troubleshooting fibre connectivity issues, managing VLANs for different customers, and ensuring that the network monitoring systems are functioning correctly. The most in-demand networking certification in Kenyan data centres is the Cisco CCNA (Cisco Certified Network Associate). While not every data centre uses Cisco equipment exclusively (Juniper, Arista, and Huawei gear is also common) the CCNA demonstrates a solid understanding of networking fundamentals (IP addressing, routing, switching, VLANs, OSPF, BGP) that transfers across vendors. For more senior roles, the CCNP (Cisco Certified Network Professional) or vendor-specific certifications in data centre networking (like the Cisco CCNP Data Centre) are valued. Systems administrators manage the server infrastructure, operating systems, virtualisation platforms (VMware, Proxmox, KVM), container orchestration (Kubernetes), and storage systems. These roles require a blend of Linux system administration skills (command line proficiency, shell scripting, configuration management with Ansible or Puppet) and understanding of virtualisation and cloud platforms. Relevant certifications include CompTIA Server+, Red Hat Certified Engineer (RHCE), and Microsoft Azure Administrator (AZ-104). ### Facility Operations and Engineering Facility operations is the discipline that most people outside the industry do not know exists, yet it is arguably more critical to a data centre's survival than any individual server or network device. Facility engineers are responsible for the physical infrastructure that keeps servers running: power distribution, cooling systems, fire suppression, and building management systems. A facility engineer in a Kenyan data centre needs to understand electrical engineering (how power flows from the Kenya Power transformer through UPS systems and PDUs to the rack), mechanical engineering (how chillers, CRAC units, and air handling units maintain temperature and humidity), and control systems (how the building management system coordinates everything). This is a specialised skill set that combines elements of electrical, mechanical, and controls engineering, and it is in critically short supply in Kenya. The gold-standard certifications for facility roles come from the Uptime Institute (ATD (Accredited Tier Designer, and AOS) Accredited Operations Specialist) and from EPI (CDCP, Certified Data Centre Professional). These certifications are recognised globally and demonstrate that the holder understands the specific requirements of data centre facility design and operations. They are not widely available in Kenya (most candidates travel to South Africa, Dubai, or Europe for training) which means holders of these certifications are scarce and highly valued. ### Network Operations Centre (NOC) Staff The NOC is the nerve centre of a data centre. Staffed 24 hours a day, 7 days a week, the NOC team monitors the entire facility (power systems, cooling, network connectivity, security cameras, and environmental sensors) and responds to alarms, outages, and customer requests. NOC operators are often the first point of contact when something goes wrong, and their ability to quickly diagnose and escalate issues directly impacts the facility's uptime record. ![Blue-lit server racks in a data hall](/images/hero-server-hall.webp) NOC roles are excellent entry points into the data centre industry. The technical barrier to entry is lower than for engineering roles, a diploma in IT or telecommunications, plus basic networking knowledge (TCP/IP, DNS, HTTP), can be sufficient for a junior NOC position. The trade-off is that NOC roles involve shift work, including nights and weekends, which not everyone can sustain long-term. However, many senior data centre professionals started their careers in NOC roles and used the experience as a foundation for engineering or management positions. ### Cabling and Physical Infrastructure Technicians Structured cabling technicians install, maintain, and repair the physical cable plant within a data centre, copper patch cables, fibre optic cables, and the cable management systems that keep thousands of connections organised and traceable. This might seem like a low-skill role, but professional cabling in a data centre environment requires precision, knowledge of fibre optic safety, and the ability to work in environments with strict contamination controls (clean rooms and controlled access areas). Cabling certifications from vendors like CommScope, Corning, or Panduit, or from industry bodies like the Fiber Optic Association (FOA), demonstrate competence and can differentiate candidates. In Kenya's market, where the skills gap is particularly acute in hands-on infrastructure roles, a certified cabling technician with demonstrated fibre termination skills can be immediately employable. ### Sales, Business Development, and Operations Not every data centre role is technical. Sales and business development professionals are responsible for acquiring new customers (banks, telecom companies, cloud providers, government agencies) and growing revenue. These roles require understanding of data centre concepts (enough to have credible conversations with technical buyers), strong relationship skills, and knowledge of the Kenyan enterprise market. Data centre sales roles in Kenya typically offer base salaries of KES 150,000–300,000 plus commission, with top performers earning significantly more. ## Certification Roadmap: Where to Start For someone entering the data centre industry in Kenya, the certification landscape can be confusing. Here is a practical roadmap, organised by career path and budget. ### Budget-Conscious Path (Under KES 50,000 total) For entry-level candidates with limited budgets, the best starting certifications are CompTIA A+ (general IT hardware and software, approximately KES 20,000 for training and exam), CompTIA Network+ (networking fundamentals, approximately KES 25,000), and free online learning through Cisco Networking Academy (which provides CCNA-level course content at no cost, though the exam itself costs approximately KES 30,000). These certifications demonstrate baseline competence and are recognised by Kenyan employers. ### Mid-Level Path (KES 50,000–200,000) For IT professionals with some experience who want to specialise in data centre infrastructure, the next tier includes CCNA (approximately KES 30,000–50,000 for training and exam), CompTIA Server+ (approximately KES 35,000), and ITIL 4 Foundation (approximately KES 30,000). These certifications, combined with a degree or diploma and some hands-on experience, qualify candidates for mid-level engineering roles in data centres. ### Specialist Path (KES 200,000–500,000+) For professionals targeting senior or specialised roles, the investment is higher but the returns are proportionate. The Uptime Institute ATD or AOS certification (approximately KES 300,000–500,000 including travel to a training centre) is the gold standard for facility engineering roles. The CDCP from EPI (approximately KES 200,000–300,000) is the most widely recognised data centre-specific certification in Africa. CCNP Data Centre (approximately KES 100,000–150,000) is valued for senior networking roles. ![Comparison of network engineer and cybersecurity engineer roles](/images/diagram-network-vs-cyber-career.webp) ## Salary Ranges in Kenya's Data Centre Industry Transparency on salaries in Kenya's data centre industry is limited, as it is in most sectors. Based on job postings, industry conversations, and data from recruitment firms, here are realistic salary ranges for common roles in Nairobi-based data centres. These ranges reflect total monthly compensation including basic salary and any standard allowances, but not performance bonuses or overtime. | Role | Experience | Monthly Range (KES) | Annual Equivalent (USD) | |------|-----------|---------------------|------------------------| | Remote Hands Technician | 0–2 years | 60,000–100,000 | 5,000–8,000 | | NOC Operator | 1–3 years | 80,000–150,000 | 6,500–12,000 | | Cabling Technician | 2–5 years | 70,000–130,000 | 5,500–10,500 | | Network Engineer | 2–5 years | 120,000–250,000 | 10,000–20,000 | | Systems Administrator | 2–5 years | 120,000–220,000 | 10,000–18,000 | | Senior Network Engineer | 5–10 years | 200,000–400,000 | 16,000–32,000 | | Facility Engineer | 3–7 years | 150,000–350,000 | 12,000–28,000 | | Facility Manager | 7–15 years | 300,000–600,000 | 24,000–48,000 | Several factors influence where an individual falls within these ranges. International operators like iXAfrica and Africa Data Centres typically pay 15–25% more than local operators. Roles requiring shift work (NOC, security) often include shift allowances of 10–20%. Certifications, particularly Uptime Institute and CDCP, can push compensation toward the higher end of the range. ## Where the Jobs Are Kenya's data centre industry is concentrated in Nairobi, with the majority of facilities located along three corridors: Mombasa Road (Enterprise Road and surrounding industrial areas), Westlands, and the Sameer Business Park area. The largest employers in the sector are iXAfrica, Africa Data Centres, and Safaricom, which together account for the majority of data centre jobs in the country. ![iXAfrica and Moringa School representatives signing a partnership](/images/moringa-ixafrica-training-signing.webp) To find job openings, monitor these channels: LinkedIn (follow the operators and set job alerts for "data centre", "NOC", "network engineer", "facility manager" in Nairobi), the operators' own careers pages, and Kenyan job boards like BrighterMonday and Fuzu. Industry events organised by TESPOK (Technology Service Providers of Kenya), the Kenya ICT and Innovation Forum, and data centre-specific conferences are also valuable for networking and discovering unadvertised positions. ![Westlands commercial district in Nairobi](/images/nairobi-westlands.webp) ## Practical Steps to Get Started If you are reading this and want to work in a Kenyan data centre, here is a concrete action plan. First, identify which career path suits your background (networking, systems, facility, cabling, or sales) and research the certifications required. Second, start learning: Cisco Networking Academy offers free CCNA-level courses, and platforms like Coursera and Udemy have affordable data centre-specific courses. Third, build hands-on experience through homelabs (running your own servers, networking equipment, or virtualisation platforms at home), volunteer work, or internships. Fourth, network: attend industry events, join relevant LinkedIn groups, and connect with people already working in the industry. Fifth, apply strategically, target the operators directly rather than only responding to job postings, and tailor your CV to highlight relevant skills and certifications. The data centre industry in Kenya is small enough that reputation and relationships matter, but growing fast enough that opportunities are genuine and increasing. The skills gap is real, and employers are actively looking for qualified candidates. The path is not easy (it requires investment in certifications, willingness to work shifts, and continuous learning) but for those who commit, the career trajectory and compensation are among the best in Kenya's technology sector. ## IBTC's Data Centre Academy: Kenya's Talent Pipeline URL: https://data-centers-254.vercel.app/articles/ibtc-data-centre-academy-kenya-talent-pipeline Cluster: Careers Published: 2026-09-24 Updated: 2026-09-24 Summary: IBTC's Data Centre Academy signed iXAfrica and IMEX in Nairobi. How the EU-backed source-train-place pipeline works, who pays, and what graduates get. At ITW & Datacloud Africa in Nairobi on 9 September 2026, a quiet signing took place that says more about the region's data centre future than most megawatt announcements. The IBTC Data Centre Academy formally expanded into Kenya, signing iXAfrica Data Centre and IMEX as East African delivery partners alongside the Africa Data Centres Association (ADCA), Schneider Electric University and DC Elite. The significance is simple. Africa's data centre industry has a demand problem it talks about constantly, but the constraint it complains about most in private is people. This is the first structured attempt to build the bridge from Kenyan engineering graduates into data centre careers, anchored in a live hyperscale Nairobi facility. ## What the IBTC Data Centre Academy actually is IBTC (International Business Training College) is a South African tuition provider that has offered globally recognised qualifications since 2006. Its Data Centre Academy, launched in April 2026, is the continent's flagship workforce-development programme for the industry, built on a "source, train, place" model. The model has three moving parts: internationally recognised certifications, structured online learning, and a three-week internship on a live data centre floor. Certification content comes from Schneider Electric University, with EPI as the certification partner. IBTC demonstrated the model before exporting it. It sponsored 30 recently graduated engineers through the inaugural Schneider Electric University Data Centre Certified Associate programme, and it has separately trained and certified an entire operator workforce in Qatar to set a baseline standard across live sites. Rack Centre in Lagos was the first operator to join, sponsoring candidates and hosting internships in the pilot phase. "This initiative creates a practical pathway into the industry, and we're proud to be part of building that future," said Lars Johannisson, Rack Centre's CEO, at launch. ## The Kenya signing: who does what ![Engineer studying on a laptop in a data centre environment](/images/dc-woman-engineer-laptop.webp) The Kenya programme formalised at ITW & Datacloud Africa, where IBTC had just presented "Growing the DC Talent Pool - an African Context" on the Datacloud Keynote Stage. Six organisations signed or stood alongside, each with a distinct job. **iXAfrica Data Centre** is the live environment. East Africa's first hyperscale, carrier-neutral operator, headquartered on Mombasa Road and majority-owned by Helios Investment Partners since December 2022, it has nominated the initial cohort: five technical staff, each taking a three-week practical placement on live equipment. **IMEX (Imexolutions Ltd)** supplies the hands-on component. The Nairobi-based electrical engineering firm, established in 2009, is East Africa's only certified Schneider Electric EcoStruxure PME and Critical Power EcoXpert partner. Its gear is what trainees will actually touch. "Critical power systems can't be learned from a textbook," said Shinu Vattathara, IMEX's Managing Director. "Giving this cohort hands-on time on real OEM equipment is exactly the gap this programme closes." **ADCA** convenes industry governance through the programme's Steering Committee, keeping training aligned with what operators actually need. **DC Elite** manages the recruitment and placement pipeline that carries graduates from certification into employer interviews. **Schneider Electric University and EPI/EXIN** provide the curriculum and certifications respectively. ![Data centre technician working at iXAfrica facility](/images/ixafrica-datacentre-technician.webp) ## The certification ladder: DCCA I to CDCA The pathway runs in two stages. DCCA I, built on Schneider Electric University's curriculum, covers infrastructure, power, cooling, controls, security and risk: the full stack a facilities-side engineer needs before touching production systems. DCCA II builds toward the EPI/EXIN CDCA (Certified Data Centre Associate), a globally recognised designation. That portability matters: a Kenyan engineer certified this way carries a credential understood in Frankfurt, Singapore and Johannesburg alike. Steve Santini, Vice President for Secure Power in Sub-Saharan Africa at Schneider Electric, framed the rationale plainly: a sustainable talent pipeline is critical infrastructure for the industry's growth, not a nice-to-have. The design choice that distinguishes this from most training is the placement. "Training on live infrastructure, not just theory, is what makes graduates job-ready," said Snehar Shah, iXAfrica's CEO. "That's the value we bring to this partnership." ## Who pays: the EU's Digital Investment Facility ![Aerial view of iXAfrica's NBOX1 campus in Nairobi](/images/ixafrica-nbox1-campus-aerial.webp) The Kenya expansion anchors the country's participation in the Digital Investment Facility (DIF), an EU Commission co-funded programme. DIF funded the industry research that shaped the Academy, and its mandate is specific: grow the workforce, not merely certify people already in it. The distinction matters. Most data centre training certifies people already working on-site, which shifts talent between employers without growing the pool. DIF was designed to fund the part of the pipeline the commercial market will not: building a bridge from recent engineering and telecoms graduates, who have the aptitude but no visibility into the industry, into careers they did not know existed. "Developing sustainable digital infrastructure requires investment not only in technology, but in people," a DIF representative said at the Academy's launch. For Kenya, the timing is pointed. The country is positioning itself as a green compute destination on the strength of its geothermal grid, and every announced campus generates the same question from investors: who will staff it? ## What it means for Kenyan job-seekers The honest answer today: five people are on the pathway, and the programme is open. IBTC has invited East African operators to certify existing technical staff, sponsor emerging talent, host interns or build structured graduate pipelines. For a graduate, the realistic entry route is through a sponsoring operator or partner rather than open enrolment. That makes the ecosystem partners worth watching, because each new operator that joins widens the door. ![Data centre engineer inspecting a server rack](/images/dc-engineer-rack-inspection.webp) Our guides on [data centre careers in Kenya](/articles/data-centre-careers-kenya) and [how to get a job in a Kenyan data centre](/articles/how-to-get-job-kenyan-data-centre) cover what employers actually screen for today. The [certifications that matter in Kenya](/articles/data-centre-certifications-kenya) sit neatly alongside DCCA I, and [career progression patterns](/articles/career-progression-african-data-centres) show where a facilities-side path leads after entry. ## What to watch Three markers will tell you whether this becomes a pipeline or a press release. Watch the cohort size: five is a pilot, and scaling to successive intakes across iXAfrica's NBOX phases is the first proof of intent. Watch partner adoption: IBTC has opened the Academy to new East African partners, and every additional operator or OEM that signs widens the funnel the way Rack Centre did in Nigeria. And watch placements: DC Elite's job is to carry graduates into employment, and placement rates, not certificates issued, are the number that will test the "place" in source-train-place. ## ISO 27001 for Data Centres in Kenya: What It Takes URL: https://data-centers-254.vercel.app/articles/iso-27001-data-centre-kenya Cluster: Security Published: 2026-09-28 Updated: 2026-09-28 Summary: What ISO 27001 certification requires from a Kenyan data centre: the ISMS, the audit cycle, the physical controls, and how it maps to the Data Protection Act. ISO 27001 is the international standard for an information security management system (ISMS), and it is the single most requested certificate in enterprise data centre procurement. In Kenya the pattern is clear: the leading operators are either certified or mid-process, because banks, cloud providers and multinationals treat the certificate as a filter. Without it, a facility is effectively excluded from the most valuable conversations. This page explains what the standard actually certifies, what the audit cycle looks like, and how it maps onto Kenyan law. ## What ISO 27001 actually certifies The most common misconception is that ISO 27001 certifies a building. It does not. It certifies a management system: the documented policies, risk assessments, controls, and review cycles through which an organisation manages information security. The same certificate framework applies to a bank, a law firm, or a colocation hall. For a data centre, what gets certified is the operator's systematic approach to identifying, assessing and treating security risks, and the independent audit that confirms the system exists and functions. That distinction matters when you read a certificate. The scope statement on the certificate says exactly which entity, which facility, and which services are covered. "Our group is ISO 27001 certified" and "this facility's operations are in scope" are different claims, and a buyer is entitled to the second one in writing. ## The audit cycle, start to finish Certification runs on a repeating cycle. It begins with a readiness review, where the certification body checks that the ISMS documentation exists and the organisation actually ran its own internal audit and management review. Then comes the certification audit, where auditors test whether the system works as documented, sampling evidence across controls. Once granted, the certificate typically runs on a three-year cycle with annual surveillance audits, and a full recertification at the end of the term. Failure to maintain the system, not just the controls but the reviews and records, is what suspends certificates in practice. For a Kenyan operator, the preparation phase dominates the effort: writing and living the policies, building the risk register, running the internal audit, and collecting evidence. Audit fees are the smaller line item; the organisational work is the real cost, which is why the cost question has no honest single-number answer. ## The controls that bite hardest in a data centre The standard's control set is broad, but for a colocation or managed facility a handful of themes carry most of the weight. Physical and environmental security: perimeter protection, entry control, monitoring, equipment siting, power and cooling resilience. Operations security and logging: who did what, when, with records that survive an audit. Supplier relationships: contractors and remote-hands staff inside the trust boundary. And capacity and continuity management, which overlaps with the facility's core product. These are precisely the controls a buyer should map against the facility tour: if the certificate is real, the [physical security layers](/articles/data-centre-physical-security-kenya), biometric readers, mantraps, CCTV retention, rack locks, and access logs, should be observable in operation, not just in a policy document. ## How it maps onto Kenyan law Kenya does not mandate ISO 27001, but three legal frameworks make it commercially decisive. The Data Protection Act 2019 requires data processors and controllers to implement "appropriate technical and organisational measures" for personal data, and certification is the most widely accepted demonstration of exactly that; the [Kenya Data Protection Act obligations](/articles/kenya-data-protection-act-data-centres) fall directly on data centre operators acting as processors. The Computer Misuse and Cybercrimes Act 2018 makes evidence discipline a legal asset, which aligns with the ISMS logging culture. And the Communications Authority's licensing regime, under which operators hold their [NFP-T2 licence](/articles/kenya-data-centre-licensing-framework), adds regulatory expectations that a documented ISMS makes far easier to answer. The [Policy Intelligence hub](/policy) tracks these instruments as they evolve. ## The failure modes that actually suspend certificates Certificates rarely lapse because a control failed. They lapse because the system around the controls stopped being lived: management reviews that stopped happening, internal audits skipped to save a quarter, risk registers untouched since certification day, and new services launched without going through the ISMS change process. Auditors sample for exactly this, which is why the operational rhythm, regular reviews, current records, evidence of decisions, matters as much as the controls themselves. A Kenyan operator planning certification should budget for that ongoing discipline from the start, not treat it as a one-time project. ## What to verify before you trust the certificate Three checks separate a meaningful certificate from wallpaper. One, scope: confirm the certificate names the operating entity and the facility, not just a corporate parent. Two, accreditation: confirm the certification body itself is accredited, so the audit means what it claims. Three, currency: check the issue date, the surveillance audit history, and the expiry, a certificate in year three without surveillance visits is a risk signal, not a comfort. If you are comparing facilities, our [security explainer](/articles/data-centre-security-explained) shows how certification fits alongside the physical, network and data layers, and the [threat patterns seen in Kenya](/articles/data-centre-security-threats-kenya) show what the controls are actually for. And if you are building a career rather than a facility, the individual training paths live in our [certifications and careers guide](/articles/data-centre-certifications-kenya); facility certification and personal certification are different tracks that are easy to confuse. ## iXAfrica: Kenya's First Hyperscale AI-Ready Facility URL: https://data-centers-254.vercel.app/articles/ixafrica-data-centres-kenya Cluster: Operators Published: 2026-08-29 Updated: 2026-09-20 Summary: iXAfrica operates East Africa's first hyperscale, AI-ready data centre on Mombasa Road, Nairobi. Actis-backed NBOX1 targets 22.5 MW. When the first phase of iXAfrica's NBOX1 facility went live in early 2025, it marked a turning point for East Africa's digital infrastructure. For the first time, the region had a purpose-built, carrier-neutral, hyperscale data centre engineered to handle the power and cooling demands of modern AI workloads. Backed by Actis, one of the world's most active emerging market investors, iXAfrica has moved quickly from a greenfield startup to the largest operational data centre platform in East and Central Africa, serving everyone from Kenyan banks to global satellite operators. ![Aerial view of the iXAfrica NBOX1 data centre campus beside Nairobi's Mombasa Road](/images/ixafrica-nbox1-campus-aerial.webp) This profile examines how iXAfrica was built, what its facilities deliver, who leads it, and why its trajectory matters for Kenya's position as a digital hub for a region of over 300 million people. ## Founding Story and Investment Backing iXAfrica Data Centres was co-founded in 2018 by Guy Willner, a veteran of the global data centre industry whose career stretches back to 1999. Willner previously co-founded IXEurope, a pioneer in European carrier-neutral colocation that he grew from startup to a publicly listed company before its acquisition by Equinix. After decades building data centres across Europe, the Middle East, and Russia, Willner identified East Africa - and Kenya in particular - as the next frontier for hyperscale infrastructure investment. He co-founded iXAfrica with the explicit mission of building the region's first facility capable of serving hyperscale cloud providers and high-density AI compute customers. The company attracted early backing from Helios Investment Partners, which invested $50 million into iXAfrica in 2022. Helios, one of Africa's most prominent private equity firms, brought not just capital but also a continent-wide platform of infrastructure investments. Actis subsequently joined as a significant investor, adding iXAfrica to its growing portfolio of data centre assets across emerging markets. Actis brings deep expertise in building and scaling infrastructure businesses in Africa, and its involvement has strengthened iXAfrica's ability to pursue an aggressive campus expansion strategy. The combination of Willner's operational experience and the financial backing of two leading Africa-focused investors gave iXAfrica a foundation that most entrants to the Kenyan market lack. Rather than retrofitting an existing building or starting with a small pilot, the company designed its campus from the ground up for the workloads it expected to serve five and ten years out. ## NBOX1: The Live Facility on Mombasa Road iXAfrica's first facility, NBOX1, sits on a 4.3-acre campus along Mombasa Road in Nairobi's Industrial Area. This location is not accidental. Mombasa Road is the primary fibre corridor connecting Nairobi to the submarine cable landing stations in Mombasa, including the EASSy, SEACOM, TEAMS, and DARE1 cables. Every major terrestrial fibre route from the coast passes through this corridor on its way to Uganda, Rwanda, Burundi, and the wider Great Lakes region. For a carrier-neutral facility whose value depends on connectivity, this is arguably the best address in East Africa. NBOX1.1, the first phase now in commercial operation, delivers the following specifications: ![Cutaway infographic of an AI data centre's systems](/images/whats-inside-ai-data-center.webp) - **IT power capacity:** 4.5 MW - **Rack capacity:** 780 racks across three data halls - **White space:** 1,653 square metres - **Rack power density:** up to 40 kW per rack (tested to 50 kW per rack) - **Power Utilisation Effectiveness (PUE):** 1.25 - **Meet-Me Rooms (MMRs):** four carrier-neutral MMRs - **Fibre entry points:** four physically diverse and secure fibre entry routes - **Building footprint:** 15,587 square metres across the full NBOX1 campus These specifications place NBOX1 in a different category from every other operational facility in Kenya. The 40 to 50 kW per rack density is the critical differentiator. Standard enterprise data centres in Nairobi typically deliver 3 to 5 kW per rack, sufficient for traditional servers but entirely inadequate for GPU clusters used in AI training and inference. iXAfrica's facility was engineered with solid floors capable of supporting heavier and taller rack configurations, adiabatic cooling systems that consume only 25% of server power for cooling (compared to 100% in older facilities), and the structural headroom to incorporate liquid cooling as GPU densities continue to increase beyond 100 kW per rack. The facility achieved its PUE of 1.25 partly by leveraging Kenya's energy profile. Approximately 93% of Kenya's electricity grid is sourced from renewables, with geothermal energy from the Olkaria geothermal field in the Rift Valley being the single largest contributor. Kenya is the sixth-largest geothermal energy producer globally, a fact that gives its data centres a structural ESG advantage over competitors in markets reliant on coal or gas-fired generation. iXAfrica holds EcoVadis certification, reflecting independent verification of its sustainability practices. ## NBOX1.2 and the Mombasa Road Campus Master Plan The NBOX1 site is designed as a multi-phase campus. The next phase, designated NBOX1.2, will add a further 18 MW of IT load capacity with 3,744 racks on the same Mombasa Road campus. When NBOX1.2 is complete, the total NBOX1 campus will deliver 22.5 MW of IT power, making it by far the largest single-site data centre campus in the Greater East African region. In September 2025, iXAfrica announced it had secured financing for this 20 MW expansion, signalling that the growth phase is moving from planning to construction. The expansion is designed to accommodate the type of large-footprint deployments that hyperscale cloud providers require: dedicated halls, custom build-outs, and the power and cooling infrastructure to support multi-megawatt commitments from a single tenant. The full NBOX1 campus sits on 17,300 square metres of land with a total building footprint of 15,587 square metres, providing significant room for the additional data halls and supporting infrastructure that NBOX1.2 will require. The site's power allocation from the grid is 22.5 MW, ensuring that the full campus build-out is supported by committed utility power rather than speculative capacity. ![An iXAfrica technician standing in front of switchgear cabinets inside the facility's electrical room](/images/ixafrica-datacentre-technician.webp) ## NBOX2: The Tilisi Campus Expansion In August 2023, iXAfrica acquired 11 acres of land in the Tilisi development, a 400-acre master-planned mixed-use development located approximately 30 kilometres from the NBOX1 campus along the Nairobi-Nakuru highway near Limuru. This second campus, designated NBOX2, is planned for over 53 MW of IT load capacity. The Tilisi location was chosen strategically. While Mombasa Road offers the best connectivity to coastal submarine cables, Tilisi provides access to different fibre routes, a separate power feed, and geographic diversity that allows iXAfrica to offer disaster recovery and business continuity options to customers who need their primary and secondary infrastructure in different physical locations. The Tilisi development itself is designed with modern infrastructure including dedicated power and water systems, making it well-suited for data centre operations. Between the NBOX1 campus on Mombasa Road and the NBOX2 campus at Tilisi, iXAfrica's total planned capacity exceeds 75 MW of IT load. This scale positions the company not just as a Kenyan operator but as a platform capable of serving as the primary digital infrastructure hub for the entire East and Central African region. ## Carrier-Neutral Model and Connectivity Ecosystem A data centre is only as valuable as the networks and services reachable from within it. iXAfrica's carrier-neutral model is central to its value proposition, and it has invested heavily in building a connectivity ecosystem that makes NBOX1 the most interconnected facility in the region. As of mid-2026, NBOX1 hosts connections from more than 20 network providers, including all major Kenyan telecommunications operators, international wholesale carriers, and content delivery networks. The facility's four MMRs provide physically separated interconnection spaces where these providers can exchange traffic with each other and with enterprise customers without traversing the public internet. Two internet exchange points operate within NBOX1, which is a significant competitive advantage: - **KIXP (Kenya Internet Exchange Point):** The country's primary domestic internet exchange, operated by the Telecommunications Service Providers Association of Kenya (TESPOK). KIXP established a new peering point of presence at NBOX1, which localises traffic exchange and reduces both latency and costs for participating networks. - **LINX Nairobi:** A point of presence operated by the London Internet Exchange, one of the world's largest internet exchanges by membership and traffic volume. The LINX presence at NBOX1 enables direct peering between East African networks and global carriers, content providers, and cloud platforms without traffic needing to transit through Europe first. Beyond traditional connectivity, iXAfrica also hosts SpaceX's Starlink teleport infrastructure, serving as the regional ground station for satellite broadband services across East Africa. This is a notable win because it demonstrates the facility's ability to attract and serve non-traditional data centre customers who require specialised infrastructure. The company has also established a strategic partnership with Safaricom, East Africa's largest telecommunications operator. Under this arrangement, iXAfrica provides the physical data centre platform while Safaricom contributes its brand recognition, established enterprise relationships, and last-mile connectivity. The partnership is jointly targeting enterprise clients and cloud providers who want the confidence of a Safaricom-branded solution delivered from a purpose-built hyperscale facility. ## Leadership Team iXAfrica's leadership combines deep global data centre expertise with local market knowledge. **Guy Willner, Co-Founder and Executive Chairman.** Willner has worked in the data centre industry since 1999. He co-founded IXEurope, one of Europe's first carrier-neutral colocation providers, and served as its CEO through its growth from startup to a publicly traded company. His experience building IXEurope, and later serving in leadership roles at IXcellerate in Russia and India, gave him the playbook for identifying underserved markets and building greenfield data centre platforms. He has been based in Nairobi since co-founding iXAfrica in 2018. **Snehar Shah, Chief Executive Officer.** Shah was appointed CEO in September 2023, bringing 25 years of cross-industry experience spanning telecommunications, investment banking, and infrastructure development. Notably, Shah previously helped build Moringa School, one of East Africa's prominent technology education institutions, giving him both an understanding of Kenya's tech ecosystem and a personal commitment to talent development. Under his leadership, iXAfrica has gone from pre-launch to commercial operations, signed the Safaricom partnership, and begun discussions with global hyperscalers. The board of directors includes William Barney (Bill), founder and chairman of Asian Century Equity and a former award-winning telecoms CEO, providing additional governance experience at the intersection of telecommunications and digital infrastructure. ## Role in Kenya's Data Centre Ecosystem iXAfrica occupies a specific and important position in Kenya's data centre landscape. The market has three broad categories of provider: - **Telecom-integrated facilities**, led by Safaricom, which primarily serve their own network and enterprise cloud requirements. - **Africa Data Centres** (a subsidiary of the Remgro-owned Digital Realty competitor), which operates facilities at Sameer Park and Westlands. - **iXAfrica**, the only provider building a greenfield, purpose-designed hyperscale campus. This distinction matters. Africa Data Centres' Nairobi facilities are capable and well-established, but they were not originally designed for the rack densities that AI workloads demand. iXAfrica's NBOX1 was designed from its first architectural drawings to support 40 to 50 kW per rack and to be upgradeable to liquid cooling for even higher densities. In practice, this means iXAfrica is currently the only facility in Kenya that can host NVIDIA GPU clusters and similar AI infrastructure without requiring the customer to accept design compromises. The company's impact extends beyond its own facilities. The launch of NBOX1 and the establishment of the KIXP and LINX peering points within it have effectively created a new interconnection hub for the region. Networks that previously had to establish separate peering relationships in multiple locations can now interconnect at a single facility. This network effect - where the value of the facility increases with each new network that connects to it - is a powerful competitive moat. The Kenyan government has recognised iXAfrica's strategic importance. At the NBOX1 launch in February 2025, Kenya Investment Authority (KenInvest) CEO John Mwendwa highlighted the facility's role in strengthening data sovereignty and positioning Kenya as a technology hub. The company is also engaged in AI policy discussions, contributing to regulatory frameworks on data sovereignty, digital policy, and AI governance in collaboration with the Kenyan government's tech envoy and the United Nations AI board. ## Significance for East Africa Kenya's data centre market does not exist in isolation. The country serves as the digital gateway to a landlocked hinterland of over 200 million people across Uganda, Rwanda, Burundi, South Sudan, the eastern Democratic Republic of Congo, and parts of Tanzania and Ethiopia. Any infrastructure built in Nairobi has a potential addressable market that extends far beyond Kenya's 50 million citizens. iXAfrica's facility is positioned to capture this regional demand. The Mombasa Road location sits directly on the fibre routes that carry international traffic from Mombasa's cable landing stations to these landlocked markets. Enterprises in Kampala, Kigali, and Bujumbura that want to access cloud services, AI compute, or content delivery with low latency will find iXAfrica's facility the natural interconnection point. The AI dimension adds urgency. As global hyperscalers look to establish presence in Africa beyond South Africa, Kenya is consistently mentioned alongside Nigeria as the next priority market. Oracle has announced plans for a sovereign cloud in Kenya. Microsoft has committed to investing $1 billion to establish a regional cloud zone in the country. These deployments require the kind of hyperscale, high-density facilities that iXAfrica has built, and the company reports that it is in active discussions with several hyperscale providers. The submarine cable landscape is also evolving in ways that favour Kenya. Meta's 2Africa cable, one of the largest subsea projects globally, is expected to go live with Kenyan landing capacity. Google's Africa-1 cable is under development. These next-generation cables will dramatically increase international bandwidth availability at the Kenyan coast, reducing the cost per megabit and enabling new categories of data-intensive services. iXAfrica's campus, sitting on the fibre corridor that carries this bandwidth inland, is architecturally positioned to be the primary on-ramp for this capacity. ## Sustainability and ESG Positioning Sustainability is not an afterthought for iXAfrica - it is a structural advantage of operating in Kenya. The country's 93% renewable energy grid, anchored by geothermal production from Olkaria, means that every kilowatt-hour consumed by NBOX1 has a significantly lower carbon footprint than equivalent facilities in markets dependent on fossil fuels. iXAfrica has gone further by pursuing EcoVadis certification, which evaluates companies across environmental, labour, ethics, and sustainable procurement criteria. The certification provides independently verified ESG credentials that are increasingly required by international enterprise customers and cloud providers as part of their own sustainability reporting obligations. Looking ahead, iXAfrica is preparing for the enactment of Kenya's power wheeling legislation, which has been passed by Parliament and is awaiting gazettement. Power wheeling will allow the company to purchase electricity directly from renewable generators, potentially including geothermal producers at Olkaria, rather than buying through the grid at blended tariffs. This could further reduce both energy costs and the carbon intensity of operations. ## Talent Development and Community Impact One of the most significant long-term constraints on Africa's data centre industry is the shortage of skilled professionals. iXAfrica has addressed this proactively through partnerships with Nomad Futurists, a global organisation focused on developing data centre talent, and through its own internship programme that recruits university students and trains them in mechanical and electrical engineering, cloud engineering, and data centre operations. Shah's background at Moringa School has influenced this approach. The company views talent development not as corporate social responsibility but as a strategic investment: the more skilled professionals in the Kenyan market, the more effectively the broader ecosystem can grow. Several interns have progressed into full-time engineering roles at iXAfrica, and the company is working toward a more structured, scalable framework for talent pipeline development that could serve the wider East African market. ## Future Plans and Expansion iXAfrica's near-term priorities are clear. The construction of NBOX1.2, the 18 MW expansion on the Mombasa Road campus, is the immediate focus, with financing secured and construction underway. The development of the NBOX2 campus at Tilisi, with its 53 MW planned capacity, represents the medium-term growth vector. On the commercial side, the company is focused on converting hyperscale discussions into signed commitments. The entry of Oracle and Microsoft into the Kenyan market creates a window of opportunity, and iXAfrica's position as the only facility engineered for the workloads these companies require gives it a significant first-mover advantage. The Safaricom partnership is expected to accelerate enterprise customer acquisition, particularly among Kenyan financial institutions and government agencies that already have trusted relationships with the telecom. Beyond Kenya, Helios Investment Partners' portfolio includes Morocco Data Centre, suggesting the possibility of a pan-African platform. CEO Snehar Shah has identified Egypt, Nigeria, and Morocco as additional Tier 1 African markets for potential expansion. Kenya, however, remains the anchor and the priority. For Kenya's data centre ecosystem, iXAfrica's trajectory is unambiguously positive. The company has taken the upfront risk of building a large-scale, purpose-designed facility in a market that many international investors still view as emerging. Its success validates Kenya's positioning as East Africa's digital infrastructure hub and raises the competitive bar for every other operator in the market. Whether you are an enterprise customer evaluating colocation options, a network engineer planning interconnection, or a policy analyst tracking digital infrastructure development, iXAfrica is a company that will shape the trajectory of Kenya's data centre industry for the next decade. For a full list of data centre operators and facilities in Kenya, visit our [data centre directory](/directory). For broader market analysis, see our [Kenya data centre market outlook](/articles/kenya-data-centre-market-outlook-2025-2030). If you are new to colocation, our guide to [colocation in Kenya](/articles/what-is-colocation-kenya) covers the fundamentals. ## Kenya Recorded 11.1 Billion Cyber-Threat Events URL: https://data-centers-254.vercel.app/articles/kenya-11-billion-cyber-threats-data-centres Cluster: Security Published: 2026-10-07 Updated: 2026-10-07 Summary: Kenya recorded 11.1 billion detected cyber-threat events in FY2025/26. What the figure means for data centres, KIXP, national monitoring and digital infrastructure. In October 2026 the Communications Authority of Kenya announced that the country had detected 11.1 billion cyber-threat events during the 2025/26 financial year, up 29.0 percent on the year before, alongside 83.1 million advisories issued to affected ICT users, up 60.8 percent. Most coverage will stop at the number, add the word alarming, and move on. The number deserves a different reading, because detection at that scale is not a software story. It is an infrastructure story, and it is one of the cleanest demand signals for Kenyan data centres published all year. > **What the 11.1 billion means.** The figure counts detected cyber-threat events reported through Kenya's national cyber-response infrastructure. It does not mean that 11.1 billion individual systems were breached, and it does not mean every event was a successful attack. That distinction matters for anyone making infrastructure decisions on the back of this number, so it is kept unavoidable throughout this article. ## What the number is, and what it is not The bulletins behind the announcement are precise about what is being counted, and the precision is what makes the figure useful rather than frightening. Read it as a national systems-health reading: hostile traffic seen and logged by monitoring that exists, is staffed and publishes its methods. Not as a body count of victims, and not as a loss figure either. | The figure is | The figure is not | | --- | --- | | Detected events reported through national monitoring | Confirmed breaches | | A national monitoring output | 11.1 billion separate victims | | A volume signal | A measure of successful attacks | | Based on reported detection | A direct measure of financial loss | None of that diminishes the signal; it frames it. What grew 29.0 percent is what the country's monitoring can see, and building the capacity to see more is an infrastructure decision. ### Sources used in this article **Primary** - [National KE-CIRT/CC Quarterly Cyber Security Reports FY 2025/26 (index)](https://ke-cirt.go.ke/quarterly-reports/), read directly by quarter: [Q1 Jul-Sep 2025](https://ke-cirt.go.ke/wp-content/uploads/2025/10/2025-26-Q1-Cyber-Security-Report.pdf), [Q2 Oct-Dec 2025](https://ke-cirt.go.ke/wp-content/uploads/2026/01/2025-26-Q2-Cyber-Security-Report.pdf), [Q3 Jan-Mar 2026](https://ke-cirt.go.ke/wp-content/uploads/2026/04/2025-26-Q3-Cyber-Security-Report.pdf) and [Q4 Apr-Jun 2026](https://ke-cirt.go.ke/wp-content/uploads/2026/07/2025-26-Q4-Cyber-Security-Report.pdf) - [Communications Authority of Kenya announcement, 6 October 2026: 11.1 billion detected cyber threats and 83.1 million advisories in FY 2025/26](https://www.linkedin.com/posts/communications-authority-of-kenya_111-billion-cyber-threats-detected-831-activity-7513169843496550400-GP_g) - Computer Misuse and Cybercrime (Critical Information Infrastructure and Cybercrime Management) Regulations 2024, as quoted in the KE-CIRT/CC bulletins **Supporting** - [ISOC Pulse IXP Tracker, Kenya country page (PeeringDB sync, October 2026)](https://pulse.internetsociety.org/en/ixp-tracker/country/KE/) - [PeeringDB, KIXP Nairobi record, ix 236 (accessed October 2026)](https://www.peeringdb.com/ix/236) - [BBC News on the 2023 eCitizen incident (28 July 2023)](https://www.bbc.com/news/world-africa-66332346) - [The Record on the July 2026 presidential website defacement (21 July 2026)](https://therecord.media/kenya-presidential-website-hack-bitcoin-ransom) *Dataset checked: 7 October 2026.* The quarter-by-quarter build of the total is public, and it is what separates this article from a press release. The National KE-CIRT/CC, the country's 24/7 computer incident response team operating within the Authority, publishes quarterly cyber security reports, and the four quarters of FY 2025/26 add up almost exactly to the headline: 842,320,667 detected events in July to September 2025, 4,559,229,985 in October to December, 3,367,113,840 in January to March 2026, and 2,355,938,192 in April to June. That sums to 11,124,602,684, which rounds to the 11.1 billion in the announcement. The gap between the quietest quarter and the loudest is more than five times, with October to December spiking 441.27 percent before the year cooled off. Swings of that size are normal in national telemetry; the honest reading is the annual total and its trend, not any single quarter. ## What a billion detections actually requires A detected threat event is the output of a chain that has to exist somewhere physically. Telemetry has to be collected on networks, exchanged between operators, correlated against known patterns, stored, and reviewed by analysts on shift. The National KE-CIRT/CC sits at the national end of that chain, and since the Computer Misuse and Cybercrime (Critical Information Infrastructure and Cybercrime Management) Regulations 2024 came into force, the Authority's mandate explicitly includes establishing and operating a Cyber Security Operations Centre (CSOC) for the ICT and telecommunications sector. The 83.1 million advisories, 20,748,489 of them in the final quarter alone, are the human-readable output of all that machinery: every advisory assumed an ICT user somewhere had the connectivity to receive it and the infrastructure to act on it. The rest of the chain looks like this: users generate traffic, carrier networks carry it, [the Kenya Internet Exchange Point](/articles/kixp-internet-exchange-point-kenya) keeps local traffic local, data centres and cloud platforms host the services being attacked, facility teams watch their own halls, and the national team correlates what everyone sees. ```text Traffic ↓ Carrier networks ↓ IXPs and data centres ↓ Telemetry collection ↓ National CSOC / KE-CIRT ↓ Advisories and response ``` Each rung of that ladder is a physical system somebody has to build, power, connect and staff, which is why the chain reads as an infrastructure bill of materials rather than a diagram in a strategy deck. Kenya has a relatively mature version of that chain for the region, combining national response capability, an established IXP ecosystem and a growing data-centre market. The [ISOC Pulse IXP Tracker](https://pulse.internetsociety.org/en/ixp-tracker/country/KE/) reports, on its October 2026 PeeringDB sync, that 89 percent of Kenya's 225 active networks are either IXP members themselves or customers of IXP members, and PeeringDB's record for the Nairobi exchange lists 144 member networks with about 2,985 Gbps of connected port capacity. The exchange fabric is therefore a genuine national vantage point rather than a boutique facility. Detection capacity and peering capacity are, in practice, the same investment. ![A technician connecting fibre optic patch cables on a distribution panel](/images/fibre-patch-panel-technician.webp) ## Where the detection infrastructure sits The DC254 map answers the where question directly. Nairobi holds 19 of Kenya's 27 tracked facilities, 13 of them operational, and that cluster is where the halls, NOCs and interconnection fabric concentrate. Mombasa is the other half of the story: all four of its tracked facilities are operational, and it is the coast where 7 in-service submarine cable systems come ashore (of 10 tracked, with Africa-1 landed and still awaiting RFS, plus Daraja announced and LuLu planned). The market's published live IT load is 10.5 MW, with 8 of the 20 operational facilities publishing figures, live designed capacity stands at 42.9 MW, and the announced pipeline adds 230 MW more. The build-out creates more of the physical capacity on which detection, storage, analysis and response can run, but only where security workloads, connectivity and skilled teams are actually provisioned. That is the defensible version of the claim, and it is also the commercially interesting one: the facilities that provision for it are the ones that win the regulated tenants. ![Data centre cooling plant with overhead pipework and fan units beside server racks](/images/dc-cooling-plant-pipes.webp) ![Rows of server racks in a data centre aisle lit by red status lighting](/images/dc-aisle-red-status-lighting.webp) The bulletins make the security relevance of that geography explicit. In the final quarter of FY 2025/26 the national team counted 17,406,495 web application attack attempts, a 43.68 percent increase on the quarter before, targeted at the critical information infrastructure sector, with government systems and internet service providers the primary targets. Kenya's [documented breach history](/articles/kenya-data-breach-timeline) says the same thing from the incident side: the 2023 eCitizen DDoS degraded the gateway to more than 5,000 public services (BBC, 28 July 2023), and in July 2026 the president's own website was defaced with a Bitcoin ransom note (The Record, 21 July 2026). The attack types are the ones catalogued in [our attack scenarios guide](/articles/data-centre-attack-scenarios-kenya); what is growing is the volume behind them, and the [critical information infrastructure rules](/articles/kenya-ict-policy-framework-data-centres) now formally assume operators are wired into the national detection effort. ## Why the number is a demand signal Every one of those 11.1 billion events consumed something real: bandwidth on a carrier network, packets across an exchange switch, log storage on a rack, minutes of an analyst's shift. Every one of the 83.1 million advisories travelled over Kenyan networks to someone who then had to patch, reconfigure or at least read. Some of that monitoring can be delivered remotely, but response latency, regulatory accountability, data governance and access to critical infrastructure make local and regionally accountable capability increasingly important, and the 2024 regulations assume the systems watching Kenya's infrastructure are reachable and answerable. Facilities that can demonstrate layered controls, from the network edge down to [the physical layer of data centre security](/articles/data-centre-physical-security-kenya), are the ones that win the regulated workloads, the banks and government tenants, whose telemetry the CSOC exists to watch in the first place. ![An uninterruptible power supply and switchgear room in a data centre](/images/dc-ups-power-room.webp) That is why this statistic belongs in the infrastructure conversation rather than the fear conversation. Kenya's digital economy is adding load from AI, cloud computing, financial services and digital public platforms at the same time as the threat volume grows, and both curves land on the same racks. The 230 MW pipeline is the country building the capacity to absorb both curves at once. Read the number as a national systems-health reading: hostile traffic is rising, and so is the country's ability to see it, count it and answer it. ## What this means for operators For the people running Kenyan facilities, the FY 2025/26 numbers translate into a working checklist rather than an abstraction. Detection at national scale assumes somebody retained the logs, so log retention and storage capacity stop being housekeeping and become part of the country's visibility. The bulletins' own attribution of the volume to unpatched systems makes patching discipline and security monitoring coverage the first-order controls, and DDoS protection matters on the evidence too, with 819,325 denial-of-service attempts counted in the final quarter alone. Network segmentation is what keeps a compromise in one tenant zone from becoming a compromise of the hall. The physical and the human controls belong on the same list. Physical access controls are the layer most often underestimated, because tailgating and unlogged access defeat the best firewalls; [the attack types that hit Kenyan data infrastructure](/articles/data-centre-security-threats-kenya) mostly arrive over the network, but the incidents that do lasting damage tend to combine both. Incident-response staffing, recovery testing and customer notification processes decide whether an incident is an event or a headline, and the expectations are set out in [our disaster recovery and business continuity guide](/articles/disaster-recovery-business-continuity-kenya). For facilities designated as critical information infrastructure, incident reporting to KE-CIRT/CC and compliance with the national security standards are already obligations under the 2024 regulations, not aspirations. None of this requires exotic spending. It requires the boring, auditable kind: documented access control, tested recovery, evidence of monitoring coverage and a named person who answers when the national team calls. Operators that already run layered defences treat the regulations as documentation work; those that have not should start at [the physical layer of data centre security](/articles/data-centre-physical-security-kenya) and work upward. Cybersecurity is not separate from digital infrastructure. It is one of the reasons digital infrastructure exists. The 11.1 billion is worth knowing, but the more useful question sits behind it: what does a country need to have built, powered, connected and staffed to see that number at all? Kenya's current answer, visible on [the DC254 map](/infrastructure/map), is 27 tracked facilities, 20 operational, an exchange point that touches nearly every network in the country, and a pipeline that keeps growing. The threat data and the construction data are the same story told twice. ## Kenya's AI Data-Centre Race: Who Can Secure the Power? URL: https://data-centers-254.vercel.app/articles/kenya-ai-data-centre-power-race Cluster: AI & Infrastructure Published: 2026-10-13 Updated: 2026-10-13 Summary: Announced megawatts are not operating megawatts. What DC254's tracker, power ledger and the Hercules proposal reveal about who can secure electricity. Kenya's drive for digitalisation has intensified the conversation around data-centre development. Investment announcements, new facilities and plans for AI infrastructure keep raising expectations about the country's position in Africa's digital economy. The ambition is no longer in question. The market has moved from whether Kenya will host meaningful computing capacity to how much, how soon, and where. But one question deserves greater scrutiny than it usually receives: **who can secure the reliable, affordable electricity required to operate the next generation of data centres?** The buildings matter. So do fibre networks, subsea cables, land and computing hardware. Yet without dependable power, a data centre cannot deliver its intended computing capacity. Power is the one input every other investment depends on, and it is also the input most often summarised in a headline number rather than examined in detail. For Kenya, the challenge is becoming more specific. Can developers secure electricity at the scale, location, cost and reliability their projects require, and demonstrate that the supporting infrastructure is ready to operate? Those are measurable questions, and this article works through them with the evidence available today. ![High-voltage substation yard in Kenya with transmission towers and insulator strings](/images/kenya-hv-substation-yard.webp) ## The numbers behind Kenya's data-centre market The [DC254 tracker](/tracker), last verified in October 2026, is the starting point for understanding the gap between existing infrastructure and announced ambition. The dataset tracks 27 Kenyan data-centre facilities, of which 20 are operational. Those operating facilities publish 10.5 MW of live IT load, figures disclosed by eight of the twenty operators, against 42.9 MW of designed capacity. Behind them sits a further 230 MW of announced pipeline: 77 MW under construction, 53 MW committed and 100 MW at early stage, led by the Microsoft-G42 proposal near Olkaria. These figures describe different stages of infrastructure development, and they should not be treated as interchangeable measures of available capacity. A project can announce hundreds of megawatts without having secured financing, completed construction, commissioned its electrical systems or installed computing equipment. Likewise, a facility's designed capacity does not tell you how much IT load is installed and operating: across the 20 operational facilities, published live load runs at less than a quarter of designed capacity. The distinction matters. **For investors and customers, the question is not simply how much capacity Kenya has announced. It is how much capacity can actually be delivered.** Power is central to that distinction, because electricity must arrive at a specific site, in sufficient quantity, at a contracted price, before anything else in the facility can function. The quarterly [State of the Market 2026-Q3](/research/state-of-the-market-2026-q3) report documents the same figures together with their methodology, and [Kenya's data centre market in numbers](/articles/kenya-data-centre-market-numbers) unpacks what each measure does and does not tell you. ![GPU servers installed in a data centre hall](/images/ai-gpu-servers.webp) ## Three approaches to powering Kenya's next data centres Kenya's emerging data-centre market presents different approaches to electricity supply. Three models deserve closer examination: grid-connected facilities, dedicated generation, and renewable-backed power. The baseline for all three comes from [the DC254 Rack Report](/rack-report) power review published on 11 October 2026, which re-checked the power record for all 27 tracked facilities field by field and source by source. Every single record starts from Kenya Power. Fourteen facilities disclose on-site backup generation alongside their grid supply, six run on grid service alone with one further grid connection recorded as planned, three pursue redundancy-first designs, and two disclose no arrangement at all. Not one operator has yet put their name to a renewable power purchase agreement. ### 1. Grid-connected facilities The first model relies on electricity supplied through Kenya's national power system. The sector's legal framework is the Energy Act 2019, under which the regulator, the Energy and Petroleum Regulatory Authority, reviews and approves the power purchase agreements that govern electricity sales, with Kenya Power as the principal distributor and transmission and distribution infrastructure connecting generation to customers (EPRA). For data-centre operators, the important questions extend beyond the amount of electricity generated nationally. They need to establish whether sufficient power can be delivered to a particular site, whether the connection can support the facility's intended load, what the commercial terms are, and how continuity will be maintained during outages or maintenance. A country can have substantial generation resources while a particular development still faces connection, transmission or site-level constraints. The cautionary example is the Microsoft-G42 campus announced near Olkaria in May 2024 at about $1 billion, which has stalled partly over a power request that grew from 60 MW to 1,000 MW, roughly a third of Kenya's total installed generation (Kenyans.co.ke, 22 August 2026; Business Daily, 10 September 2026). Kenya's record peak demand stands at 2,439 MW, reached in December 2025 (TechCabal, 20 August 2026), so requests of that size stop being a connection query and become a national planning question. These questions require project-specific evidence. National generation figures alone cannot establish whether a particular data centre has secured the power it needs. [Kenya's power infrastructure and the data centre constraint](/articles/kenya-power-infrastructure-data-centres) examines the delivery chain in detail, and [how power reliability shapes data centre design](/articles/kenya-power-reliability-data-centres) explains what operators build to compensate for it. ### 2. Dedicated generation: the Hercules proposal A more distinctive approach is emerging in Mombasa. AMACO Energy Group, a Greek energy and infrastructure firm, has announced plans for Hercules, a proposed AI data-centre and energy infrastructure project estimated at US$1.5 billion, first reported on 17 August 2026 (Business Daily). In September 2026, AMACO announced a cooperation agreement with GE Vernova covering gas-turbine technology for a planned offshore power barge (Business Daily, 10 September 2026). The concept is to develop electricity generation alongside the proposed computing infrastructure rather than depend on a conventional grid connection. ![A floating power barge with gas turbine exhaust stacks moored near a city coastline](/images/amaco-power-barge.webp) The proposal is significant because it places power generation at the centre of the data-centre investment model: the fuel, the turbines and the computing load are designed as one system. The candidate sites are Dongo Kundu and Kilindini near the Port of Mombasa, and the underlying LNG-to-power registration sits in the permitting phase with a 2027 target start year (Business Daily, 10 September 2026; AMACO project page). However, an agreement with an equipment supplier is not evidence that a power plant or a data centre has been commissioned. The project's final capacity, financing, confirmed site, construction schedule, fuel arrangements and contracted customer demand all require further verification, and none of these has been disclosed. The energy model also raises questions about fuel costs, emissions, grid integration, backup arrangements and the long-term commercial viability of LNG-powered computing infrastructure, which we examine in [the full Hercules case study](/articles/amaco-hercules-mombasa-ai-data-centre). Hercules should therefore be tracked as a proposed project, not as operating Kenyan data-centre capacity. It is not counted in the DC254 directory's 27 facilities, because a proposal without approval, financing or a confirmed site does not meet the directory's verification bar, whatever the size of its headline number. ### 3. Renewable-backed power The third approach involves renewable electricity, whether through direct generation, contractual supply arrangements or other structures that support a facility's energy requirements. Kenya's geothermal, hydro and wind resources make this an important area of investigation: the national grid draws over 90 percent of its electricity from renewable sources, one of the greenest generation mixes in the world (DC254 Rack Report, 11 October 2026). ![Olkaria geothermal power plant in Kenya's Rift Valley with steam rising](/images/kenya-geothermal-plant-olkaria.webp) But the presence of renewable generation in the national electricity mix does not automatically establish that a particular data centre is supplied by a dedicated renewable-energy arrangement. The grid mix is a national generation statistic, not a guarantee of site-specific power delivery. The same power review found only three records in the directory carrying any renewable-energy claim at all, and none of the three is an operator-disclosed renewable power purchase agreement. The distinction is important. A credible assessment needs to establish the operator's actual supply arrangements, the contractual structure, the source of electricity and any claims about renewable-energy coverage. It should also examine how the facility maintains continuity when a particular generation source is unavailable. Renewable supply, grid connections and backup systems are not necessarily competing choices; a data centre may combine several of them to meet its operating requirements. For DC254, the objective is to document the arrangement each operator actually uses rather than assign facilities an energy profile based on assumptions. [Kenya's renewables and industrial power](/articles/kenya-renewables-industrial-power-data-centres) tracks what would change this picture, and what a genuine renewable supply claim would look like. ## The real test: can power be delivered when computing needs it? The AI infrastructure debate often focuses on headline investment figures and planned megawatts. Yet several distinct milestones sit between an announcement and a functioning AI data centre: 1. **Power identified:** The developer has described a potential source of electricity. 2. **Power contracted:** A supply or generation arrangement has been agreed, subject to its terms and conditions. 3. **Power infrastructure delivered:** Required generation, transmission, connection and distribution equipment has been installed. 4. **Systems commissioned:** Electrical and backup systems have passed the necessary commissioning tests. 5. **IT capacity operational:** Computing equipment has been installed, energised and made available for service. These milestones should not be collapsed into a single claim that a project has secured power. A signed agreement may be an important step, but the evidence required to establish operational readiness is different in kind, not just in degree. Hercules currently sits between the first milestone and the second: the power source is described and an equipment partner is engaged, but no supply contract, financing or construction schedule has been published. ![Uninterruptible power supply modules and switchgear in a data centre power room](/images/dc-ups-power-room.webp) For large AI workloads, operators must also plan cooling, power distribution, redundancy and the electrical requirements of high-density computing equipment. These systems influence both the facility's design and its operating costs, and they are where paper projects quietly fail: commissioning tests exist precisely to expose the gap between what a design promises and what the installed system delivers. The relevant question is whether the complete infrastructure system can support the intended workload, not simply whether a developer has announced an electricity source. [UPS systems and backup power for Kenyan data centres](/articles/ups-backup-power-kenyan-data-centres) covers the continuity layer in detail. ## What this means for Kenya's competitiveness Reliable power is one component of a competitive data-centre market. It must work alongside connectivity, suitable sites, financing, skilled personnel, regulatory clarity and customers willing to purchase computing services. A weakness in any one of them eventually shows up as an empty hall or a delayed opening, whatever the launch ceremony said. Geography offers Kenya's two candidate hubs different starting hands. Mombasa has a potential strategic advantage in its international connectivity and port infrastructure: seven live submarine cable systems make landfall on its coast, and the port can deliver fuel and heavy equipment by sea. Nairobi has an established concentration of digital infrastructure and network interconnection, anchored by operating facilities such as [iXAfrica's NBOX1](/directory/ixafrica-nbox1) and [iColo MBA1](/directory/icolo-mba1). ![Nairobi skyline at night with the KICC tower lit against the city lights](/images/nairobi-skyline-night-kicc.webp) But geographic advantages do not establish that a proposed facility is commercially ready. A developer must still demonstrate that it can finance the project, secure the required approvals, deliver the power architecture, connect to the relevant networks and attract enough demand to justify the investment. For AI infrastructure, these questions become especially important because the costs of computing hardware, electricity, cooling and supporting infrastructure must be considered together: a cheap kilowatt-hour delivered unreliably can cost more than an expensive one delivered with proper redundancy. Kenya's opportunity is therefore not measured solely by the size of its announced project pipeline. It also depends on the country's ability to turn credible proposals into reliable, operating infrastructure. That is a question of delivery, and delivery is verifiable. ## What should Kenya disclose? There is a practical opportunity to improve the quality of public information about data-centre development, and the market itself supplies the argument for it. Today only eight of the twenty operational Kenyan facilities publish a live IT load figure at all, which means most capacity conversations run on designed numbers rather than operating ones. Operators, developers and relevant public agencies could help investors, customers and policymakers by publishing consistent information on: - **Operational status:** Announced, financed, under construction, commissioned or operating. - **IT capacity:** Installed and operating load, clearly distinguished from designed or proposed capacity. - **Electricity arrangements:** Grid connection, contracted supply, dedicated generation and backup systems. - **Power readiness:** Connection milestones, commissioning status and material constraints, where disclosure is appropriate. - **Energy and cooling:** Relevant consumption data, cooling systems and documented sustainability commitments. - **Project delivery:** Financing milestones, construction progress and expected commissioning dates, clearly labelled as estimates when unconfirmed. Such disclosure would make it easier to distinguish infrastructure that exists from capacity that remains on paper. It would give customers a stronger basis for evaluating service availability, and give policymakers better evidence for planning electricity, transmission and digital infrastructure. Our earlier analysis of [what operators should disclose](/articles/kenya-data-centres-what-operators-should-disclose) sets out the case in full. ## The DC254 position Kenya's AI data-centre race should be assessed through evidence of delivery, not announcements alone. DC254's role is to track the progression from proposed investment to secured infrastructure, commissioned systems and operating computing capacity. That means asking the same questions of every project: What is confirmed? What remains proposed? What capacity is actually operating? What evidence supports the claim? The Hercules proposal provides a useful case study because it makes the relationship between electricity generation and AI infrastructure explicit. But the same verification standard should apply to grid-connected facilities and to projects pursuing renewable-backed power. The objective is not to favour one energy model before the evidence is available. It is to establish which arrangements can meet each project's requirements, at what cost, with what risks and under what conditions. **As AI infrastructure investment grows, which developers can demonstrate that they have secured the power, financing, connectivity and operational capability required to deliver working data-centre capacity?** The answer will tell us more about Kenya's readiness for AI infrastructure than the announced megawatts alone. DC254 tracks Kenya's data-centre facilities, capacity, development pipeline and supporting digital infrastructure in the [public tracker](/tracker). Project statuses and capacity figures should always be read alongside their sources, verification dates and stated definitions. ## Frequently asked questions ### How much data-centre capacity does Kenya actually have operating? As of the DC254 tracker, last verified in October 2026, Kenya has 27 tracked facilities, of which 20 are operational. Those operating facilities publish 10.5 MW of live IT load, disclosed by eight of the twenty operators, against 42.9 MW of designed capacity. A further 230 MW of pipeline is announced: 77 MW under construction, 53 MW committed and 100 MW at early stage. These are different stages of development and should not be read as one number. ### Does Kenya have enough electricity for AI data centres? The honest answer is project-specific rather than national. Kenya's grid draws over 90 percent of its electricity from renewable sources, but a national generation mix does not guarantee delivery to a particular site at a particular size and price. The measurable questions are whether each developer can secure power at the scale, location, cost and reliability its project requires, and whether the connection, backup and commissioning evidence supports that. This article deliberately avoids a blanket shortage or surplus claim because the project-level evidence, not the headline, decides outcomes. ### What is the Hercules project in Mombasa? Hercules is a proposed AI data-centre and energy infrastructure project in Mombasa announced by AMACO Energy Group in August 2026, estimated at US$1.5 billion. In September 2026 AMACO announced a cooperation agreement with GE Vernova covering gas-turbine technology for a planned offshore power barge. The design would generate electricity alongside the computing load instead of relying on the grid. Final capacity, financing, a confirmed site, construction schedules and customers have not been disclosed, so DC254 tracks it as a proposed project and it is not counted in the directory's 27 facilities. ### Are Kenya's data centres powered by renewable energy? The grid is: over 90 percent of Kenya's electricity comes from renewable sources, stated as a national generation mix. The operators' own claims are thinner. DC254's power review published on 11 October 2026 found only three records in the directory carrying any renewable-energy claim at all, and none of them is an operator-disclosed renewable power purchase agreement. A renewable grid mix and a renewable supply contract for a specific facility are different things, and only the second can be verified per site. ### What is the difference between announced and operational capacity? Five milestones sit between them: power identified, power contracted, power infrastructure delivered, systems commissioned, and IT capacity operational. An announcement usually covers only the first. A signed agreement advances the second, but capacity only becomes operational when computing equipment is installed, energised and available for service after commissioning tests. Treating these stages as interchangeable is the most common error in reading data-centre market numbers. ## CA Consultation: Standalone Data Centre Licence URL: https://data-centers-254.vercel.app/articles/kenya-ca-standalone-data-centre-licence Cluster: Policy Published: 2026-09-10 Updated: 2026-09-24 Summary: The CA's 30-day consultation on a standalone Data Centre licence would move colocation out of the NFP-T2 regime. Fees, timeline, implications. On 6 March 2026, the Communications Authority of Kenya brought commercial data centres under telecom licensing for the first time, via Gazette Notice No. 3335 and the Revised Telecommunications Market Structure. The message then was that data centres are network facilities, and the Network Facilities Provider framework was where they belonged. Six months later, the message has changed. On 8 September 2026, the CA opened a public consultation on a proposal that would pull colocation data centres out of that framework entirely and give them their own licence category, a standalone Data Centre licence. This is not a routine rule tweak. It is the regulator formally acknowledging that the classification it chose in March was a compromise, and that the compromises that fit telecom network builders do not necessarily fit companies whose business is floor space, power, cooling, and physical security. For the roughly nineteen operational facilities across Nairobi and Mombasa, and for the investors backing the much larger pipeline behind them, the consultation window that opened on 8 September is the moment to shape the rules they will live under. ## What the CA Actually Proposed The proposal is precise about what would change. Colocation data centres would be removed from the Network Facilities Provider Tier 2 (NFP-T2) licensing category, where the March 2026 framework placed them, and rehosted under a new, dedicated Data Centre licence. Stakeholders have 30 days from 8 September 2026 to review and comment on the proposed framework, which puts the comment deadline at approximately 8 October 2026 through the CA's open consultations portal. The CA's own stated reasoning is worth reading carefully. The Authority said the current classification was intended to give it visibility over data centre operations (a real gap before March 2026) but places facilities within a licensing category primarily designed for companies that establish and operate telecommunications network infrastructure. In other words: the March framework solved a visibility problem, but at the cost of forcing data centres into rules written for a different kind of business. The consultation is an attempt to keep the visibility while dropping the misfit. ## Why the NFP Framework Never Quite Fit A network facilities provider builds towers, runs transmission backbone, and sells connectivity. A colocation data centre provides a building: secure floor space, redundant power, precision cooling, and cross-connects to the networks that customers bring with them. The facility itself does not deliver telecom services to end users, its tenants do. Under the proposed framework, the CA explicitly recognises this, describing the licence as reflecting the role data centres play in providing "colocation, power, cooling, storage and computing infrastructure rather than telecommunications services directly to end users." The distinction is not academic. Licence obligations designed for network operators, things like coverage rollouts, quality-of-service metrics for voice and data services, or universal service contributions tied to network revenue, translate awkwardly onto a facility whose revenue is rack space and power. The CA's fuller statement is worth quoting: "The proposed approach is intended to provide regulatory clarity, enhance visibility over data centre operations, support investment in digital infrastructure, and align Kenya's framework with proportionate approaches adopted in comparable jurisdictions." The Authority also scoped the licence to cover "entities that provide colocation data centre services, including the attendant supporting services", language that pulls the power, cooling and physical-security envelope into the licence's core definition rather than treating it as incidental. The "proportionate approaches" phrase is the tell: the Authority has benchmarked how comparable regulators treat data centres, and concluded the telecom regime over-applies here. That phrasing also suggests the Authority has been listening to the operators it regulates, several of whom have spent 2026 building out GPU-ready capacity for AI workloads under rules written for Telecoms Act compliance. ## The Numbers: Fees, Validity, and Who Is Exempt The proposed fee structure is modest by licensing standards, and that is a deliberate signal. The application fee would be KES 5,000, with an initial licence fee of KES 100,000. Annual operating fees would be KES 80,000 or 0.4% of annual gross turnover, whichever is higher, meaning the flat fee covers smaller facilities, while the turnover-based tier scales for large commercial operations. Licences would be valid for 15 years, giving operators a long runway for the kind of capital investment data centre campuses require. The contrast with the current NFP-T2 regime is stark, and it is the strongest evidence yet that the consultation is an investment-climate document as much as a regulatory one. Under NFP-T2 as it stands, an operator pays the same KES 5,000 application fee but then a 15-year licence fee of KES 15 million, or 0.4% of gross annual turnover, whichever is higher. The proposed standalone licence cuts that upfront commitment to KES 100,000: a reduction of more than 99% for a pure-play operator, while keeping the same 0.4% turnover test at the annual tier. One cost line belongs in every operator's model, though: the Universal Service Fund levy still applies on top of licence fees (presently 0.5% of annual gross revenue) and the consultation leaves room for that figure to change before the framework is finalised. Two carve-outs matter for the broader market. Entities already holding NFP licences would be permitted to establish and operate data centres without obtaining the separate Data Centre licence, so large integrated players keep their existing path. The same exemption extends to Application Service Provider (ASP) licence holders. In practice, the standalone licence targets pure-play colocation operators: the companies whose entire business is hosting other people's servers. That is the segment that grew from a handful of facilities to the nineteen-plus operational sites the industry counts today, and the segment most exposed to a licence category that was never designed for it. ## The Roadmap: Consultation Now, Implementation Later The CA is not promising a quick flip. The proposed data centre regime is scheduled to be finalised and implemented across two financial years. During FY2026/27, the roadmap calls for the consultation and finalisation of the framework (the 30-day window that opened on 8 September is the start of that process) and, in the same financial year, the consequential revision of the telecommunications market structure itself. During FY2027/28, the new regime would be implemented, with facilities migrating to the new category. That sequencing is worth pausing on, because press coverage at the time got it wrong. Early reports placed the consequential market-structure changes in FY2027/28 alongside implementation; the consultation document's own roadmap table puts them in FY2026/27, the same year as finalisation, with only the go-live in FY2027/28. This correction came out of capturing the regulator's full document on 24 September 2026 - the table reads: framework (FY2026/2027), public consultation (FY2026/2027), finalisation (FY2026/2027), consequential revision of the market structure (FY2026/2027), implementation (FY2027/2028). For operators, the practical reading is that NFP-T2 remains the licence of record through at least FY2026/27, and anyone entering the market now should still plan around the March 2026 framework while tracking the consultation's outcome. The two-step roadmap also means the market structure documents themselves get amended between finalisation and implementation. That is where transition mechanics (existing licence holders, renewal dates, fee harmonisation) will be settled. ## Where the ODPC Fits In A standalone CA licence licenses the building, not the data. The Data Protection Act 2019 runs in parallel and applies to the same facility regardless of what the Communications Authority decides: the operator is typically a data processor (and often a controller for its own staff and security data), and tenants are controllers of the personal data on their racks. In practice three DPA obligations land on the facility first. Registration with the Office of the Data Protection Commissioner (ODPC) where thresholds are met. Breach notification duties, where a facility that suffers a security incident must notify affected customers promptly so controllers can meet their own 72-hour ODPC notification window. And Section 48 transfer restrictions, which govern what tenants may do with personal data that leaves the country, rules our [Data Protection Act explainer](/articles/kenya-data-protection-act-data-centres) unpacks in full. For the consultation, the intersection is mostly about what the new licence does NOT settle. Data residency commitments, audit access and cross-transfer safeguards live in customer contracts and DPA compliance, not in CA licence conditions, and regulated tenants (banks under CBK guidance, SACCOs under SASRA) will keep demanding contractual in-country storage regardless of which regime licenses the rack. There is also a convergence point worth watching: both regulators now collect information about the same facilities, the CA for visibility over data centre operations, the ODPC through registration and breach reporting. Operators responding to the consultation may reasonably ask the Authority to align data-related conditions with ODPC requirements rather than invent parallel ones. For the sovereignty backdrop behind all of this, see [our data localisation deep dive](/articles/kenya-data-sovereignty-localisation). ## What It Means for Operators and Investors The direction of travel is regulatory clarity, and clarity is what investors have asked for since the AI investment cycle reached Kenya. Business Daily's reporting framed the move within the government's ambition to position Kenya as the digital gateway for Eastern and Central Africa, citing PwC's 2026 outlook work on data centre investment, the same research that projects Africa attracting $255 billion of cumulative data centre capex through 2050. A regulator that splits data centres out of a telecom framework into a purpose-built licence is making the market easier to underwrite: licence conditions matched to the business, fees scaled to facility economics, and a 15-year validity period that survives political cycles. The risks in the transition are real but manageable. Facilities licensed under NFP-T2 since March will face a migration process whose mechanics are not yet public. Operators with mixed businesses (connectivity plus colocation) will need advice on whether the NFP exemption already covers them or whether a separate application is cleaner. And the turnover-based annual fee introduces a revenue-visibility obligation that some private operators may push back on during comments. None of these are reasons against the reform; they are exactly the kind of detail the 30-day consultation exists to settle. ## What Is Not Clear Yet Reading the full consultation document sharpens some questions and exposes others that the seven pages simply do not answer. The scope phrase "attendant supporting services" is the biggest: it pulls power, cooling and physical security into the licence's core definition, but nothing in the document says where the boundary sits. A facility offering managed services, private cloud, or disaster-recovery seats could read itself in or out of that phrase, and the answer changes who must hold the licence. Migration mechanics are likewise unstated. Facilities that took an NFP-T2 licence after March 2026 face a transition whose shape (mid-term migration, fee credit for licence already paid, renewal alignment) is nowhere in the document. The Universal Service Fund treatment is cited to statute but not restated for the new category. And there is a dating discrepancy worth flagging in comments: the document describes the current market structure as "revised in April 2026", while the gazette record we hold dates the Revised Telecommunications Market Structure to 6 March 2026 - either a drafting slip or a reference to a later effective revision, and worth asking the Authority to clarify in writing. None of these gaps argue against the reform; they are the questions a serious submission would put on the record while the window is open. ## How to Participate The consultation is open through the CA's consultations portal at ca.go.ke, with comments due within 30 days of 8 September 2026. Operators, industry bodies, tenants, and investors with Kenyan data centre exposure all have standing to comment. For context on the framework being revised, see our full explainer on [Kenya's data centre licensing framework](/articles/kenya-data-centre-licensing-framework), and for the facilities this licence will cover, browse the [Kenya data centre directory](/directory). Sources: Communications Authority of Kenya open consultations; CA Proposed Licensing Framework for Data Centres (September 2026, Consultation Version) captured in full text on 24 September 2026; Developing Telecoms (9 September 2026); Business Daily (8 September 2026); TechAfrica News (8 September 2026); The Star (8 September 2026); w.media (September 2026). Facts verified against multiple independent outlets on 10 September 2026; fee comparison and Universal Service Fund detail verified against w.media on 12 September 2026; ODPC intersection section added 16 September 2026. On 24 September 2026 the regulator's consultation document was captured in full and every fee figure, the 15-year term, the NFP/ASP exemption and the roadmap were verified exact against it; the roadmap sequencing above was corrected on the same date (consequential market-structure revision sits in FY2026/27, not FY2027/28), and the document's "April 2026" dating of the current market structure was flagged as an open discrepancy against the 6 March 2026 gazette record. ## Kenya's Data Disasters: Breaches That Made History URL: https://data-centers-254.vercel.app/articles/kenya-data-breach-timeline Cluster: Security Published: 2026-09-16 Updated: 2026-09-16 Summary: From the 2017 election hacking claim to the 2025 companies registry breach: Kenya's documented data disasters, what actually happened, and how each ended. Kenya has been digitally attacked more often, and more visibly, than almost any country in Africa. That is not an insult; it is what happens when a country moves 5,000 government services onto one portal, runs its elections on biometric databases, and builds an economy on mobile money. The same connectivity that makes Kenyan institutions modern makes them targets. This article is the documented record. Every incident below is real, sourced, and dated, and each one ends with the part most coverage skips: how it was resolved, and what changed afterwards. If you operate infrastructure in Kenya, this is the list of things that have already happened once. ![Security monitoring screens showing alerts in a dark operations room](/images/dc-security.webp) ## 2017: the election systems and the claim that shook them The first nationally defining cyber crisis arrived in election season. On 9 August 2017, opposition leader Raila Odinga stood before the press and declared that Kenya's election results transmission system had been hacked, saying preliminary results showing him behind President Uhuru Kenyatta were a fraud (The New York Times, 9 August 2017). The Independent Electoral and Boundaries Commission's response was immediate and specific: a hacking attempt had been made, but it did not succeed (Al Jazeera, 11 August 2017). Two facts give that season its weight. First, on 1 August 2017, days before the vote, IEBC acting ICT director Chris Msando was tortured and killed; his murder was never solved, and it turned every technical question about the commission's systems into a question about trust (Al Jazeera, 1 August 2017). Second, on 1 September 2017, the Supreme Court annulled the presidential election, the first such ruling anywhere in Africa, citing procedural and transmission irregularities rather than a proven hack. **How it resolved:** the election was rerun in October 2017, the courts never confirmed a successful breach, and no group was ever charged with one. What Kenya learned, at the cost of a constitutional crisis, was that confidence in a national system is part of its security. A system that works but cannot prove it was not tampered with fails in every way that matters. ## July 2023: Anonymous Sudan takes on the whole government The most disruptive cyberattack in Kenyan history began on 27 July 2023. A group calling itself Anonymous Sudan launched a distributed denial of service attack, a flood of junk traffic aimed at overwhelming systems, against Kenya's digital public infrastructure. The eCitizen portal, the single gateway to more than 5,000 government services, went down or degraded (BBC, 28 July 2023). The attack landed on services Kenyans depend on daily: people could not buy electricity tokens, and services tied to mobile money and government payments stalled (CIPIT Strathmore, 22 August 2023). The government confirmed the cyberattack publicly and attributed the disruption to the DDoS technique (Anadolu Agency, 28 July 2023). Anonymous Sudan claimed responsibility and framed the attacks as retaliation for Kenya's position on the Sudan conflict, a political motive rather than a criminal one (Nigerian computer security agency advisory, July 2023; CIPIT, 22 August 2023). **How it resolved:** services were restored, and the government refused the group's framing that anything had been conceded. No ransom payment was ever confirmed. The lasting effects were structural: the attacks became the reference case for Kenya's dependency on single portals, they sharpened the mandate of the national cybersecurity coordination structures, and they remain the clearest proof that availability, not data theft, is the fastest way to hurt a country. ## August 2023: WorldCoin and the harvesting problem Weeks after the DDoS crisis, a different kind of data disaster appeared, this time invited through the front door. WorldCoin, a global identity project, was collecting Kenyans' iris scans and personal data in exchange for crypto tokens. On 2 August 2023 the Office of the Data Protection Commissioner opened an investigation, and on 3 August 2023 it suspended WorldCoin's operations in Kenya (Data Guidance, 1 August 2023). **How it resolved:** slowly, and with teeth. A High Court case, Republic v Tools for Humanity Corporation, ended in May 2025 with the court finding WorldCoin's activities illegal and in violation of Kenya's data protection framework (Kenya Law, 5 May 2025). In January 2026, the ODPC announced that WorldCoin had deleted the personal data collected from Kenyans in 2023. It is the country's clearest case of data harvested in bulk being forced back out again. ![A pocket copy of the Constitution of Kenya held in a hand](/images/constitution-kenya-pocket-hand.webp) ## January 2025: the companies registry breach The largest documented data exposure in Kenyan history hit an institution most Kenyans never think about until they need it. On the night of 31 January 2025, the Business Registration Service, the registry that holds every company's records, suffered a cyberattack that exposed private details (Nation Africa, 1 February 2025). The exposed data reportedly included confidential company ownership, directorship and beneficial owner information, and it was connected to records covering roughly two million companies (ITWeb Africa, 6 September 2026; Techpoint Africa, 3 February 2025). **How it resolved:** the breach became a long-running investigation. ITWeb Africa reported in September 2026 that the probe was still active, with the data believed to have been temporarily accessible before containment (ITWeb Africa, 6 September 2026). The lesson is uncomfortable: registries are attack surfaces. A country's corporate records are an intelligence product, and whoever holds them holds a leak waiting for a port scan. ## July 2026: the presidency's website, defaced for ransom The most recent entry brought the problem to the most visible address in the country. On Saturday 18 July 2026, hackers breached president.go.ke, replaced the homepage with an anti-government message, and demanded a ransom of five Bitcoin, roughly $320,000 or Sh41 million (Anadolu Agency, 18 July 2026; Eastleigh Voice, 19 July 2026). The site was taken offline, cleaned, and restored within about two days (ITWeb Africa, 20 July 2026). **How it resolved:** the ransom was not paid, the site was restored, and Kenya opened an investigation into the intrusion (The Record, 21 July 2026). A website defacement is not a database breach, but the choice of target was the point: the attackers bought national attention for the price of one compromised web server. ![Close-up of a server motherboard and processor](/images/dc-server-chip-2.webp) ## The backdrop: how often Kenya is actually attacked The headline incidents sit on top of a much larger, mostly invisible volume. The national computer incident response team, KE-CIRT/CC, detects threats at national scale and publishes the numbers: over 1.1 billion cyber threat events in the second quarter of 2024, over 840 million in the fourth quarter of 2024, and roughly 842 million in the third quarter of 2025 (Communications Authority quarterly reports). The overwhelming majority are automated attempts exploiting unpatched systems, and they fail without anyone outside a security operations centre ever hearing about them. Enforcement has matured alongside the threat. The Office of the Data Protection Commissioner issued its first penalty notices in September 2023 (Clyde & Co, 6 October 2023), had received over 7,100 complaints by March 2025, and can fine violators up to five million shillings or one percent of annual turnover. The Data Protection Act 2019 also imposes breach notification duties, which is why every incident above now has a legal tail as well as a technical one. ## What the record teaches Read together, the timeline is not random. It has a pattern that anyone running Kenyan infrastructure should memorise. **Availability is the first casualty.** The two most nationally painful incidents, 2023 and 2026, were about access, not theft. Attackers who want attention attack uptime, and Kenya has proven how much downtime a country will tolerate before it becomes politics. **Registries and identity are the crown jewels.** Election rolls, iris scans, corporate ownership: the same theme repeats. Data that identifies people and companies is worth more, and hurts more, than the transactional data everyone assumes is the target. **"Solved" usually means contained.** With the partial exception of WorldCoin, no Kenyan incident ended with hackers caught, prosecuted and jailed. Endings look like restored services, investigations that run for years, and laws tightened afterwards. Resilience, not retribution, is the realistic goal, and it is what we design data centre security for in the first place. ## Kenya's Data Centre and Compute Hub Plan, Explained URL: https://data-centers-254.vercel.app/articles/kenya-data-centre-compute-hub-plan-explained Cluster: Policy & Regulation Published: 2026-09-16 Updated: 2026-09-16 Summary: Kenya's ICT Principal Secretary says the cables are done and compute is next. We break down his plan, the numbers and what is actually built. On 15 September 2026, Eng. John Kipchumba Tanui, CBS, Kenya's Principal Secretary for ICT and the Digital Economy, published a long essay on LinkedIn with an ambitious title: Positioning Kenya as Africa's Preferred Data-Centre and Compute Hub. It is the most complete public statement yet of how Kenya's government sees this industry. It tells the story of how the country got here, sizes up the competition, admits the risks, and sets out what the state should do next. Pieces like this usually stay buried in policy language, so we read every word, checked the load-bearing claims against our verified directory, and translated it into plain English. ![Eng. John Kipchumba Tanui, Principal Secretary for ICT and the Digital Economy, speaking at a Konza Technopolis event](/images/tanui-ps-konza-podium.webp) ## The one-sentence version Kenya built the cables. Now it needs the compute. Tanui's opening argument is that every digital transaction, from an M-PESA payment to an AI chatbot answer, has a physical home: a building full of servers, storage and network equipment. Countries that host those buildings gain speed, control over their own data, technology investment and the base required for artificial intelligence. Countries that do not host them stay dependent. His framing is worth quoting exactly, because it is the whole essay in two lines: "Kenya has already built the cables. The next national assignment is to build the compute." For a newcomer, the word compute is the one to unpack. Cables are the roads that carry data in and out of the country. Compute is the factory work at the end of the road: the machines that actually process your mobile money record, your eCitizen application or a Kiswahili AI model's training data. Today, far too much of that factory work for African data happens outside Africa. The PS's sharpest line makes the stakes plain: if African data continues to be processed mainly outside the continent, Africa will "export the raw material of its digital economy and import the resulting intelligence, platforms and services." ## The numbers he builds the case on The essay leans on a handful of figures, mostly from industry and energy bodies. Here they are, with their sources, so you can quote them responsibly. - **Africa has about 18 percent of the world's people but under 1 percent of global data centre capacity.** This is the essay's central injustice, and it is consistent with the capacity estimates used widely across industry research. - **The global colocation and interconnection market was worth about US$119 billion in 2025 and is projected to approach US$147 billion in 2026** (PS essay, 15 Sep 2026). That is the pool of investment Kenya is competing for. - **Global data centre electricity consumption could more than double to about 945 terawatt-hours by 2030** (International Energy Agency, Energy and AI, April 2025). Slightly more than Japan uses in a year. - **Ireland's data centres grew from 5 percent of national electricity in 2015 to 22 percent in 2024** (PS essay, citing Irish statistics). This is presented as a warning, not a trophy. - **A large data centre may engage more than a thousand people at peak construction but keep only about 50 permanent workers** (PS essay, citing Northern Virginia, the world's largest market). Jobs, in other words, are not the main prize. One habit worth copying from the essay: it consistently separates operating capacity from pipeline. Announced projects and operating megawatts are treated as different things, which is exactly the discipline our own directory applies to every facility we track. ## How Kenya got here: the four phases Tanui tells Kenya's data centre history in four stages, and the arc explains why the current moment feels different. 1. **Private halls.** Government agencies, universities, banks and large companies built their own server rooms for records and internal systems. Banking digitisation, from ATMs to mobile banking, made financial institutions the first serious demand base. 2. **The telecoms era.** Operators built facilities for voice, internet and mobile money. Safaricom began offering cloud services around 2010, and those investments carried M-PESA, proving a globally significant platform could run on infrastructure hosted in Kenya. 3. **Commercial specialised hosting.** The East Africa Data Centre opened in Nairobi in 2013 under Liquid Intelligent Technologies and Africa Data Centres, marking the shift to purpose-built commercial facilities. Then iColo's carrier-neutral Nairobi One launched in 2019, letting many networks, banks and content providers interconnect under one roof without being tied to a single operator. That model turned Nairobi from a national telecoms centre into a regional interconnection market. 4. **The current stage.** International operators, public cloud infrastructure and facilities designed for dense cloud and AI workloads. This is the phase the essay is really about, and the one where the "preferred hub" question will be decided. We track the whole market this history produced in [our Kenya data centre market outlook for 2025 to 2030](/articles/kenya-data-centre-market-outlook-2025-2030) and in [our verified facility directory](/directory). ## What is actually real today (and what is still promised) This is where the essay earns its credibility, because it is more sceptical than most government writing. Here is the state of play it describes, cross-checked against what we track. ![Tatu City boulevard with Tatu City signage and office buildings under construction](/images/tatu-city-aerial.webp) **The Oracle cloud region: announced, not live.** President Ruto announced Kenya's first Oracle Cloud Infrastructure (OCI) public cloud region in January 2024. In January 2026, iXAfrica was named as the host partner in Nairobi (iXAfrica announcement, 27 Jan 2026). The region is under development and comes into service only after construction, power and cloud fit-out are done. When it lands, it will be the first real test of whether Kenya can host a global public cloud region, not just interconnection. You can see the facility it will run on, [iXAfrica NBOX1 in our verified facility directory](/directory/ixafrica-nbox1). ![Oracle's Kenya country leader and iXAfrica CEO Snehar Shah meet at the iXAfrica data centre in Nairobi](/images/oracle-ixafrica-partnership.webp) **Tatu City: the biggest thing being built right now.** Nxtra by Airtel, Airtel Africa's data-centre subsidiary, is developing a two-phase, 44 MW campus inside the Tatu City special planning area on Nairobi's northern edge. Trade press reported the groundbreaking and a US$150 million investment in September 2025 (DataCenterDynamics, 12 Sep 2025), and the PS cites commissioning targeted in 2027. No completion date is independently confirmed, so our verified directory has tracked it as Under Construction since the groundbreaking, and it now appears on our national infrastructure map too. **Digital Realty NBO2: open.** The new Nairobi Two (NBO2) facility in Karen, about 6.5 MW of design load (the PS's essay cites 6.4 MW), expands the former iColo campus and connects more than 100 networks, with access to two internet exchange points. It is real, operating capacity, and it completes iColo's transition to the Digital Realty brand in Kenya. **Microsoft and G42: still a discussion.** The geothermal-powered data centre and cloud project announced in May 2024, initially framed at US$1 billion near Naivasha, is described as strategically important but under discussion, with scale, power requirements and commercial structure unresolved. The PS's caution here is deliberate: pursue the investment, but do not place disproportionate guarantees on public institutions. That matches how our directory has treated the project all along. **Konza: the sovereign card.** The National Data Centre at Konza Technopolis provides sovereign infrastructure for public services and is positioned as a future site for high-performance computing and sovereign AI. It is the government's own facility, and the essay treats it as a strategic asset rather than a commercial rival. His own verdict on all of it is refreshingly honest: these projects "do not yet make Kenya a global compute hub. They do make it one of Africa's most credible contenders." ## What Kenya has that rivals cannot easily copy ![Geothermal power plant steam plumes at Olkaria in Kenya's Rift Valley](/images/kenya-geothermal-plant-olkaria.webp) The essay's strongest section is its inventory of national advantages, and it is worth reading as a package rather than a list, because the package is the point. - **Connectivity.** Multiple submarine cables land at Mombasa, linking Kenya to Europe, Asia, the Middle East and the rest of Africa, and national fibre pushes that reach across borders. We explain [why Kenya is positioned as East Africa's digital hub](/articles/kenya-east-africa-digital-hub) in a separate guide. - **Geothermal power.** Unlike solar or wind, geothermal runs around the clock, and Kenya's best fields (Olkaria near Naivasha, Menengai near Nakuru, and prospects toward Baringo) sit along a corridor that could host dedicated green-compute zones. Our guide to [how geothermal energy feeds Kenya's data centres](/articles/geothermal-energy-kenya-data-centres) covers the mechanics. - **Local demand.** M-PESA, digital banking, eCitizen, e-commerce and a growing tech sector create workloads that can anchor investment. Cloud regions follow customers, not the other way round. - **A policy stack that is moving.** The Kenya Cloud Policy, the National AI Strategy 2025 to 2030, the Data Protection Act and the removal of the mandatory 30 percent local shareholding requirement in ICT all push the same direction, and the Communications Authority is reviewing a dedicated data centre licence. We break down [the Communications Authority's standalone data centre licence consultation](/articles/kenya-ca-standalone-data-centre-licence) and [how tax incentives shape data centre investment in Kenya](/articles/tax-incentives-data-centre-investment-kenya) elsewhere on this site. - **Talent and regional reach.** Engineers, developers, financial-services professionals and universities, plus Nairobi's role as a regional headquarters city, which lets facilities here serve markets far beyond Kenya. On the rules side, one policy thread connects directly to the essay's compute argument: Kenya's data protection framework expects Kenyan personal data to have a Kenyan home, which is the legal pull behind local compute. We unpack [Kenya's data sovereignty and localisation rules](/articles/kenya-data-sovereignty-localisation) for anyone building under them. ## The warnings he gives his own industry ![Principal Secretary John Tanui addressing an audience at a Konza Technopolis podium](/images/tanui-ps-konza-address.webp) The essay is unusual for a government piece because it names the failure modes. Ireland shows what happens when data centre growth outruns the grid: 22 percent of national electricity by 2024, and new sites now expected to bring additional renewable generation with them. Northern Virginia shows the employment trap: thousands of construction jobs, then roughly 50 permanent staff per large facility, which means success must be measured by the digital economy the facilities enable, not headcount inside the halls. Malaysia shows the incentive trap from the other side: it attracted over RM110 billion in approved investment by tying incentives to outcomes, and the PS argues Kenya should reward operating capacity, local cloud services, additional renewable generation and skills, with clawbacks where outcomes are missed. He also lists the local risks plainly: electricity availability and price, competition with households for infrastructure, noise and land-use friction, cybersecurity and over-dependence on a single cloud provider, and the possibility of building capacity faster than workloads arrive. For context, Kenya's entire national peak demand is around 2,439MW, which is why every large campus conversation in Kenya is really an energy conversation. The practical discipline he proposes is disclosure: each major project should publish its expected electricity demand, the source of additional generation, water requirements, cooling technology and its contribution to local skills. ## His to-do list for the government The final stretch of the essay is essentially a memo to the state, and it is the part investors will read most closely. Condensed, the asks are: 1. A national programme tied to available power, fibre and water, with a published grid-connection queue. 2. A transparent large-electricity-load framework, with connection timelines, curtailment rules and clarity on who funds substations and transmission. 3. Dedicated geothermal and other renewable capacity for green computing along the Naivasha, Nakuru and Baringo corridor, built as additional megawatts, not reallocated household supply. 4. Incentives tied to local value, occupancy and sustainability, with clawbacks when outcomes are missed. 5. Aggregated government and regulated-sector cloud demand, with a hard cap on unsustainable guarantees. 6. Affordable GPU access for universities, start-ups and public institutions, with clear allocation rules. 7. A Nairobi, Konza, Naivasha, Nakuru and Baringo green-compute corridor, including land-use and community-benefit agreements. 8. Mombasa positioned as the international cable, edge-computing and disaster-recovery hub, not the default home of the largest AI training campuses. 9. Continued engagement with Oracle, Microsoft, Google and others, with a published list of what the state will and will not guarantee. 10. A public scoreboard: live megawatts, commissioned cloud regions, occupancy, renewable share, jobs, local procurement and compute hours available to Kenyan innovators. That last one should sound familiar. A public scoreboard of live megawatts and occupancy is close to describing what this site already does with its directory, and we would happily see the government publish its own version. ## Our fact-check verdict We ran the essay's checkable claims against our own verified data, and it holds up well. The iXAfrica OCI host-partner announcement (27 January 2026) matches our directory entry exactly. The Digital Realty NBO2 opening and its 100-plus networks match our tracked records. The Nxtra Tatu City figures align with trade-press reporting from September 2025 (groundbreaking, US$150 million) and with our own directory entry, which records the launch target slipping to July 2027; we have now also placed the campus on our national infrastructure map. The cautious framing of Microsoft and G42 matches the line our directory has carried since launch. The Africa capacity gap he cites (about 18 percent of population, under 1 percent of capacity) is consistent with the estimates used across industry research. What we could not verify: any completion dates. The 2027 commissioning target for Tatu City is a stated plan, not a confirmed date, and the OCI region has no public launch date. To his credit, the essay itself flags that gap between confidence and capacity. As he puts it, the test is whether "the innovation, computing and intelligence produced by this infrastructure also happen in Kenya." On the evidence in this piece, the government's thinking and the observable pipeline are, for once, pointed in the same direction. ## What this means for you - **If you buy infrastructure (CTO, founder, IT lead):** the pipeline is real but the live map is still short. Buy against operating capacity today, and negotiate hard against announced capacity tomorrow. Our [Nairobi colocation buyer's guide](/articles/colocation-data-centres-nairobi-buyers-guide) is the practical starting point. - **If you invest or operate:** watch the asks, not the speeches. The licence review, the incentive design with clawbacks, and the grid-connection queue are the signals that will decide where capital lands next. - **If you are building a career:** the essay's own numbers say construction jobs fade and permanent ones are few, but the enabled digital economy is the prize. The careers that last are in power, networking and the software that runs on top, which our [careers and business guides](/careers) cover from the ground up. ## Frequently asked questions ### Who is Eng. John Tanui and why does his essay matter? Eng. John Kipchumba Tanui, CBS, is Kenya's Principal Secretary for ICT and the Digital Economy, the senior civil servant who runs the ministry day to day under the Cabinet Secretary. When he publishes a detailed essay on data centres and computing, it signals what the government is actually thinking: which projects it counts as real, which it treats as talk, and which policy changes it is preparing. The essay was published on LinkedIn on 15 September 2026. ### What does "compute" mean in this plan? Compute is short for computing capacity: the servers, storage and specialised chips that actually process data, housed in data centres. Cables only carry information to and from a country. Compute is where the work happens. The PS's argument is that Kenya spent two decades building the cables (the roads for data) and now needs to build the factories at the end of those roads, so that Kenyan data is processed in Kenya instead of being sent abroad and bought back as services. ### Is the Oracle Cloud region live in Kenya yet? No. President Ruto announced Kenya's first Oracle Cloud Infrastructure (OCI) public cloud region in January 2024. In January 2026, iXAfrica was named as the host partner in Nairobi. As of September 2026 the region is still under development, with no public launch date. When it goes live it will let government agencies, banks and enterprises run cloud services hosted physically in Kenya, with lower latency and clearer data-residency options. ### What is being built at Tatu City? Nxtra by Airtel, the data-centre subsidiary of Airtel Africa, is developing a two-phase, 44 MW data centre campus inside the Tatu City special planning area on Nairobi's northern edge. Trade press reported the groundbreaking and a US$150 million investment in September 2025 (DataCenterDynamics, 12 Sep 2025), and the PS's essay cites commissioning targeted in 2027. Our verified directory lists the project as Under Construction, since no completion date is independently confirmed yet. ### What did the PS say about the Microsoft and G42 project? He was notably careful. The project, announced in May 2024 as a roughly US$1 billion geothermal-powered data centre and cloud venture near Naivasha, is described as strategically important but still under discussion, with its scale, power requirements and commercial structure unresolved. That matches how our directory treats it: an ambition, not committed capacity. His advice was to keep pursuing the investment without placing unsustainable guarantees on public institutions. ### Why does Ireland keep coming up in Kenya's data centre debate? Ireland is the cautionary tale of what success looks like when it outruns planning. Data centres there grew from 5 percent of the country's metered electricity in 2015 to 22 percent in 2024, and new sites are now expected to bring their own additional generation. The PS uses Ireland, and Northern Virginia's construction-heavy-then-operations-light job pattern, to argue that Kenya should design its rules before the pressure arrives, not after. ## Kenya Data Centre Licensing: NFP-T1 and NFP-T2 Explained URL: https://data-centers-254.vercel.app/articles/kenya-data-centre-licensing-framework Cluster: Policy Published: 2026-08-21 Updated: 2026-08-27 Summary: Kenya now requires data centre operators to hold an NFP-T2 licence (KES 15M, 15-year term). What the CA framework means for operators and investors. On 6 March 2026, the Communications Authority of Kenya formally brought commercial data centres under its telecommunications licensing framework for the first time. Gazette Notice No. 3335 gave effect to the **Revised Telecommunications Market Structure**, a document that explicitly categorises data centre operations as a licensable activity under the Network Facilities Provider regime. For the roughly nineteen operational facilities across Nairobi and Mombasa, and for the billions of shillings in planned investment, this is the regulatory framework they will operate under. ## Why Did the CA Decide to Licence Data Centres? Data centres have evolved from simple server rooms into critical national infrastructure. The CA's decision followed a formal public consultation that began with a paper published in December 2024, where the Authority argued that data centres, by virtue of mediating access to digital services, are functionally equivalent to communications tower companies and fibre operators, both of which have long been regulated as Network Facilities Providers. The Authority wrote that "given that the facility owner in the second type of data centre arrangement significantly influences data accessibility, it is necessary to bring these arrangements within the licensing framework to protect users' data access rights." This was not a cosmetic reclassification. It was an assertion that data centres are the physical layer of the digital economy, and that leaving them unregulated was a gap that needed closing. The urgency was demonstrated in June 2026 when a power outage at the Huduma Kenya data centre in Nakuru brought government services offline across all forty-seven counties simultaneously. Millions of citizens were unable to access identity documents, business registrations, and essential services. The incident proved that data centre failures are no longer theoretical risks, they are events with immediate, nationwide consequences. ![Nairobi skyline at sunset](/images/nairobi-sunset-wide.webp) ## How Does NFP-T1 Differ from NFP-T2? The CA did not create a standalone "data centre licence" in the March 2026 framework. Instead, it placed data centres within the existing Network Facilities Provider framework. **NFP-T2** is the primary route for pure-play data centre operators, while **NFP-T1** is available for large integrated operators who also require nationwide spectrum rights. > **Update (8 September 2026):** The CA has opened a 30-day public consultation on a proposed standalone **Data Centre licence** that would remove colocation facilities from NFP-T2 entirely, with fees of KES 5,000/100,000 at application and issue, annual operating fees of KES 80,000 or 0.4% of turnover, 15-year validity, and implementation targeted across FY2026/27–FY2027/28. Full analysis: [Kenya's CA opens consultation on a standalone data centre licence](/articles/kenya-ca-standalone-data-centre-licence). Until that regime is finalised and implemented, the NFP-T2 route described below remains the licence of record. ### NFP-T1 vs NFP-T2 for Data Centre Operators | Aspect | NFP-T1 (Tier 1) | NFP-T2 (Tier 2) | |---|---|---| | Data Centre Right | Permitted without additional licence | Primary licence category for DCs | | Geographic Scope | Countrywide with national spectrum | Countrywide, county-by-county build-out | | Spectrum Fees | Reservation + utilisation fees (national stack) | Utilisation fees only (approved regions) | | Licence Term | 15 years or 25 years (optional) | 15 years only | | Application Fee | KES 5,000 | KES 5,000 | | Initial Licence Fee | KES 15M (15yr) / KES 45M (25yr) | KES 15 million | | Annual Operating Fee | 0.4% of AGT or KES 4M, whichever is higher | 0.4% of AGT or KES 800K, whichever is higher | *Source: Communications Authority of Kenya, Revised Telecommunications Market Structure (June 2026)* During the consultation process, the CA had initially proposed placing data centres under NFP-T3, which covers limited geographic areas. Industry stakeholders pushed back, arguing that data centres inherently serve a national or regional function. The Authority accepted this reasoning, and the final framework places data centres under the two highest infrastructure tiers. ![Westlands commercial district in Nairobi](/images/nairobi-westlands.webp) ## Kenya's Data Centre Market: The Numbers Kenya is home to **19 operational data centres**: 15 in Nairobi and 4 in Mombasa. The coastal facilities are positioned near [submarine cables landing in Mombasa](/infrastructure), giving Kenya its competitive advantage as an East African connectivity hub. Nairobi serves as the primary market for domestic and regional enterprise demand, hosting the largest concentration of carrier-neutral colocation facilities in the region. The market's total IT power capacity stood at approximately **15 megawatts in 2025**, projected to reach **25 megawatts by 2030** (CAGR 10.76%, Mordor Intelligence). In monetary terms, the Kenya data centre market was valued at **USD 266 million in 2025** and is forecast to reach **USD 805 million by 2031** (CAGR 20.27%, ResearchAndMarkets). These figures represent one of the fastest-growing data centre markets on the African continent. Key operators include **Africa Data Centres** (Liquid Intelligent Technologies), which operates multiple facilities including a significant presence along Mombasa Road; **iXAfrica Data Centres**, developer of East Africa's first hyper-scale, AI-ready facility at NBOX1.1 with a 4.5 megawatt IT load; and various government-owned facilities supporting public sector digital services. Safaricom and Telkom Kenya also operate data centre infrastructure for their telecommunications and enterprise services. ![Nairobi skyline lit up at night](/images/nairobi-skyline-night.webp) ## Is Power Supply a Bigger Problem Than Licensing? While the regulatory framework has taken a significant step forward, the single largest obstacle to data centre growth in Kenya remains the power supply. In May 2026, Microsoft and G42's planned one-billion-dollar AI data centre project (potentially the largest single foreign direct investment in Kenya's digital infrastructure) stalled after it became clear that the national grid could not reliably deliver the required power capacity. Government officials indicated that meeting the project's energy demands would require rationing power to other consumers. ![Data centre rooftop chillers and generators](/images/dc-chillers-roof.webp) Kenya's total installed generation capacity stands at approximately 3,500 megawatts, predominantly from geothermal and hydroelectric sources. But the Microsoft project alone was reported to require several hundred megawatts, a demand that would represent a significant share of the national grid's firm capacity. For more detail on this constraint, see our analysis of [data centre power and energy challenges](/energy). The Huduma Kenya outage in June 2026 further illustrated the power reliability challenge: a single power supply failure at one data centre in Nakuru disrupted government services nationwide. This is not a resilience problem that licensing alone can solve, it requires investment in backup power systems, redundant feeds, and the expansion of Kenya's electricity generation and distribution infrastructure. ![High-voltage pylons on the backbone grid](/images/kenya-transmission-pylons.webp) ## What Other Changes Did the CA Introduce? The revised market structure included changes beyond data centre licensing that affect the broader digital infrastructure ecosystem: - **Landing Rights Authorisation (LRA)**: A new licence category for entities transmitting signals into Kenya via submarine cables or satellite. Previously, landing rights were bundled within the IGSS licence, creating barriers for international cable operators. - **Expanded NFP-T3**: Now covers up to three counties instead of one, making it viable for smaller infrastructure investors building across county clusters. - **Micro Network and Services Provider (MNSP)**: A new licence for operators serving limited areas from residential estates up to county level, covering voice, data, and internet services. These changes signal a broader strategy: make it easier for international operators to establish infrastructure in Kenya, while also promoting competition and lowering costs for consumers in underserved areas. ## What Should Operators and Investors Do Now? For existing operators, the immediate priority is ensuring they hold the appropriate NFP licence. Those already licensed under NFP-T2 can establish commercial data centres without additional authorisation. Those operating under other licence classes, or without any licence, need to assess their activities against the new framework. For prospective investors, particularly international operators evaluating East African market entry, the framework provides predictability that was previously absent. The fifteen-year licence term provides sufficient runway for return on investment, and the annual fee structure (percentage of gross turnover with a minimum floor) aligns the Authority's revenue with operator success. Kenya has progressively liberalised its ICT investment framework: the removal of the 30% local shareholding requirement, the introduction of Special Economic Zones with customs and tax relief on ICT equipment, and now the formal licensing of data centres all signal a deliberate strategy to position the country as a destination for digital infrastructure investment in Africa. The question is no longer whether Kenya is open for investment, it is whether the supporting infrastructure, particularly power, can scale fast enough. ## Sources and References | Source | Reference | |---|---| | Communications Authority of Kenya | Revised Telecommunications Market Structure (June 2026) | | Kenya Gazette Notice No. 3335 | 6 March 2026, DG David Mugonyi | | Bowmans (Law Firm) | Kenya: Revised Telecommunications Market Structure (2026) | | McKay Advocates | Kenya Strengthens ICT Investment Appeal (April 2026) | | Mordor Intelligence | Kenya Data Center Market Size & 2030 Growth Trends (2025) | | ResearchAndMarkets | Kenya Data Center Investment Analysis 2026 | ## Kenya Data Centre Market: USD 266M to USD 805M by 2031 URL: https://data-centers-254.vercel.app/articles/kenya-data-centre-market-266m-to-805m-arizton-outlook Cluster: Market Analysis Published: 2026-09-24 Updated: 2026-09-24 Summary: Arizton's March 2026 forecast puts Kenya's data centre market at USD 266M in 2025 and USD 805M by 2031. We cross-check the numbers against the ground truth. ## A new forecast, and a familiar disagreement The market research firm Arizton published its Kenya data centre market edition in March 2026, and the headline is bold. The market was valued at USD 266 million in 2025 and is projected to reach USD 805 million by 2031, rising at a compound annual growth rate of 20.27%. Numbers like that travel fast in investment decks, and they deserve scrutiny before they harden into "fact". We read the report's public summary, cross-checked every major claim against our own directory and coverage, and found a picture that is directionally right but definition-sensitive. Here is what the forecast says, what it gets right, and where it needs a health warning. ## What sits inside the USD 805 million Arizton's market sizing covers colocation revenue, facility area in square feet and IT load capacity in megawatts, segmented across IT infrastructure, electrical and mechanical systems, and general construction. Most of Kenya's colocation capacity is being built to Tier III standards, the level that supports concurrent maintainability. That segmentation matters when you compare forecasts. Our earlier market outlook for 2025-2030 placed the 2025 market at USD 180-220 million, heading to USD 400-500 million by 2030, and Arizton's numbers sit well above both ends. The difference is not sloppiness on either side; it is what gets counted, and forecasts that blend colocation revenue with the value of services consumed inside data centres will always look bigger. ## Seven cables now, two more coming Connectivity is the quiet foundation of every Kenyan data centre forecast, and Arizton's cable list checks out against our own coverage. The report names seven operational systems: 2Africa, DARE 1, EASSy, LION2, PEACE Cable, SEACOM/Tata TGN-Eurasia and TEAMS. Two more are funded and under way. Africa-1 and Daraja are expected to become operational by 2026-2027, each adding landing capacity on the coast. The pattern matters more than the count: every new cable lowers the marginal cost of international bandwidth and widens the redundancy that hyperscale tenants demand before they commit to a region. ## The 13-facility question Arizton counts 13 operational colocation data centres in Kenya, with Nairobi hosting 8 of them and 7 more upcoming in the city as of September 2025. The DC254 directory, which tracks the region the way operators actually describe themselves, currently lists 23 Kenyan facilities across 14 operators. Neither number is wrong. Arizton's lens is third-party colocation only, which excludes enterprise and in-house facilities and anything outside its September 2025 snapshot window. This is the same methodological gap we documented when explaining why Kenya's market estimates disagree, and it is the first question to ask of any headline figure: what universe was counted? ## Where the money is actually going The vendor landscape in the report reads like our directory's bookmarks bar. Incumbent investors named are Africa Data Centres, iColo (Digital Realty), iXAfrica Data Centres, Safaricom and Telkom Kenya, with Nxtra by Airtel, Cloudoon and G42 flagged as new entrants. The new money is concrete. In September 2025, Airtel's Nxtra started construction on its Nairobi facility, a two-phase build expected to absorb about KSh 19 billion (roughly USD 147 million) with operations targeted for the first quarter of 2027. In the same month, iXAfrica secured financing from Rand Merchant Bank to expand its NBOX1 campus, which had earlier been backed by a USD 50 million Helios investment. ## The snapshot problem: G42 Here is where the forecast needs its health warning. Arizton still lists G42 as a Kenyan new entrant, citing the March 2024 memorandum of understanding with EcoCloud for a green campus that would start at 100 MW of IT capacity and scale toward 1 GW. That was true when the snapshot closed in September 2025. Our own reporting records what happened next: the Microsoft and G42 campus was suspended in May 2026 after the Treasury concept note was never approved. A forecast is only as current as its cutoff, and this entry is now a reminder to date-check every "upcoming" list before it reaches an investment memo. ## Power is the enabler, and Kenya keeps stacking it The report's energy signals align with what we track on this site. In July 2025, KenGen installed a battery energy storage system to power its modular data centre in Nairobi, designed to keep supply stable through periods of low grid demand. The government continues to pilot arrangements that would let data centres purchase renewable energy directly from generators. Kenya's grid already draws more than 90% of its electricity from renewables, anchored by the Olkaria geothermal field. For a market promising a 20% compound growth rate, the binding constraint will not be demand or connectivity; it will be whether power purchase structures and grid build-out keep pace with the rack load being sold. ## How to read this forecast responsibly The honest summary is that Kenya's data centre market is growing fast, that USD 805 million by 2031 is plausible if definitions hold and nothing structural breaks, and that the same sentence carries three conditions. Watch the delivery dates rather than the projections: whether Nxtra hits its Q1 2027 target, whether Africa-1 and Daraja land on schedule, and whether the Communications Authority's proposed standalone data centre licence clarifies or complicates the regulatory ground. For investors, the report's own scope note is the best guidance: 13 existing facilities, 9 upcoming, 4-plus locations. That is the counted universe behind the USD 805 million. For everyone else, our directory and tracker stay live, and the gap between any forecast and the ground truth is exactly what we exist to measure. ## Kenya's Data Centre Market in Numbers (2026) URL: https://data-centers-254.vercel.app/articles/kenya-data-centre-market-numbers Cluster: Market Analysis Published: 2026-09-08 Updated: 2026-09-16 Summary: The verified numbers behind Kenya's data centre market: 27 tracked facilities, six independent estimates compared, and what each measures. Kenya's data centre market generates more confident numbers per shilling of revenue than almost any industry in the country. Market size figures vary by a factor of twenty depending on who is counting. Facility counts vary by a factor of three. Capacity figures get quoted in MW without anyone saying whether they mean IT load, building capacity or a master plan. This page is our reference layer: the numbers we can stand behind, what each one actually measures, and where it came from. For the full reasoning on why the headline market-size figures disagree with each other, read [our companion piece on the estimates](/articles/why-kenya-data-centre-market-estimates-disagree). ![Investment analysis of Kenya's data centre market](/images/dc-business-investment.webp) ## The facility count: 27 verified entries Start with the number everyone argues about. The PeeringDB facility register (the industry's community-maintained, cross-operator database) lists 14 facility records for Kenya as of 8 September 2026, including cable landing stations and one apparent duplicate registration. Our [DC Directory](/directory) tracks 27 verified entries: the register's genuine facilities, plus operator-owned sites (Safaricom Waiyaki Way, Safaricom Red Hill), government facilities (Konza National Data Centre) and announced projects, each labelled by status so operating capacity is never conflated with pipeline. Three facilities dominate the carrier-neutral layer. iColo (a Digital Realty company) runs the largest footprint with four facilities, two in Karen, Nairobi, and two in Mombasa. Africa Data Centres' NBO1 at Sameer Business Park is the most interconnected building in the country, with 122 networks and four internet exchanges registered on-site. iXAfrica's NBOX1 on Mombasa Road is the newest large entrant and the only facility describing itself as hyperscale and AI-ready. PAIX, inside Britam Tower in Upper Hill, rounds out the carrier-neutral four, the smallest by interconnection but the most central. The count matters less than what it excludes. Several well-known round-ups list facilities that fail basic verification, we removed one such entry from our own directory in September 2026 after checking the operator's own website. That is exactly why every entry in our directory carries named sources, a last-verified date and a PeeringDB cross-reference; our [data methodology](/methodology) documents the counting rules in full. ## Capacity: what is actually operating Verified, published operating capacity in Kenya is modest. Summing the facility-level figures operators actually publish (many do not publish any), total facility capacity across operational carrier-neutral and operator sites comes to roughly 28 MW, with individual buildings ranging from 0.9 MW at iColo Mombasa One to 7.5 MW of available site capacity at ADC Nairobi 1. The EU Digital Investment Facility's Xalam-authored Kenya briefing put live IT load at approximately 13 MW, with around 30 MW under construction, as of its most recent edition. Live IT load is always lower than total building capacity, the difference is commissioning, halls not yet fitted out, and headroom. The pipeline is where the numbers get noisy. iXAfrica has announced an 18 MW second building (3,744 racks) to take its campus to a planned 22.5 MW. Africa Data Centres broke ground in January 2023 on a Sameer expansion described as starting at 5 MW and scaling to an additional 15 MW, with no completion confirmation published since. The Microsoft–G42 joint venture, announced at US$1 billion and a 100 MW class, remains stalled on grid power delivery at single-site scale. Every one of these is a plan, not a megawatt, and our directory marks them accordingly. ## The money: market size, honestly scoped The most-cited market size figures for Kenya (every one fetched and dated on 8 September 2026) are: - **~US$25–30 million a year**, Xalam Analytics' estimate of annual *commercial colocation revenue* (the money operators actually earn from selling space and power), published openly in the EU DIF Kenya briefing. This is the only methodology-visible public figure, and it reconciles with verified megawatts at regional pricing. - **US$266 million (2025) → US$805 million (2031)**, Arizton's syndicated estimate, a *total market* scope covering the value chain well beyond colocation. - **US$266.0 million (2025) → US$805.0 million (2031)**, Ken Research's July 2026 report. Compare the decimals: identical to Arizton, to the point, at the same CAGR. Whether by licence, shared upstream data, or circular citation, that is **one data point wearing two logos**, not independent confirmation. - **US$369 million (2025)**, 6Wresearch (July 2026), total-market scope, 11.3% CAGR. - **US$509 million (2024)**, Credence Research, undated on its public page. - **US$606 million (2025) → US$1,240 million (2035)**, DCMarketInsights, undated. - **15 MW → 25 MW (capacity, not dollars)**, Mordor Intelligence (Dec 2025), honest about measuring capacity. A 20x spread is not a rounding disagreement, it is a scope disagreement plus a citation chain, and [we explain the mechanics](/articles/why-kenya-data-centre-market-estimates-disagree) in detail. The practical takeaway: if someone quotes you a Kenya market size, the first question is whether it measures operator revenue or the whole construction-plus-services value chain, and the second is whether the source shows its method. What the reputable sources do agree on is direction. Xalam projects roughly 30% compound annual growth for Kenyan colocation through 2030. Ken Research's July 2026 report projects the total market more than tripling between 2025 and 2031, with commissioned IT load growing from 20 MW (2025) to a projected 90 MW (2031). Occupancy across operating facilities runs at around 60%, healthy, with room to absorb new supply. ## Connectivity: the numbers that actually differentiate buildings In a market where most facilities claim "Tier III" without third-party certification, interconnection is the honest differentiator. The PeeringDB register shows a steep hierarchy: ADC Nairobi 1 leads with 122 networks and four exchanges (KIXP Nairobi, LINX Nairobi, BGP.Exchange and PLUGINS IX all on site); iColo Mombasa One is the densest building on the coast with 94 networks, a function of its proximity to the submarine cable landing stations; iColo Nairobi One carries 62 networks and four exchanges; iXAfrica NBOX1 has 44; PAIX has 37. At the other end, several registered facilities carry fewer than five networks, real buildings, but not yet interconnection destinations. Kenya's submarine position underwrites all of it: seven live cable systems land on the Kenyan coast (SEACOM and TEAMS from 2009, EASSy, LION2, DARE1, PEACE and 2Africa, the largest cable system ever built), with Africa-1 landed and awaiting service, and the Meta-backed Daraja cable, hosted by Safaricom, in development. On [our infrastructure map](/infrastructure/map) you can see each facility, its registered networks and the cable routes together. ## Power: Kenya's unfair advantage and its hard ceiling About 93% of Kenya's electricity generation in 2024 came from renewables (geothermal, hydro and wind) giving workloads hosted in Nairobi one of the cleanest grid mixes of any data centre market anywhere. The grid had roughly 3,192 MW of connected capacity against a peak demand around 2,316 MW as of late 2024 and early 2025: comfortable in aggregate, tight in specific places. That tightness is the practical constraint behind the stalled Microsoft–G42 project, delivering 100+ MW to a single site is a different engineering and commercial problem from serving a city. PUE (power usage effectiveness) is improving as new stock comes online: Ken Research's report puts Kenya's weighted average at around 1.52 historically, moving toward 1.32 in newer builds. iXAfrica claims a campus PUE of 1.25 using free-air cooling, plausible at Nairobi's altitude and temperate climate, and an operator design claim rather than a certified figure. ## How to use these numbers Three rules. First, match the metric to the question: operator revenue for commercial attractiveness, IT load for capacity planning, network counts for interconnection decisions. Second, check the date, figures older than a year in a market growing 30% annually are decoration. Third, check the label: "announced", "under construction" and "operational" are different universes, and any source that blurs them is inflating something. Every figure on this page carries its source and verification date; the [directory](/directory) applies the same discipline to all 27 entries, and we correct in the open when the market moves. ## Kenya Data Centre Market Outlook 2025-2030 URL: https://data-centers-254.vercel.app/articles/kenya-data-centre-market-outlook-2025-2030 Cluster: Market Analysis Published: 2026-08-28 Updated: 2026-09-16 Summary: Kenya's data centre market is projected to grow from $180M to $450M by 2030. The investment drivers, expansions, and opportunities behind the curve. Kenya's data centre market is entering its most significant growth phase since the first commercial facilities were built in Nairobi a decade ago. With an estimated market value of $180–220 million in 2025 and projections pointing toward $400–500 million by 2030, the country is positioning itself as the digital infrastructure capital of East Africa. This is not speculation, it is a trend backed by concrete capacity expansions, new market entrants, and structural demand drivers that are unlikely to reverse. ![Nairobi skyline at sunset](/images/nairobi-sunset-wide.webp) Understanding this market requires looking beyond the headline numbers. The value of a data centre market is not just in the physical facilities themselves, but in the ecosystem of connectivity, power, skilled labour, regulatory frameworks, and demand sources that determine whether those facilities are economically viable. Kenya has strengths in several of these areas and weaknesses in others, and the next five years will determine how quickly the gaps close. ## Current State of the Market in 2025 Kenya currently has approximately 10 operational data centre facilities, concentrated heavily in Nairobi and its immediate environs. The total IT power capacity across these facilities is estimated at 15 megawatts, with the majority housed in three large colocation providers: iXAfrica, Africa Data Centres, and Safaricom. A handful of smaller operator-built and enterprise facilities account for the remainder. The market can be segmented into three tiers. The first tier consists of carrier-neutral colocation providers, iXAfrica's NBOX1 and NBOX1.1 facilities in Mombasa Road, and Africa Data Centres' facilities in Sameer Park and Westlands. These offer retail and wholesale colocation, interconnection to multiple networks, and are the facilities that attract international cloud providers, content delivery networks, and enterprises. The second tier includes telecom-integrated facilities, primarily Safaricom's data centres that support its mobile network, M-Pesa processing, and enterprise cloud services. The third tier comprises smaller, often single-tenant facilities built by banks, government agencies, and technology companies for their own use. ![KICC tower and the Nairobi skyline at night](/images/nairobi-skyline-night-kicc.webp) Revenue in the Kenyan data centre market comes from several streams. Colocation space rental (where customers pay per rack, per kilowatt, or per square metre) is the largest single segment. Interconnection revenue, earned when carriers, cloud providers, and enterprises connect to each other within a facility, is growing faster than colocation itself. Managed services, where the facility operator also manages the customer's servers, storage, and network, commands premium pricing but remains a smaller portion of the market. ancillary services like backup power, cooling, and remote hands round out the revenue mix. Kenya's colocation pricing is competitive by African standards. A full rack in a Tier III Nairobi facility typically costs between KES 60,000 and KES 120,000 per month, depending on power density, location, and contract length. This compares favourably with South Africa, where equivalent space in Johannesburg ranges from KES 80,000 to KES 150,000, and is significantly cheaper than Lagos, where power costs and generator fuel push rack rates above KES 150,000. ## The Key Demand Drivers Several structural forces are driving data centre demand in Kenya, and understanding these is essential for any investment analysis. The first and most significant is enterprise cloud migration. Kenyan banks, telcos, insurance companies, and government agencies are moving workloads from on-premises server rooms to cloud environments, either public clouds (AWS, Azure, Google Cloud) or private clouds hosted in colocation facilities. This transition is still in its early stages. Industry estimates suggest that only 20–30% of Kenyan enterprise workloads have moved to the cloud, compared to 60–70% in mature markets. The remaining migration represents years of demand growth. The second driver is artificial intelligence and machine learning. AI workloads require GPU-dense computing environments that are fundamentally different from traditional server infrastructure. GPU servers consume 4–10 times more power than standard servers and generate correspondingly more heat. This is creating demand for high-density colocation, facilities that can deliver 20–40 kW per rack instead of the traditional 5–10 kW. Kenya's first dedicated GPU-capable facilities are being planned, and the demand from AI startups, research institutions, and international companies running inference for African markets is substantial. ![Nairobi Expressway stretching across the city](/images/nairobi-expressway-wide.webp) The third driver is government digitalisation. Kenya's e-Citizen platform, which now hosts over 5,000 government services, requires domestic data processing. The government's data sovereignty requirements, codified in the 2019 Data Protection Act, mean that certain categories of government and citizen data must be processed within Kenya's borders. This creates a floor of demand that is independent of private sector cycles. The Huduma Namba biometric registration programme, digital tax systems (iTax, TIMS), and the ongoing computerisation of land records, court systems, and health records all contribute to growing government demand for data centre capacity. The fourth driver is financial services modernisation. Kenya's banking sector is one of the most sophisticated in Africa, and the reliance on digital channels (mobile banking, USSD, agent banking) means that every transaction, every account query, every loan application generates data processing demand. The National Payment System alone processes billions of shillings daily, and the backend infrastructure supporting M-Pesa, Airtel Money, and bank-to-bank transfers requires robust, always-on data centre facilities. ## Investment and Expansion Plans The most visible investment in Kenya's data centre market is iXAfrica's NBOX1.1 expansion. iXAfrica, backed by African infrastructure investment funds, opened its first facility (NBOX1) in 2023 with an initial IT load capacity of 2.5 megawatts. NBOX1.1, currently under construction on the same campus along Mombasa Road, will add a further 4.5 megawatts, making the combined campus the largest single data centre site in East Africa at 7 megawatts. The facility is designed to Tier III standards with N+1 redundancy on power and cooling, and will include dedicated GPU-ready zones for AI workloads. Africa Data Centres, a subsidiary of Liquid Intelligent Technologies (now part of the Cassava Technologies group), operates multiple facilities in Nairobi. Their Sameer Business Park facility and their Westlands site together provide several megawatts of IT capacity. The company has publicly stated ambitions to expand its East African footprint, with Kenya as the anchor market. Their advantage is the integration with Liquid's fibre network, which spans over 100,000 kilometres across the continent and provides a built-in customer base of connectivity clients who also need data centre services. ![Westlands commercial district in Nairobi](/images/nairobi-westlands.webp) Safaricom, Kenya's largest telecom operator by revenue and subscriber base, operates data centres that primarily serve its own needs, mobile network switching, M-Pesa transaction processing, and its Cloud and Hosting business unit. Safaricom has been expanding its enterprise cloud offerings, competing directly with international providers by offering local data residency, lower latency, and bundles that combine connectivity, cloud, and security. While Safaricom's facilities are not carrier-neutral colocation in the traditional sense, the company is a significant source of data centre demand in its own right and a potential wholesale partner. Beyond the established players, several new entrants are evaluating the Kenyan market. Kenya Data Centres, a local company, has announced plans for a facility targeting SME colocation customers. Global TLS, a smaller operator, runs a facility in Nairobi focused on the mid-market. The entry of a hyperscale provider (AWS, Microsoft Azure, or Google Cloud building a direct presence rather than partnering through a colocation provider) would be a transformative event for the market, though none has publicly committed to a Nairobi facility as of 2025. ## Kenya vs. Competitor Markets Understanding Kenya's position requires comparing it against Africa's other major data centre markets. South Africa dominates the continent with over 200 megawatts of IT load capacity across dozens of facilities in Johannesburg, Cape Town, and Durban. South Africa's advantage is a large domestic enterprise market, mature financial services sector, and proximity to the rest of Southern Africa. However, South Africa faces power grid challenges (load shedding has been a persistent issue) and higher energy costs compared to Kenya's renewable-heavy grid. Nigeria's market is centred on Lagos, with an estimated 15–20 megawatts of capacity. Nigeria's enormous population (220 million) and large economy create substantial demand, but infrastructure challenges (unreliable power, right-of-way issues for fibre, and regulatory uncertainty) have constrained growth. Nigeria is a high-potential, high-difficulty market. Egypt's market, concentrated in Cairo and the new administrative capital, has grown rapidly with investments from Gulf-based operators. Egypt's advantage is access to European and Asian markets via submarine cables, but its market is more domestically focused and less integrated into the broader African internet ecosystem than Kenya's. ![Rooftop chiller units at a data centre](/images/dc-chillers-roof-2.webp) Kenya's competitive advantages are specific and structural. The country has four submarine cable landing stations in Mombasa (more than any other East African country) providing direct connectivity to Europe, Asia, the Middle East, and Southern Africa. Kenya Power's grid, while not without reliability issues in some areas, draws over 90% from renewable sources (geothermal, hydro, wind, solar), giving Kenya some of the lowest carbon-intensity electricity in the world and a structural cost advantage as global companies face pressure to reduce their carbon footprint. Kenya's strategic location (within 100 milliseconds of latency for the entire East African Community of 300 million people) makes it the natural hub for regional digital services. ## Market Projections 2025–2030 Based on current expansion plans, demand trends, and the structural drivers outlined above, Kenya's data centre market is projected to grow at a compound annual rate of 15–20% between 2025 and 2030. Total IT power capacity is expected to increase from approximately 15 megawatts to 25–30 megawatts, driven primarily by iXAfrica's expansion, Africa Data Centres' growth, and new entrants. Colocation revenue is expected to grow from approximately $100 million to $220–250 million, while managed services and interconnection revenue will grow faster, roughly doubling from $60–80 million to $150–200 million. The total addressable market, including the value of services consumed within data centres (cloud computing, content delivery, financial transaction processing), is substantially larger, potentially $1–2 billion by 2030. The investment required to support this growth is estimated at $300–500 million over the five-year period. This includes new facility construction, expansion of existing sites, power infrastructure upgrades, and the development of specialised capabilities like GPU-ready zones and liquid cooling systems. The returns on this investment are attractive by African infrastructure standards, colocation facilities in Nairobi typically achieve 12–18% unlevered IRR, with higher returns available for operators who can differentiate on services, connectivity, or energy efficiency. The Kenyan government has now put its own weight behind this trajectory. Writing in September 2026, the Principal Secretary for ICT and the Digital Economy, Eng. John Kipchumba Tanui, framed the global context (a colocation and interconnection market worth about US$119 billion in 2025, heading toward roughly US$147 billion in 2026) against Africa's share of it (under 1 percent of global data centre capacity, for a continent holding about 18 percent of the world's population). His essay, which we fact-checked and unpacked in [Kenya's data centre and compute hub plan, explained](/articles/kenya-data-centre-compute-hub-plan-explained), commits the state to converting that gap into operating capacity, which is exactly the demand side this projection assumes. ## Risks and Challenges No investment analysis is complete without an honest assessment of risks. Kenya's data centre market faces several. Power reliability remains a concern, while Nairobi's central business districts and industrial areas have relatively stable supply, the grid's overall reliability has degraded in some peri-urban areas, forcing data centres to rely heavily on diesel generators during outages. This increases operating costs and carbon emissions, partially negating the advantage of Kenya's renewable generation mix. Currency risk is significant. Colocation contracts are typically denominated in Kenya Shillings, but equipment, generator fuel, and many construction materials are priced in US dollars. The KES has depreciated from approximately 100 per USD in 2020 to over 150 per USD in 2025, creating a cost-revenue mismatch that operators must manage through hedging, contractual escalation clauses, or USD-denominated pricing for international clients. The regulatory environment, while improving after the Communications Authority's decision to bring data centres under the [licensing framework](/articles/kenya-data-centre-licensing-framework), still has uncertainties. The KES 15 million NFP-T2 licence fee is manageable for established operators but may be a barrier for new entrants or smaller facilities. Tax policy, including VAT treatment of data centre services and import duties on equipment, also affects competitiveness. Finally, the skills gap is a real constraint. Kenya produces many software engineers and IT graduates, but the specialised skills required to design, build, and operate data centres (power engineering, cooling system management, structured cabling, facility management) are in short supply. Operators are investing in training programmes, but scaling these to support a 2x growth in capacity over five years is a non-trivial challenge. ## The Opportunity Despite these risks, the investment case for Kenyan data centres is compelling. The market is growing from a small base, demand drivers are structural and multi-year, and the competitive landscape is less crowded than in South Africa or Nigeria. Kenya's renewable energy advantage will become increasingly valuable as global companies face stricter environmental reporting requirements and carbon reduction targets. The country's connectivity advantage (four submarine cables, a well-functioning internet exchange point, and a growing fibre network) makes it the natural digital hub for a region of 300 million people. For investors, the opportunity is not just in building facilities, but in the entire value chain: connectivity, managed services, edge computing, and the emerging GPU infrastructure needed for AI workloads. The next five years will determine whether Kenya cements its position as East Africa's data centre capital, or cedes ground to faster-moving competitors. All current indicators point toward growth. ## Kenya Data Centre Market Update: 15 October 2026 URL: https://data-centers-254.vercel.app/articles/kenya-data-centre-market-update-15-october-2026 Cluster: Market Analysis Published: 2026-10-15 Updated: 2026-10-15 Summary: Verified signals for 15 Oct 2026: Tanzania's $42bn LNG deal nears final agreements, AI infrastructure debt gets repriced, Kenya's tracker holds at 27 facilities. **News assessment, 15 October 2026: [Unverified]** No clearly verified new Kenyan data-centre opening or operational-capacity announcement surfaced in today's search. The strongest signals concern regional energy investment and the rising financial risks of AI infrastructure. The Kenya tracker holds at its verified baseline: 27 facilities, 20 operational, designed capacity of about 42.9 MW at operational sites, published live IT load of about 10.45 MW ([tracker](/tracker), dataset 2026-Q3, last checked 2026-09). ## Signal 1: Tanzania advances its $42 billion LNG project | Field | Reading | |---|---| | Project | Tanzania LNG, led by Equinor and Shell | | Location | Tanzania (offshore gas, LNG export plant) | | Status | Negotiations; final agreements not yet confirmed | | What changed | Finance minister says the $42 billion deal is close to finalisation | | Evidence | Reuters, 13 Oct 2026; Africa Business Insider, 14 Oct 2026 | | Supplier relevance | Regional energy economics, not direct DC procurement | | Confidence | Verified (deal pending); Reported (timing) | | Next question | How much new generation capacity is actually contracted, and for whom? | | Last checked | 2026-10-15 | Tanzania is nearing completion of agreements for a proposed $42 billion liquefied natural gas project, with the government also reaffirming a target of 8,000 MW of electricity-generation capacity by 2030 and renewables at 75% of the mix (Reuters, 13 Oct; Newsbase, 9 Oct). The project has missed earlier target dates, including a June 2026 window, so "close" here means closer than last year, not signed. **Why it matters to DC254:** energy infrastructure shapes the long-term competitiveness of East African digital infrastructure. A larger, cheaper regional power pool would strengthen the case for energy-hungry compute in the corridor from Mombasa inland. But national generation targets do not establish how much electricity will be available to data centres, at what tariff, or when. The signal is context for the [power economics](/tracker/power) behind the region's build-out, and it is labelled accordingly. ## Signal 2: Credit markets reprice AI infrastructure debt | Field | Reading | |---|---| | Project | AI infrastructure financing (global) | | Location | Global markets, with implications for African investment | | Status | Confirmed market development | | What changed | Oracle's five-year credit-default-swap spread hit a record 288 basis points; CoreWeave's rose past 1,000 | | Evidence | Reuters, 15 Oct 2026 | | Supplier relevance | Anyone selling into projects funded by AI-linked capital | | Confidence | Verified | | Next question | Which African projects have committed customers and secured power behind their financing? | | Last checked | 2026-10-15 | Reuters reports that the cost of insuring against defaults in parts of the AI sector rose sharply this week: Oracle's five-year CDS spread widened to a record 288 basis points, and CoreWeave, the AI cloud provider carrying heavy debt, saw a much larger increase past 1,000 basis points. Investors are questioning the relationship between infrastructure spending, debt and future returns. **Why it matters to DC254:** African data-centre projects also depend on long-term capital commitments, and this market has seen investment disputes before ([examined here](/articles/africa-data-centre-investment-disputes)). When global credit reprice, the projects that survive scrutiny are the ones with documented power arrangements, credible delivery milestones and committed customers. That raises the value of exactly the discipline this site applies: keep announced capacity separate from verified operational capacity, and treat financing announcements as claims until documents exist ([the IMF's power-constraint analysis](/articles/africa-160-data-centres-imf-power-constraint) makes the same point from the demand side). ## Also on the radar Three items crossed the wire today that belong in a supplier's peripheral vision rather than the tracker. A new Africa AI Council is forming to scrutinise technology policy ahead of Kenya's 2027 general election (KBC, 15 Oct), which matters because AI policy posture is becoming an investment signal in its own right. DAMA Kenya formally launched to strengthen data governance practice (CIO Africa, 15 Oct); governance maturity is a quiet precondition for the compliance-sensitive workloads data centres want to attract. And Vertiv's managing director for Africa, Wojtek Piorko, gave a structured interview on AI reshaping data-centre requirements and infrastructure readiness (TechCabal, 14 Oct), worth reading for how a major equipment vendor frames the same power-readiness questions this site tracks. None of the three changes a facility's verified status. ## What this changes on the tracker Nothing today, and that is the point. Reporting priority on DC254 is to update the [Kenya tracker](/tracker) only when new evidence changes a facility's status, capacity or power-readiness assessment. A quiet day for verification is still a loud day for financing signals, and conflating the two is how market noise gets mistaken for market progress. Suppliers who need the verified picture without the daily noise can check the [directory](/directory) any time; the announcement-versus-evidence gap is exactly what the [Opportunity Brief](/opportunity-brief) exists to convert into targeting decisions. ## The reporting angle this update feeds The week's signals sharpen one question for the next verification cycle: **who is funding the power behind the promise?** Three checks follow from it. Which announced Kenyan projects have secured financing that can be verified in filings? Which have documented power arrangements and delivery milestones rather than memoranda of understanding? And how much proposed capacity has moved into construction, commissioning or operation, measured against the 42.9 MW operational baseline? Suppliers watching specific accounts should hold vendors to the same standard: ask for the evidence, not the press release. ## Kenya Data-Centre Boom: What Should Operators Disclose? URL: https://data-centers-254.vercel.app/articles/kenya-data-centres-what-operators-should-disclose Cluster: Policy Published: 2026-10-10 Updated: 2026-10-10 Summary: Kenya is about to license data centres directly. We checked what operators publish today on capacity, power and water, and what disclosure should require. Kenya's data centre market has crossed the point where the interesting question is no longer how much is being built. Twenty of the 27 facilities on the DC254 verified board are operational, carrying 42.9 MW of designed capacity, with 10.5 MW of published live IT load between them. Another 230 MW sits in the pipeline at various stages of commitment, from the 77 MW rising out of the ground in three under-construction projects to the 100 MW stated in the Microsoft and G42 announcement that remains at the intention stage. On 8 September 2026, the Communications Authority opened a consultation on a standalone data centre licence, and with it, the question changed again: not just what gets built, but what the industry should have to say about itself in public. Because that is the uncomfortable part of the current record. The market publishes capacity enthusiastically and discloses almost nothing else. Operators state design figures, certification lists and connectivity counts. Almost none publish what a building actually draws from the grid, what it drinks, or how those numbers change as the hall fills. This piece asks a simple question of the verified record: what can a reader establish today about a Kenyan data centre beyond the press release, and what should a licensing framework require to be on the table? ## A licence framework puts disclosure on the agenda On 8 September 2026 the Communications Authority of Kenya published a notice proposing a standalone data centre licence category, replacing the current treatment of colocation operators under the Network Facilities Provider Tier 2 regime. The Authority's stated rationale is to "provide regulatory clarity, enhance visibility over data centre operations, support investment in digital infrastructure, and align Kenya's framework with proportionate approaches adopted in comparable jurisdictions." The proposed licence would cover "entities that provide colocation data centre services, including the attendant supporting services." ![Institutional building at dusk, the standing imagery of regulatory authority](/images/dc-policy-regulation.webp) The proposal sat on the Authority's open consultations page through our verified captures of 24 September, 4 October and 6 October, with a comment window of thirty days from publication, closing on or about 8 October 2026. Outcome searches we ran on 10 October, including sweeps for a gazette supplement and industry-association submissions, found no decision notice; the most recent coverage on record is still from the proposal window itself. What the final framework contains, fees, obligations, and critically whether it attaches any disclosure or reporting requirements to the licence, is not yet public. We will report it with the instrument captured the day it is. The advisory market is not waiting. AfriTrade Group published commentary on the framework in mid-September, and on 9 October Access Partnership and AfriTrade announced a joint practice covering digital infrastructure across Kenya and the wider East African region, with data centre approvals named in the initial scope (TechAfrica News, ITWeb Africa, Capital FM, 8–9 October 2026). When the consultants mobilise before the rules are final, it is a fair signal that compliance, and the paperwork around it, is about to become a real cost line. ## Five questions a reader should be able to answer The framework under discussion is a licensing instrument, but the debate around it has converged on a disclosure question. Five indicators keep recurring in how this market is argued about, and it is worth being precise about where the verified record stands on each of them for Kenya today. | Indicator | What the verified record shows today | What is not published | |---|---|---| | Operational status | Fully answerable. Every one of the 27 tracked facilities carries a status, operational, under construction, committed or early stage, with its own verification date and source list | Nothing material; status is the sector's best-published fact | | IT capacity | Partially answerable. 42.9 MW of designed capacity across the 20 operational facilities; 10.5 MW of published live IT load; capacity figures labelled by type on the DC254 board | Per-facility live IT load for most operational sites, and a stated measurement basis for several headline figures | | Electricity | Thin. Power architecture claims (2N systems, N+2 cooling) and a handful of design statements, such as NBO2's master plan substation above 20 MW at full build | Actual consumption by any tracked facility, contracted supply volumes, and load factor over time | | Water | Thinner still. Cooling method is occasionally stated as an operator claim | Consumption, source and treatment for every tracked facility, there is no published Kenyan water figure in the dataset | | Regulatory status | Partially answerable. Licence categories sit on CA registers, and the standalone licence proposal is documented through its consultation trail | The final framework terms, fees and any disclosure obligations the licence will carry | The pattern is consistent: the industry publishes what helps it sell, and the reader fills the rest with assumptions. A licensing moment is the natural point to reset that, because a licence is enforced paperwork, and paperwork is disclosure by another name. ## What "6.4 MW" actually tells you The measurement-basis problem is easiest to see in a single live example. When Digital Realty announced the opening of NBO2 in Nairobi on 7 September 2026, the announcement's headline figure was "6.4 MW of capacity", with no statement of whether that is commissioned IT load, installed design capacity, or a campus-level planning number. An earlier figure of 6.5 MW of IT load, from Developing Telecoms in August 2024 and repeated by the operator before launch, sits on the same record as a documented pre-launch discrepancy. The DC254 record carries both, labels the 6.4 MW as designed capacity, and leaves live IT load deliberately empty because no commissioned figure has been published. ![Officials cutting the ribbon at the NBO2 data centre launch in Nairobi](/images/nbo2-launch-ribbon-cutting.webp) Down the road on the same Karen campus, NBO1, the facility PeeringDB shows carrying more than 60 networks and four internet exchanges, publishes no IT load figure at all. The operator's own listing does not state one, and the DC254 divergence note for the record says exactly that: ask for committed and available capacity in kW before sizing a deployment. None of this is unusual, and that is the point. Across the 27 tracked records there are 29 standing divergence notes, each recording a place where the public record conflicts, drifts or goes quiet. The directory's capacity chips, LIVE, DESIGNED, PLANNED, ANNOUNCED, exist because a bare megawatt number, repeated without its basis, is the single most common misreading in this market. Disclosure policy should start from the same insight: a capacity figure without a stated measurement basis is not transparency, it is marketing with a decimal point. ## Electricity: the number everyone wants, nobody publishes Power is the first question every serious buyer, investor and regulator asks about a data centre, and it is where Kenya's public record is thinnest relative to interest. What exists is architecture and design talk: 2N power distribution claims, N+2 cooling, and design statements like the NBO2 master plan's captive substation consuming over 20 MW at full build. What does not exist, for any tracked facility, is a consumption figure, how many megawatt-hours the building actually drew last year, against what contracted supply, at what load factor. ![Power distribution equipment inside a data centre facility](/images/dc-power-systems.webp) The closest the public record comes to a power story is the one attached to the market's most-announced project. The Microsoft and G42 joint venture, stated at 100 MW and a billion US dollars, remains an early-stage record on the DC254 board, and our power-corridor tracking notes the single-site power-delivery constraint that has been reported around it. Geothermal from Olkaria is discussed as the primary source, discussed being the operative word, since no PPA, no grid agreement and no consumption profile has been published for verification. This is where the licence consultation matters most. Grid-connected load of this scale is a planning question for Kenyan institutions, and consumption data is exactly what a licensing or reporting framework could require operators to file, even if commercial confidentiality keeps the raw numbers out of public view, an aggregate or audited disclosure would transform the quality of public debate. ## Water and cooling: almost nothing on the record If electricity is thin, water is absent. Cooling method occasionally appears as an operator claim, the tracked Kenya records variously state N+2 cooling and similar architectures, but consumption, source and treatment are unpublished across the board. There is no published Kenyan water figure in the verified dataset, not because nobody has looked, but because no operator has put one on the record. ![Cooling plant pipework serving a data centre white space](/images/dc-cooling-plant-pipes.webp) The regional context makes this worth attention now rather than later. In South Africa, civil rights groups have called for a halt to new data centre construction amid water and power concerns; the country's Human Rights Commission told the Associated Press it had received more than 250 submissions after inviting public input in May, and President Ramaphosa's push to attract further digital infrastructure investment now runs alongside that scrutiny (AP, 4 September 2026). The claims in that debate are contested and, on several points, unverified, which is precisely the problem. A market that had published consumption figures per facility would be arguing with numbers instead of about them. Kenya has the advantage of watching that argument arrive before its own build-out peaks. The proposed licence framework is the instrument that could require resource-use disclosure from the start, for every licensee, on a consistent basis, the difference between designing transparency in and retrofitting it under pressure. ## What good disclosure would look like Between the verified record and the consultation, a short checklist is coming into focus, five things a licensed Kenyan data centre could reasonably be expected to state, each currently missing or unstandardised in public reporting: 1. **Status, dated.** The sector already does this well; keep it, and make the verification date part of the licence record. 2. **Capacity with a basis.** Every published megawatt figure labelled as commissioned IT load, installed design capacity or planned build, the distinction the NBO2 announcement shows is still optional today. 3. **Electricity in and out.** Contracted supply and annual consumption, even in aggregate or audited form, so the grid question can be argued with numbers. 4. **Water by the litre.** Consumption, source and cooling method, the three inputs to every water-use argument the South African debate is currently having without data. 5. **Regulatory position.** Licence category, conditions and any reporting obligations, published by the regulator as a register rather than assembled piecemeal by researchers. None of this is exotic. Most of it is what operators already compute internally, and several regional peers have begun publishing pieces of it voluntarily. The cost of standardisation is small; the cost of a contested, data-free public argument, the South African path, is larger, and Kenya can still choose to skip it. ### Does the proposed licence actually require any of this? Unknown, honestly. The consultation text we captured proposes the licence category and its scope; the reporting obligations that would attach to it have not been published. We will document the final framework's disclosure terms, or its silence on them, as soon as the instrument is out. ### Is the market hiding something, then? No, that is not the finding. Published design capacity, certification claims and connectivity counts are real, and this market publishes more than most on the continent. The gap is specific: consumption, water and measurement bases are not standardised, so the reader cannot compare facilities or hold the sector to its own numbers. A gap is an evidence problem, not an accusation. ### How is DC254's own record kept honest? Every facility record carries its own verification date, its sources, and a divergence note wherever the public record conflicts. Records are re-verified on rolling dates, so a September date beside an October one is the system working, not drift. Where we cannot verify a figure, commissioned IT load at most operational sites, water everywhere, the field is left empty and the gap is stated. --- The bottom line: Kenya is about to hand a growing industry a licence, and the framework's drafting is the cheapest moment it will ever have to decide what the public record of that industry should contain. The verified board shows a market that publishes capacity well and consumption not at all. Five disclosure lines would fix that, and every one of them is already being computed somewhere inside the operator's own systems. All figures in this piece come from the DC254 verified dataset, captured operator announcements and regulator documents, with evidence gaps marked as gaps, the standard we would like to see the licence require of everyone. ## Kenya Data Protection Act: What Data Centres Must Know URL: https://data-centers-254.vercel.app/articles/kenya-data-protection-act-data-centres Cluster: Policy Published: 2026-08-28 Updated: 2026-08-28 Summary: Kenya's Data Protection Act 2019 sets real obligations for data centres: consent, localisation, breach notification. What it means for operators. When Kenya's Data Protection Act came into effect in November 2019, it fundamentally changed the legal landscape for any organisation that handles personal data, and that includes every data centre operator and every customer who places servers in a Kenyan facility. The Act, modelled closely on the European Union's General Data Protection Regulation (GDPR), established rights for individuals, obligations for organisations, and enforcement powers for the Office of the Data Protection Commissioner (ODPC) that, in principle, match the standards set by the world's most comprehensive data protection laws. ![The Constitution of Kenya 2010, whose Article 31 privacy guarantee underpins the Data Protection Act](/images/constitution-of-kenya-2010.webp) In practice, Kenya's data protection regime is still maturing. The ODPC, established in 2020, has been building its institutional capacity gradually. Enforcement actions have been limited, guidance documents are still being developed, and many organisations remain unclear on exactly what compliance requires. But the law is the law, and for data centre operators and their customers, understanding the Data Protection Act is not optional, it is a legal obligation with financial penalties for non-compliance of up to KES 5 million or 1% of annual turnover, whichever is higher. ## How the Act Classifies Data Centre Participants The Data Protection Act uses two key roles that map directly onto the data centre industry. The **data controller** is the entity that determines the purposes and means of processing personal data. In a data centre context, the data controller is typically the customer, the bank that decides what customer data to store, how to process it, and why. The **data processor** is the entity that processes personal data on behalf of the controller. The data centre operator, which provides the physical infrastructure to store and process the controller's data, is a data processor. This distinction matters because the Act imposes different obligations on controllers and processors. The controller bears the primary responsibility for ensuring lawful processing, obtaining consent, respecting data subject rights, and conducting data protection impact assessments. The processor must process data only as instructed by the controller, implement appropriate security measures, and support the controller's compliance obligations. A data centre operator's obligations as a processor are defined by its contract with the customer. A well-drafted data processing agreement (DPA) between a data centre and its customer should specify what data the data centre may access, what security measures it must implement, what happens in the event of a breach, and what happens to the data when the contract ends. Without a DPA, both parties are exposed to legal risk. ## Key Obligations for Data Centre Operators ### Security Measures The Act requires data processors to implement "appropriate technical and organisational measures" to protect personal data. For a data centre, this translates directly into the security infrastructure discussed in [our security guide](/articles/data-centre-security-explained): physical access controls (biometrics, mantraps, CCTV), network security (firewalls, intrusion detection, encryption), and environmental controls (fire suppression, climate control). A data centre that cannot demonstrate these measures is not just a security risk, it is a compliance risk. ISO 27001 certification, covered in our [ISO 27001 for Kenyan data centres](/articles/iso-27001-data-centre-kenya) guide, is the most widely accepted way to demonstrate exactly this. ### Processing Only as Instructed A data centre must not access, copy, or process customer data beyond what is necessary to provide the contracted services. When a technician performs remote hands services (physically pressing a button or checking a status light on a customer's server) they must not access the data on that server. Data centre operators enforce this through access controls (technicians cannot log into customer servers), monitoring (CCTV and access logs record all physical access to equipment), and contractual provisions (the master service agreement and DPA prohibit unauthorised access). ![Biometric access control at a data centre](/images/dc-biometric-access.webp) ### Supporting Breach Notification If a data centre experiences a security incident that affects customer data, whether a physical breach (unauthorised access to the server room), a cyber attack, or an equipment failure that exposes data, it must notify the affected customer promptly. The customer, as data controller, then has 72 hours to assess whether the breach poses a risk to data subjects and, if so, to notify the ODPC. A data centre that delays notifying its customer, or fails to detect a breach in the first place, could jeopardise the customer's ability to meet the 72-hour notification requirement. ### Sub-processor Management If a data centre engages third parties (security contractors, cleaning staff, maintenance technicians) who may have access to areas where personal data is stored or processed, it must ensure those sub-processors provide equivalent data protection. The data centre remains liable to its customer for the actions of its sub-processors, so vetting, contracting, and monitoring third parties is a compliance requirement, not just a security best practice. ## Cross-Border Data Transfers One of the most practically significant aspects of the Data Protection Act for data centre customers is the restriction on cross-border data transfers. Section 48 of the Act provides that personal data shall not be transferred outside Kenya unless the recipient country has been assessed by the ODPC as having an adequate level of data protection, or unless one of several exceptions applies (the data subject has consented, the transfer is necessary for a contract, the transfer is necessary for important reasons of public interest, or the transfer is made through binding corporate rules or approved codes of conduct). ![Blue-lit server racks in a data hall](/images/hero-server-hall.webp) As of 2025, the ODPC has not published a comprehensive list of countries deemed to have adequate data protection. The EU (via GDPR), the United Kingdom (via UK GDPR), and several other jurisdictions would likely qualify, but the absence of a formal adequacy determination creates uncertainty. In practice, many Kenyan organisations interpret this requirement conservatively and choose to keep personal data within Kenya's borders, which drives demand for Kenyan data centre capacity. For data centre operators, this provision is a commercial opportunity. The Data Protection Act creates a regulatory incentive for organisations to use Kenyan data centres rather than hosting in South Africa, Europe, or the cloud. A colocation facility that can demonstrate compliance with the Act, including through certifications like ISO 27001, becomes more attractive to customers who need to keep data in Kenya for regulatory reasons. ## Data Subject Rights and Their Impact on Data Centres The Act grants Kenyan data subjects several rights that, while primarily the responsibility of the data controller, can affect data centre operations. The right of access means a data subject can request a copy of their personal data. If the data is stored on servers in a data centre, the controller may need the data centre's assistance to access and extract it. The right to erasure ("the right to be forgotten") means a data subject can request deletion of their personal data. When data is stored on physical drives in a data centre, secure deletion requires the data centre's cooperation, either by the customer remotely wiping the data, or by the data centre performing physical drive destruction or secure erasure on the customer's instructions. The right to data portability means data subjects can request their data in a structured, commonly used format. For data stored in a data centre, this may involve the data centre providing access to the physical drives or supporting the transfer of data to a different facility. These rights create operational obligations for data centres that go beyond simply providing power, cooling, and connectivity. ## Registration and Compliance The Data Protection Act requires data controllers and processors to register with the ODPC. Data centre operators must register as data processors, and their customers (banks, telcos, government agencies) must register as data controllers. Registration involves providing details about the organisation, the types of data processed, the purposes of processing, and the security measures in place. ![Security cameras monitoring a colocation facility](/images/dc-security-camera.webp) Beyond registration, compliance requires several ongoing activities. Data protection impact assessments (DPIAs) must be conducted for processing activities that are likely to result in a high risk to data subjects. Records of processing activities must be maintained. Staff who handle personal data must receive data protection training. And data centres must maintain documentation of their security measures, breach response procedures, and sub-processor agreements. ## The Practical Reality Kenya's Data Protection Act is well-drafted legislation that, on paper, provides protections comparable to GDPR. The practical reality is that enforcement is still developing, many organisations are in early stages of compliance, and the ODPC has limited resources relative to the scale of its mandate. However, this is changing. The ODPC has issued guidance notices, conducted compliance assessments of government agencies, and is building its enforcement capacity. For data centre operators, the message is clear: compliance is not a future concern, it is a current obligation. The operators that invest in compliance now (through ISO 27001 certification, robust data processing agreements, staff training, and security measures) will have a competitive advantage as enforcement intensifies. Those that treat compliance as an afterthought will face increasing legal, commercial, and reputational risk. The Data Protection Act is not just a legal requirement; it is becoming a market differentiator in Kenya's data centre industry. ## Kenya Data Sovereignty and Data Localisation URL: https://data-centers-254.vercel.app/articles/kenya-data-sovereignty-localisation Cluster: Policy Published: 2026-08-28 Updated: 2026-09-16 Summary: Kenya's data sovereignty rules under the Data Protection Act 2019: cross-border transfer limits, sector obligations, and data centre demand. ## What Data Sovereignty Means in the Kenyan Context ![KICC tower and the Nairobi skyline at night](/images/nairobi-skyline-night-kicc.webp) **Data sovereignty** is the foundational principle that data generated within a nation's borders (or data pertaining to its citizens) is subject to that nation's laws and regulatory authority. In Kenya, this concept has gained significant traction since the enactment of the **Data Protection Act, 2019**, which established a comprehensive legal framework for the processing of personal data. Unlike a simple data localisation mandate that dictates physical storage location, data sovereignty encompasses the broader idea that **Kenyan law governs data about Kenyans**, regardless of where that data is processed or stored. For data centre operators in Kenya, understanding this distinction is critical. Data sovereignty creates a legal jurisdiction claim, while data localisation creates a physical infrastructure requirement. Kenya's primary legislation leans toward sovereignty, but sector-specific regulators increasingly push toward localisation, and that distinction has direct implications for where servers must be located and how data flows across borders. This article examines the legal provisions, regulatory guidance, sector-specific requirements, comparative frameworks, and the economic impact of these rules on Kenya's data centre industry. ## The Data Protection Act 2019: Cross-Border Transfer Provisions The **Data Protection Act, 2019** (DPA) is Kenya's primary data protection legislation, modelled substantially on the European Union's General Data Protection Regulation (GDPR). The Act established the **Office of the Data Protection Commissioner (ODPC)** as the supervisory authority, charged with enforcement, registration of data controllers and processors, and issuance of guidance. ### Sections 48–50: The Legal Framework for Cross-Border Transfers Sections 48, 49, and 50 of the DPA govern the transfer of personal data outside Kenya. Together, they establish a layered framework: - **Section 48** restricts cross-border transfers to jurisdictions that the ODPC has determined provide an **adequate level of data protection**, or to territories where appropriate safeguards exist. - **Section 49** permits transfers based on **contractual arrangements** between the data exporter and the data importer, provided those arrangements include binding commitments to protect the data to standards equivalent to those under the DPA. - **Section 50** allows transfers in specific circumstances, including where the **data subject has given explicit, informed consent**, or where the transfer is necessary for the performance of a contract, legal obligation, or vital interests of the data subject. The Act does **not** impose a blanket prohibition on transferring data outside Kenya. Instead, it creates a conditional framework, data may leave the country, but only under controlled circumstances that ensure the data subject's rights are not undermined. ### ODPC Guidance on Cross-Border Transfers The [Office of the Data Protection Commissioner](https://www.odpc.go.ke) has issued guidance notes to data controllers and processors on how to comply with these provisions. Key elements include: - **Transfer Impact Assessments (TIAs)**: Data controllers are expected to assess the legal and practical data protection landscape of the destination jurisdiction before initiating a transfer. - **Registration obligations**: All data controllers and processors operating in Kenya must register with the ODPC, and cross-border processing arrangements must be disclosed during registration. - **Record-keeping requirements**: Organisations must maintain records of cross-border transfers, including the legal basis, the recipient jurisdiction, and the safeguards applied. As of 2026, the ODPC has not yet published a formal **adequacy whitelist**, a list of countries deemed to have sufficient data protection standards. This means that data controllers must rely heavily on contractual safeguards, consent mechanisms, and their own risk assessments to lawfully transfer data abroad. ## Data Sovereignty vs Data Localisation: Understanding the Distinction The terms **data sovereignty** and **data localisation** are frequently used interchangeably in public discourse, but they represent fundamentally different regulatory approaches. **Data sovereignty** is the principle that data is subject to the laws of the jurisdiction in which it is collected or where the data subject resides. It is a legal concept, it is about **which laws apply**, not about where data physically sits. A company could store Kenyan citizen data in a data centre in Germany but still be subject to Kenyan data protection law. **Data localisation** is a more prescriptive requirement that data must be **physically stored and/or processed within a country's borders**. It is an infrastructure concept, it mandates that servers, storage systems, and processing facilities be located domestically. Localisation requirements directly drive demand for in-country data centre capacity. Kenya's Data Protection Act is primarily a **data sovereignty** instrument. It asserts legal jurisdiction over personal data relating to Kenyan citizens and residents. However, as explored below, certain sector regulators have effectively introduced **data localisation** requirements through their own regulatory instruments. ## Sector-Specific Localisation Expectations in Kenya While the DPA takes a sovereignty-focused approach, several Kenyan regulators have imposed stricter, sector-specific requirements that amount to de facto data localisation. ### Banking and Financial Services, CBK Guidelines The **Central Bank of Kenya (CBK)** has been the most assertive regulator on data localisation. Through its **Prudential Guidelines on ICT Risk Management** and related circulars, the CBK requires: - **Core banking systems and customer data** must be hosted within Kenya. - Financial institutions must obtain **prior CBK approval** before engaging any offshore cloud service provider or data processing arrangement. - A comprehensive **risk assessment** must demonstrate that the offshore arrangement does not compromise the confidentiality, integrity, or availability of financial data. - Banks must maintain **business continuity and disaster recovery** capabilities within Kenya. These requirements have significant implications for multinational banks operating in Kenya and for local institutions considering cloud migration. For data centre operators, the CBK's stance translates into sustained demand for **tier-certified colocation facilities** within Nairobi and its environs. You can read more about how these compliance demands shape facility requirements in our article on [Kenya's data protection framework for data centres](/articles/kenya-data-protection-act-data-centres). ### Health Records and Patient Data Health data is classified as **sensitive personal data** under Section 31 of the DPA, subject to heightened protections. The **Ministry of Health** has issued guidelines through the Kenya Health Information Systems framework, which strongly encourage (and in some cases require) that patient records, clinical data, and health management information systems be hosted on servers located within Kenya. The **Kenya Medical Research Institute (KEMRI)** and institutions handling clinical trial data face particularly stringent requirements, given the sensitivity of medical research data and the ethical obligations around participant confidentiality. ### Government Data, ICT Authority Directives The **ICT Authority** of Kenya, operating under the Ministry of Information, Communications and the Digital Economy, has issued directives requiring that **government data and e-government services** be hosted on Kenyan soil. The **Kenya Cloud Government (KCG) initiative**, managed through the [ICT Authority's partnerships](https://www.icta.go.ke) with local data centre operators, is designed to consolidate government IT infrastructure into locally hosted, secure facilities. This policy has been a major driver of data centre investment. Government agencies are required to migrate from offshore hosting arrangements to local data centres, creating a substantial and predictable demand pipeline for facilities operated by companies such as **Liquid Intelligent Technologies**, **Africa Data Centres**, and **MDXi**. ### Telecommunications, CAK Requirements The **Communications Authority of Kenya (CAK)** regulates telecommunications providers and has imposed conditions on licence holders regarding the handling of subscriber data, call detail records (CDRs), and other communications metadata. While not a strict localisation mandate, CAK's requirements effectively necessitate that this data be processed and stored within accessible Kenyan jurisdiction for law enforcement and national security purposes. ![Power hall inside a Kenyan data centre during a working site visit](/images/site-visit-power-hall-kenya.webp) ## Comparative Analysis: How Kenya Stacks Up Understanding Kenya's position requires examining how other jurisdictions approach the same issues. ### Nigeria, NDPR and NDPA Nigeria's **National Data Protection Regulation (NDPR, 2019)** and its successor, the **Nigeria Data Protection Act (NDPA, 2023)**, regulate cross-border transfers and require data controllers to conduct transfer impact assessments. The NDPA is generally considered more prescriptive than Kenya's DPA, with specific requirements for data localisation in certain sectors. Nigeria's approach has also been more aggressive in enforcement, with the Nigeria Data Protection Commission issuing significant fines for non-compliance. ### South Africa, POPIA South Africa's **Protection of Personal Information Act (POPIA, 2020)** takes a similar approach to Kenya's DPA, restricting cross-border transfers unless the recipient jurisdiction has adequate protections or appropriate safeguards are in place. POPIA is enforced by the **Information Regulator**, and South Africa has been more active in developing guidance on adequacy assessments. However, like Kenya, South Africa has not yet published a formal adequacy whitelist. ### European Union, GDPR The **EU's GDPR** is the global benchmark for data protection and heavily influenced Kenya's DPA. The GDPR permits cross-border transfers based on adequacy decisions (the EU maintains an official list of adequate countries), Standard Contractual Clauses (SCCs), binding corporate rules, or derogations. Kenya's framework mirrors this structure but operates at an earlier stage of regulatory maturity. Notably, the GDPR does not impose data localisation (it is a sovereignty instrument) and Kenya has followed the same philosophical approach at the national level. For a deeper comparative analysis of data centre regulations across the continent, see our article on [Africa's data centre regulation compared](/articles/africa-data-centre-regulation-compared). ## The Economic Argument: Data Localisation Driving Data Centre Demand One of the most significant consequences of Kenya's evolving data sovereignty and localisation framework is its impact on the **data centre market**. Even though the DPA itself does not impose strict localisation, the combined effect of sector-specific requirements from the CBK, the Ministry of Health, the ICT Authority, and CAK has created a substantial structural demand for in-country data centre capacity. This demand manifests in several ways: - **Colocation growth**: Banks, telecoms, government agencies, and health organisations need physical rack space in Kenyan facilities. This has driven occupancy rates upward at major Nairobi facilities. - **Hyperscale interest**: The regulatory environment, combined with Kenya's strategic position as an East African digital hub and the landing point for multiple submarine cables in Mombasa, makes Kenya an attractive market for potential hyperscale investments. - **Local cloud growth**: Kenyan cloud service providers such as **Africa's Talking**, **Safaricom Cloud**, and **Ilara Health** (for health-tech) benefit from the preference for local data hosting, as compliance-conscious organisations choose domestic providers over international alternatives. According to market projections covered in our [Kenya data centre market outlook 2025–2030](/articles/kenya-data-centre-market-outlook-2025-2030), regulatory compliance is expected to be one of the top three demand drivers for new data centre capacity in Kenya through the end of the decade. The government's own economic framing now says the same thing. In September 2026, the Principal Secretary for ICT and the Digital Economy warned that if African data continues to be processed mainly outside the continent, Africa will export the raw material of its digital economy and import the resulting intelligence, platforms and services. That argument, from the essay we unpack in [Kenya's data centre and compute hub plan, explained](/articles/kenya-data-centre-compute-hub-plan-explained), turns localisation from a compliance burden into an industrial strategy: the point is not only to keep data in Kenya, but to build the Kenyan compute that does something valuable with it. ## How Multinational Cloud Providers Navigate Kenya's Rules The absence of a dedicated cloud region in Kenya creates a practical compliance challenge for multinational providers. - **AWS** operates its closest region in **Cape Town, South Africa** (af-south-1), launched in April 2020. Kenyan organisations using AWS must rely on contractual safeguards, SCCs, and consent to justify the cross-border transfer. - **Microsoft Azure** offers regions in **South Africa North** (Johannesburg) and **South Africa Central** (Cape Town). Azure provides compliance documentation mapping its controls to the Kenya DPA, and offers data residency options. - **Google Cloud** also operates from South Africa and relies on similar legal mechanisms for compliance with Kenyan regulations. For **CBK-regulated institutions**, the use of these offshore regions is problematic. The CBK's expectation that core financial data remain in-country means that banks cannot simply default to AWS Cape Town or Azure South Africa for core systems. This regulatory friction has created an opening for **local data centre operators** and **local cloud platforms** to serve the financial sector. ![Nairobi skyline lit up at night](/images/nairobi-skyline-night.webp) ## Practical Implications for Data Centre Operators For companies operating or planning to build data centres in Kenya, the data sovereignty and localisation landscape presents both opportunities and obligations: 1. **Compliance as a value proposition**: Marketing a facility as "Kenya DPA-compliant" or "CBK guideline-ready" is a significant differentiator in the market. 2. **Certification matters**: Facilities that can demonstrate alignment with international standards such as **ISO 27001** (information security), **ISO 27701** (privacy information management), and **SOC 2 Type II** will be better positioned to attract regulated clients. 3. **Contractual frameworks**: Data centre operators should develop standard contractual clauses and data processing agreements that address cross-border transfer provisions, giving their clients a compliance-ready foundation. 4. **Engagement with the ODPC**: Proactive engagement with the ODPC (including registration, participation in public consultations, and adherence to guidance) builds regulatory credibility. 5. **Infrastructure investment**: The trend toward stricter localisation in key sectors means that data centre capacity in Kenya is not just a commercial opportunity but a **national infrastructure priority**. ## Conclusion Kenya's data sovereignty framework, anchored in the **Data Protection Act 2019** and reinforced by sector-specific regulators, has created a regulatory environment that increasingly favours in-country data processing and storage. While the DPA itself takes a sophisticated sovereignty-focused approach rather than a blunt localisation mandate, the practical effect (driven by CBK banking guidelines, health sector requirements, and ICT Authority directives) is a growing structural demand for Kenyan data centre capacity. For data centre operators, this represents a compelling market opportunity. For organisations handling Kenyan data, it demands careful compliance planning, robust contractual safeguards, and a clear understanding of when localisation is legally required versus when sovereignty-focused compliance is sufficient. ## Frequently Asked Questions ### Does the Kenya Data Protection Act 2019 require all data to be stored within Kenya? No. The Data Protection Act 2019 does not impose a blanket data localisation requirement. Instead, it regulates cross-border data transfers under Sections 48–50, requiring that transfers only occur to jurisdictions with adequate data protection standards or under specific safeguards such as binding corporate rules, contractual clauses, or the data subject's informed consent. The Act takes a sovereignty-focused approach rather than a strict localisation mandate, though certain sector-specific regulators impose tighter in-country storage rules. ### What is the difference between data sovereignty and data localisation? Data sovereignty refers to the principle that data is subject to the laws and regulations of the country in which it is collected or processed, regardless of where it is physically stored. Data localisation is a stricter, more specific requirement that data must physically reside within a country's borders. Kenya's framework leans toward data sovereignty (ensuring Kenyan law applies to data about Kenyan citizens) while data localisation is imposed only in specific sectors such as banking, health, and government data. ### How does the Central Bank of Kenya regulate data localisation for financial institutions? The Central Bank of Kenya (CBK) has issued guidelines requiring financial institutions, including banks, microfinance institutions, and payment service providers, to ensure that core banking data, customer financial records, and transaction data are hosted on servers located within Kenya. The CBK's Risk Management Guidelines and its Prudential Guidelines on ICT require institutions to notify the CBK before engaging any cross-border cloud or data processing service, and to conduct a risk assessment demonstrating that the offshore arrangement does not compromise data security or the interests of depositors. ### How do multinational cloud providers comply with Kenya's data transfer rules without a local region? Multinational cloud providers such as AWS, Microsoft Azure, and Google Cloud currently do not have dedicated cloud regions in Kenya. AWS operates from its Cape Town region (South Africa), and Azure from its South Africa North and South Africa Central regions. To comply with Kenya's Data Protection Act, these providers rely on legal mechanisms such as Standard Contractual Clauses (SCCs), binding corporate rules, and explicit consent from data controllers. They also offer data residency commitments and compliance documentation. However, the absence of a Kenya region means data physically resides outside Kenyan borders, which creates tension with sector-specific localisation expectations from regulators like the CBK and the Ministry of Health. ### How does Kenya's framework compare to Nigeria's NDPR and South Africa's POPIA on cross-border transfers? All three regimes regulate cross-border transfers but differ in approach. Nigeria's NDPR (2019) requires that data exporters ensure the recipient country provides an adequate level of protection, and the Nigeria Data Protection Regulation mandates registration of data controllers. South Africa's POPIA (2020) is more prescriptive, prohibiting cross-border transfers unless the receiving country has adequate data protection laws, the data subject consents, or contractual safeguards are in place, and it specifically requires the Information Regulator to maintain a list of adequate jurisdictions. Kenya's Data Protection Act 2019 is closest in structure to the EU's GDPR, allowing transfers based on adequacy decisions, appropriate safeguards, or consent. However, Kenya has not yet published a formal adequacy whitelist, leaving compliance more interpretive. ## Kenya Digital Economy and Data Centre Demand Drivers URL: https://data-centers-254.vercel.app/articles/kenya-digital-economy-data-centre-demand Cluster: Kenya Market Published: 2026-08-28 Updated: 2026-08-28 Summary: The macroeconomic forces driving data centre demand in Kenya, from M-Pesa and fintech to 5G, e-commerce, and government digitisation. ## The Macro Picture: Why Kenya's Data Centre Market Is Accelerating Kenya's data centre industry is entering a phase of structural growth driven not by a single sector but by a convergence of macroeconomic forces. From mobile money volumes that rival some European nations to a government digitisation agenda touching every public service, the demand for [colocation and cloud infrastructure](/articles/colocation-data-centre-kenya) in Kenya is compounding year on year. Understanding these demand drivers is essential for investors, operators, and enterprises making infrastructure decisions. ![Wind turbines on a Kenyan hillside](/images/kenya-wind-turbines-3.webp) Kenya's GDP grew at approximately 5.0% in 2024 and is projected to maintain 5-6% growth through 2030, according to the World Bank and IMF. Crucially, the **digital economy's share of GDP** has risen from under 3% a decade ago to an estimated 7.5-8% today. This shift is not abstract, it translates directly into servers, racks, and megawatts of data centre demand. ## Government Digitisation: The Public Sector Engine ### The Kenya Digital Masterplan 2022-2032 The **Kenya Digital Masterplan 2022-2032**, developed by the Ministry of Information, Communications and the Digital Economy, represents the most comprehensive government technology strategy in East Africa. Its ambitions are substantial: connecting all 47 counties with high-speed broadband, digitising 80% of government services, establishing national data centres, and building a workforce of 1.5 million digitally skilled Kenyans. ![Nairobi skyline lit up at night](/images/nairobi-skyline-night.webp) For data centre operators, the masterplan signals sustained public-sector demand. Government systems, including **e-Citizen** (which processed over 20 million service applications in the 2023/2024 financial year), **iTax** (the Kenya Revenue Authority's digital tax platform handling millions of filings), and the **Huduma Namba** digital identity programme, all require local server infrastructure. The [World Bank's Digital Economy Diagnostic for Kenya](https://www.worldbank.org/en/country/kenya/publication/digital-economy-diagnostic-kenya) identified digital government infrastructure as a critical gap that, when filled, will generate significant data centre demand. ### Education Digital Content The Ministry of Education's push for digital learning content, including the Digital Literacy Programme and the Competency-Based Curriculum (CBC) digital resources, requires content delivery networks and local hosting infrastructure. With over 10 million students in primary and secondary schools, the storage and delivery of educational content at scale demands data centre capacity, particularly edge nodes closer to schools in underserved areas. ## Mobile Money: The Backbone of Data Demand ### M-Pesa and Transaction Processing **M-Pesa** remains the single most significant driver of backend infrastructure demand in Kenya. According to the [Communications Authority of Kenya](https://www.ca.go.ke/sector-statistics), mobile money transactions exceeded KES 7.8 trillion (approximately USD 52 billion) in the year ending June 2024. With daily transaction volumes of 15-20 million, the M-Pesa platform requires: - **Real-time transaction processing** with sub-second latency for each payment - **Database infrastructure** capable of handling millions of concurrent read/write operations - **Compliance logging** and audit trail storage that grows terabytes per month - **Disaster recovery** facilities with real-time replication Safaricom operates multiple data centre facilities in Nairobi to support M-Pesa and its broader mobile network. As M-Pesa evolves from a simple person-to-person transfer tool into a full financial services platform, incorporating savings (M-Shwari, KCB M-Pesa), credit (Fuliza), international remittances, and merchant payments, the computational demands intensify. ### Other Mobile Money Platforms While M-Pesa dominates, **Airtel Money** and **Equity Bank's Equitel** also contribute to transaction processing demand. The competitive dynamics among these platforms drive continuous infrastructure investment as each provider seeks performance and reliability advantages. ## Fintech: Beyond Mobile Money ### The Fintech Boom Kenya is widely recognised as Africa's leading fintech hub alongside Nigeria and South Africa. The ecosystem includes: - **Flutterwave**, which established a significant East African presence in Nairobi, processing cross-border payments for businesses across the continent - **Cellulant**, a Kenyan-born payments platform operating in over 35 African countries, requiring robust infrastructure to serve its pan-African network - **The Fintech Consortium of Kenya**, an industry body representing dozens of fintech companies collectively demanding infrastructure, interoperability, and regulatory clarity - **Interswitch**, **PesaPal**, **Jenga API**, and dozens of other payment and lending platforms Each of these companies runs production workloads that require data centre infrastructure. The fintech sector's preference for [local infrastructure](/articles/kenya-east-africa-digital-hub) over distant cloud regions is driven by latency requirements (payment processing demands sub-100ms response times), regulatory compliance, and cost optimisation. ### Lending and Credit Infrastructure Digital lending platforms (including Tala, Branch, and the bank-affiliated mobile lending services) process millions of loan applications using automated credit scoring models. These require real-time API access to credit reference bureaus, mobile money transaction histories, and other data sources. The infrastructure chain from loan application to disbursement involves multiple systems, all hosted in or connected through Kenyan data centres. ![Nairobi Expressway stretching across the city](/images/nairobi-expressway-wide.webp) ## E-Commerce and Digital Retail ### Growing Online Commerce Kenya's e-commerce market has grown steadily, accelerated by the COVID-19 pandemic and the increasing digitisation of consumer behaviour. Key players include: - **Jumia Kenya**, the pan-African e-commerce platform, requiring warehousing management systems, payment processing, and customer-facing web infrastructure - **Copia Global**, which serves peri-urban and rural consumers through a network of agents, relying on logistics optimisation and inventory management systems - **Wasoko** (formerly Sokowatch), the B2B e-commerce platform serving informal retailers across East Africa, requiring real-time inventory, pricing, and delivery management systems - **Jiji**, the classifieds marketplace processing millions of listings and user interactions E-commerce platforms generate demand across the full data centre spectrum: web and application servers, database clusters, analytics and business intelligence systems, and content delivery for product images and media. ## The Startup Ecosystem: Silicon Savannah ### Nairobi as Africa's Tech Capital Nairobi has earned the nickname **"Silicon Savannah"** for good reason. The city hosts: - **Over 200 active tech startups** across fintech, agritech, healthtech, edtech, and logistics - **Venture capital inflows** exceeding USD 400 million annually in recent years, with 2024 seeing renewed investor interest after a global funding slowdown - **Incubators and accelerators** including iHub, Nairobi Garage, Nailab, and Antler - **Global technology company offices** including Microsoft ADC, Google, Amazon Web Services, Oracle, and IBM Each startup that scales beyond its initial cloud credits eventually faces infrastructure decisions. Those handling sensitive data (healthtech, fintech) or requiring low latency gravitate toward Kenyan colocation facilities. The [market outlook to 2030](/articles/kenya-data-centre-market-outlook-2025-2030) suggests this cohort will be a significant demand driver. ## 5G Rollout and Edge Computing Demand ### Safaricom and Airtel 5G Expansion **Safaricom** launched commercial 5G services in October 2023, initially covering parts of Nairobi, Mombasa, and Kisumu. **Airtel Kenya** has deployed its own 5G network. By 2026, 5G coverage is expected to extend to most major Kenyan towns. 5G fundamentally changes the data centre demand landscape in several ways: - **Edge computing nodes** are needed near 5G base stations to process data locally and reduce round-trip latency for applications like autonomous vehicles, industrial IoT, and augmented reality - **Multi-Access Edge Computing (MEC)** requires small-form-factor data centre infrastructure distributed across urban areas - **Network function virtualisation (NFV)** means telco network functions that once ran on proprietary hardware now run on standard servers in data centres This creates a new category of data centre demand, not just large centralised facilities in Nairobi, but distributed edge infrastructure across Kenya's urban geography. ## Internet Penetration and the Connected Population ### The Numbers Kenya's internet ecosystem has reached impressive scale: - **43 million+ internet users** as of 2024, representing approximately 82% of the population - **Over 66 million mobile subscriptions**, reflecting multi-SIM usage and M2M connections - **Average mobile data consumption** exceeding 3 GB per month per user and growing - **Broadband subscriptions** (fixed and wireless) growing at 15-20% annually More users, more data consumption, and more online services all translate into more traffic flowing through Kenyan data centres. Every website hosted locally, every streaming service caching content in Nairobi, and every SaaS application serving Kenyan users benefits from local data centre presence. ## Supply vs Demand: The Growing Gap ### Current Capacity Kenya's total operational data centre capacity stands at approximately **30 MW** spread across facilities in Nairobi (the vast majority) and Mombasa. Key facilities include: - **Africa Data Centres** (formerly CDN / Internet Solutions facilities) in Nairobi - **PAIX** (Pan-African Internet Exchange) data centres - **Safaricom's** owned facilities supporting mobile network and M-Pesa - **Banking sector** facilities operated by KCB, Equity, NCBA, and others - **Government data centres** including the Kenya Information Communications Technology Authority (KICTA) facilities ### Projected Demand Based on the demand drivers outlined above, Kenya's data centre requirements are projected to reach **80-120 MW by 2030**. This implies a need for **50-90 MW of new capacity**, roughly 2-4 times current supply. The gap is particularly acute in: - **Tier III+ colocation** for enterprise customers seeking carrier-neutral facilities - **GPU-ready capacity** for AI and machine learning workloads - **Edge facilities** for 5G MEC and content delivery ![Westlands commercial district in Nairobi](/images/nairobi-westlands.webp) ## Konza Technopolis: The Long-Term Supply Answer The **Konza Technopolis** development, located in Machakos County approximately 60 km from Nairobi's central business district, remains the most ambitious planned addition to Kenya's data centre supply. Conceived under Vision 2030's ICT pillar, Konza is designed as a 5,000-acre technology city with: - **Dedicated power substation** with planned capacity sufficient for data centre operations - **Fibre optic connectivity** via multiple routes connecting to both Nairobi and the Mombasa submarine cable landing stations - **Zoned land allocations** specifically for data centre development, with environmental and planning approvals streamlined - **Planned co-location** with technology companies, universities, and research institutions creating a natural demand ecosystem While Konza's development timeline has extended beyond original projections (the Phase 1 smart city core is now expected to be substantially operational by 2027-2028) the infrastructure investments already made (roads, water, power) reduce the barrier for data centre operators considering greenfield development. ## Connecting the Dots: From Policy to Racks Every demand driver discussed, government digitisation, mobile money growth, fintech expansion, e-commerce, startup scaling, 5G rollout, and rising internet penetration, ultimately converges on a single requirement: **more data centre capacity in Kenya, built to higher standards, and distributed more widely across the country**. The policy environment is increasingly supportive. The **Kenya Data Centre Licensing Framework** under the Communications Authority provides regulatory clarity. The **Kenya Digital Masterplan** explicitly calls for data centre development. The **Data Protection Act 2019** incentivises local processing. And the macroeconomic fundamentals (GDP growth, rising middle class, digital-native demographics) provide the demand foundation. The data centre industry in Kenya is not a speculative bet. It is an infrastructure category whose demand is being written into law, driven by consumer behaviour, and underpinned by the strongest economic growth story in East Africa. ## Frequently Asked Questions ### How much data centre capacity does Kenya currently have? Kenya's total operational data centre capacity is estimated at approximately 30 MW across all facilities in Nairobi and Mombasa. This includes carrier-neutral colocation facilities like Africa Data Centres (ADC), PAIX (part of the Liquid Intelligent Technologies group), and several enterprise-owned facilities operated by banks, telcos, and government agencies. By comparison, South Africa has approximately 150-200 MW of capacity, and Lagos has roughly 40-60 MW. Kenya's capacity is expected to grow to 80-120 MW by 2030 based on current demand trajectories and announced investments. ### What is the Kenya Digital Masterplan 2022-2032 and how does it affect data centres? The Kenya Digital Masterplan 2022-2032 is the government's comprehensive blueprint for transforming Kenya into a digital economy. It encompasses initiatives in digital infrastructure, digital government, digital skills, and digital innovation. For data centres, the masterplan specifically calls for the establishment of government data centres at national and county levels, the development of Konza Technopolis as a technology city with data centre capacity, and the expansion of broadband infrastructure (including 100,000 km of fibre optic network). Each of these initiatives directly or indirectly drives data centre demand, whether through government workloads migrating to local facilities, increased enterprise digital adoption, or the need for edge computing nodes to support the expanded network. ### How does M-Pesa drive data centre demand? M-Pesa processes an average of 15-20 million transactions per day, with peak volumes during salary payment periods and holidays exceeding 30 million daily transactions. Each transaction requires real-time processing, database lookups, logging, and settlement, all running on servers housed in Kenyan data centres. Safaricom operates multiple data centre facilities in Nairobi to support M-Pesa and its broader mobile network. As M-Pesa expands into new services (savings, credit, international remittances, merchant payments), the computational and storage requirements grow correspondingly. The migration of M-Pesa platform components to cloud-native architectures also increases demand for hybrid cloud and colocation infrastructure. ### What role does 5G play in data centre demand? 5G networks require significantly more edge computing infrastructure than 4G. Safaricom launched 5G commercially in 2023, and Airtel Kenya has followed with its own rollout. 5G base stations generate enormous volumes of data that benefit from local processing to reduce latency, this drives demand for edge data centres and micro data centres distributed across urban areas. Multi-Access Edge Computing (MEC) nodes, typically colocated with or near 5G base stations, require their own rack-space, power, and cooling. As 5G coverage expands beyond Nairobi to Mombasa, Kisumu, Nakuru, and other towns, the geographic distribution of data centre demand will widen significantly. ### Is Konza Technopolis actually going to host data centres? Konza Technopolis, located approximately 60 km south of Nairobi in Machakos County, has been planned as Kenya's flagship technology city since its conception under Vision 2030. The masterplan includes designated zones for data centre development, with plans for dedicated high-capacity power feeds and diverse fibre connectivity. While progress has been slower than originally envisioned (a common challenge for megaprojects) key infrastructure components including road access, water supply, and power substation capacity have been completed. The Konza Technopolis Development Authority (KoTDA) has actively marketed the site to data centre operators and technology companies. The strategic logic is sound: Konza offers land at a fraction of Nairobi industrial zone costs, planned power capacity, and proximity to the fibre corridors running between Nairobi and Mombasa. ## Kenya Digital Infrastructure: What Is Actually Live? URL: https://data-centers-254.vercel.app/articles/kenya-digital-infrastructure-what-is-live Cluster: Data Centres Published: 2026-10-02 Updated: 2026-10-02 Summary: DC254 tracks 27 facilities and 10 cable systems in Kenya. 20 are operational, 10.5 MW is verified live, and 230 MW is announced. Here is the real picture. Kenya's digital infrastructure is bigger than the headlines suggest, but smaller than the announcements. The DC254 Q3 2026 snapshot tracks 27 facilities and projects in Kenya, and not all of them are real today. 20 are operational. The other 7 are under construction, committed or early-stage. On capacity, 10.5 MW of IT load is verified as live, against a 230 MW announced pipeline. On connectivity, 7 submarine cable systems are in service at Mombasa. That gap between what exists and what is announced is not a contradiction. It is the single most useful thing to understand about Kenya's digital economy right now. ![DC254 Intelligence dashboard for Kenya, Q3 2026](/images/kenya-infrastructure-dashboard.png) ## What do the 27 tracked records actually contain? Every DC254 directory record carries a status, and the four statuses mean very different things. | Stage | Kenya records | Announced MW | What it means | |---|---|---|---| | Operational | 20 | 42.9 designed | Built and running; 10.5 MW published as live IT load | | Under construction | 3 | 77 | Capital committed, structures going up | | Committed | 1 | 53 | Developer has committed to build, site prep or design underway | | Early stage | 3 | 100 stated, 2 unquantified | Announced intent only; nothing is guaranteed | Under construction covers iXAfrica NBOX1.2 (18 MW), Africa Data Centres Nairobi 2 (15 MW) and Nxtra by Airtel at Tatu City (44 MW). The one committed record is iXAfrica NBOX2 at Tilisi (53 MW). In early stage, the Microsoft and G42 AI data centre carries a stated 100 MW, while Raxio Nairobi and the Kenya Data Centres SME Facility have no verified MW figure yet. Read the ladder from bottom to top and the pattern is clear: the further a record sits from operational, the less certain its numbers become. ## Where does Kenya's digital infrastructure actually sit? Nairobi holds 19 of the 27 tracked records, about 70 percent. Mombasa holds 4, and all four are operational. Konza, Limuru and Thika each have one operational facility, and Ruiru has the Nxtra project at Tatu City. Two facts follow from this split. First, the live base is more distributed than people assume: operators run facilities along the coast, in the capital and in satellite towns. Second, the entire announced pipeline is a Nairobi-metro story. Every one of the 7 non-operational records sits in Nairobi or its metropolitan fringe, which means the next 230 MW, if it gets built, lands where the demand, fibre and staff already are. ![Map of East Africa data centre locations](/images/africa-dc-map.webp) ## How big is the gap between 10.5 MW live and 230 MW announced? The announced pipeline is roughly 22 times the verified live base. That ratio is striking, but the honest reading is more careful. The 10.5 MW is a floor, not a ceiling: only 8 of the 20 operational facilities publish a verified live IT load figure, so Kenya's real running load is certainly higher and unmeasured in public data. The 230 MW, meanwhile, is not a forecast. It spans very different confidence levels: 77 MW is physically under construction, 53 MW is committed, and 100 MW rests on a single early-stage announcement. If the Microsoft and G42 project alone were delivered, it would multiply Kenya's live capacity nearly tenfold. If it slipped, the pipeline would shrink by 43 percent overnight. That is why DC254 reports the two numbers side by side and never adds them into one figure. ## Why do cables in Mombasa not equal compute in Nairobi? Connectivity is Kenya's most mature layer. TEAMS and SEACOM arrived in 2009, EASSy in 2010, LION2 in 2012, DARE1 in 2021, PEACE in 2022 and 2Africa in 2024. Africa-1 has already come ashore but has not announced readiness for service, so DC254 does not count it as live. Daraja is announced and LuLu is planned. Seven systems in service over fifteen years is genuine maturity. But a cable is not a data centre. International capacity lands at Mombasa, then travels over terrestrial fibre to internet exchange points in Nairobi, and only then reaches the facilities where compute actually runs. Operators cite Kenya's grid, more than 75 percent renewable, as the power draw. Each link in that chain, cable, landing station, fibre, IXP, data centre, power, has its own live-versus-announced split. Connectivity into the country is not the same thing as capacity inside it. ![Submarine cable coming ashore in Kenya](/images/mombasa-cable-landing.webp) ## How should you read infrastructure announcements? The discipline DC254 applies is simple. A facility can be operational, under construction, committed or early stage, and each label is a different kind of fact. A cable can be in service, landed with readiness pending, or still a drawing. Capacity can be verified live, built and designed, or announced pipeline. Wednesday's Inside the Map piece showed how we decide whether a cable is truly live. The same method separates Kenya's 20 live facilities from its 230 MW of announcements. When a headline says Kenya is building a hyperscaler market, check which rung of the ladder the project actually sits on. What exists is real infrastructure. What is announced is intention. DC254 tracks both, and never lets them blur. ![Server hall inside a Nairobi data centre](/images/atlancis-nairobi-datacentre-hall.webp) ## How Data Centres Make Kenya East Africa's Digital Hub URL: https://data-centers-254.vercel.app/articles/kenya-east-africa-digital-hub Cluster: Market Analysis Published: 2026-08-28 Updated: 2026-09-20 Summary: Submarine cables, renewable energy, skilled workforce, and growing capacity make Kenya East Africa's natural digital hub for 300 million people. The phrase "Silicon Savannah" has been applied to Kenya for over a decade, and while it sometimes feels more aspirational than descriptive, the underlying reality is that Kenya has built a genuine and growing digital infrastructure ecosystem that serves not just its own 55 million people but the broader East African region. Data centres are the physical foundation of this ecosystem, the buildings full of servers that make every digital service possible. Understanding Kenya's position as East Africa's digital hub requires understanding the infrastructure that makes it possible. ![Westlands commercial district in Nairobi](/images/nairobi-westlands.webp) Kenya's emergence as a regional digital hub is not the result of a single policy decision or a single infrastructure project. It is the cumulative effect of decades of investment in connectivity, energy, human capital, and regulatory reform, much of it unplanned and driven by private sector initiative rather than government direction. The result is an ecosystem that, while imperfect, offers the most comprehensive digital infrastructure in East Africa. ## The Connectivity Foundation Kenya's most fundamental advantage is connectivity. The country's geographic position (with a long Indian Ocean coastline south of the Horn of Africa) made it the natural landing point for submarine cables serving the East African coast. Four submarine cable systems land in Mombasa: SEACOM (2009), TEAMS (2009), EASSy (2010), and DARE1 (2022). Together, these cables provide multiple terabits per second of international bandwidth, connecting Kenya to Europe, Asia, the Middle East, Southern Africa, and the rest of the world. No other East African country has four submarine cable landing stations. Tanzania has two (in Dar es Salaam), and landlocked countries like Uganda, Rwanda, Burundi, and South Sudan depend entirely on cross-border fibre connections to reach submarine cables. Kenya's direct cable access means lower latency to international destinations, more routing options, and competitive bandwidth pricing. ![Nairobi Expressway stretching across the city](/images/nairobi-expressway-wide.webp) The terrestrial fibre network that distributes this connectivity inland is equally important. Kenya has an estimated 15,000–20,000 kilometres of fibre optic cable, connecting all 47 county headquarters through the National Optic Fibre Backbone Infrastructure (NOFBI) and through operator-owned networks from Liquid Intelligent Technologies, Safaricom, Telkom Kenya, and others. Cross-border fibre links connect Kenya to Uganda, Tanzania, Ethiopia, and South Sudan, making Kenya a transit hub for regional internet traffic. The [Kenya Internet Exchange Point (KIXP)](/articles/kixp-internet-exchange-point-kenya) reinforces this connectivity advantage. KIXP allows networks within Kenya to exchange traffic locally rather than routing through South Africa or Europe, reducing latency and cost for domestic traffic. KIXP is one of the largest and most active internet exchange points in sub-Saharan Africa, and its presence in Nairobi is both a cause and a consequence of the city's position as a regional hub. ## The Energy Advantage Kenya's energy mix is a structural advantage that most discussions of digital infrastructure understate. Over 90% of the country's electricity is generated from renewable sources: geothermal energy from the Rift Valley (the largest source), hydroelectric power from dams, wind energy from the Lake Turkana wind farm, and a growing contribution from solar PV. This renewable mix gives Kenya some of the lowest carbon-intensity electricity in the world and competitive industrial power costs. For data centres, this matters in two ways. First, cost: at KES 12–18 per kilowatt-hour ($0.08–0.12), Kenya's industrial electricity is cheaper than Nigeria ($0.20–0.35 including generator costs), competitive with South Africa ($0.10–0.15 plus load shedding costs), and only slightly more expensive than Egypt ($0.05–0.07, subsidised). Second, sustainability: as global companies face increasing pressure to reduce the carbon footprint of their digital infrastructure, Kenya's renewable grid offers a genuine green advantage that no amount of on-site solar can fully replicate in a country dependent on fossil fuel generation. The energy advantage also now has a cross-border dimension. Kenya imports up to 200 MW of Ethiopian hydropower over the 1,069 km, 500 kV HVDC interconnector that was energised commercially in 2022, with the volume scheduled to double to 400 MW from December 2026. The imports add system headroom during dry hydrology years and peak hours, and they reinforce the regional integration that makes Kenya's hub position durable rather than purely domestic. ![Aerial view of the Grand Ethiopian Renaissance Dam with its spillway flowing into the Blue Nile](/images/gerd-dam-aerial.webp) ## The Talent Pool Kenya's human capital is a critical component of its digital hub positioning. The country produces approximately 50,000 university graduates annually, with significant numbers in engineering, computer science, mathematics, and related fields. The University of Nairobi, Strathmore University, Jomo Kenyatta University of Agriculture and Technology (JKUAT), and Kenyatta University all have established computer science and engineering programmes. Beyond formal education, Kenya has a vibrant self-taught and community-trained technology culture. Developer communities, coding bootcamps, and technology hubs (like iHub, Nairobi Garage, and Nailab) have produced thousands of software developers, designers, and product managers who work for local companies, international technology firms, and remote-first companies based in the US and Europe. ![Engineers and officials on a technical visit inside a facility power hall](/images/site-visit-power-hall-kenya.webp) Microsoft's Africa Development Centre (ADC) in Nairobi, employing hundreds of engineers working on global products, is both a symbol and a driver of Kenya's talent position. Google, IBM, Intel, and other technology companies have research or development presences in Nairobi, attracted by the talent pool and the opportunity to build products for African markets. The presence of these companies, in turn, attracts more talent and creates a virtuous cycle. ## The Financial Services Engine Kenya's financial services sector is the engine that drives much of the demand for digital infrastructure. M-Pesa, the world's most successful mobile money platform, processes transactions equivalent to over 50% of Kenya's GDP annually. Every M-Pesa transaction (every payment, every transfer, every withdrawal) requires data processing in data centres. Safaricom's investment in data centre capacity has been driven primarily by the needs of M-Pesa and its broader mobile network, and the colocation market has grown to serve the banks, insurers, and fintech companies that form the financial ecosystem around M-Pesa. The banking sector is equally significant. Kenya's banks have been among the most aggressive in Africa in adopting digital channels. Mobile banking, USSD banking, agent banking, and internet banking have become the primary customer interaction channels for most banks, shifting transaction processing from bank branches to data centres. The Central Bank of Kenya's promotion of digital financial services and its regulatory framework for fintech have further accelerated this shift. ## Serving the Region Kenya's digital infrastructure does not serve Kenya alone. Organisations across East Africa (in Uganda, Tanzania, Rwanda, Burundi, South Sudan, and the eastern Democratic Republic of Congo) connect to services hosted in Kenyan data centres. The reasons are practical: Kenya has the most data centre capacity, the best connectivity, and the most service providers in the region. A bank in Kampala may choose to host its disaster recovery systems in a Nairobi data centre because the connectivity, expertise, and competitive pricing available in Kenya are superior to what is available domestically. This regional role is self-reinforcing. As more regional organisations use Kenyan data centres, the data centre operators invest in more capacity, more connectivity, and more services, making Kenya even more attractive. The [data centre market outlook](/articles/kenya-data-centre-market-outlook-2025-2030) projects continued growth driven in part by this regional demand. ![Turbine shaft assembly inside a hydropower powerhouse](/images/gerd-turbine-pit.webp) ## The Challenges ![Nairobi skyline at sunset](/images/nairobi-sunset-wide.webp) Kenya's digital hub position is not unassailable. Rwanda's government has been more aggressive and intentional in its digital transformation policies, offering aggressive tax incentives, fast-track business registration, and regulatory simplicity. Tanzania has been investing in connectivity and has a larger population than Kenya (65 million). Ethiopia, with 120 million people, is opening its telecommunications market and could become a significant digital economy if it manages the transition effectively. Domestically, the challenges include power reliability in some areas, a skills gap in specialised data centre roles, currency volatility that complicates investment planning, and the fact that much of Kenya's digital infrastructure advantage was built by private sector initiative rather than coordinated government strategy. The Konza Technopolis project, intended to be the physical embodiment of Silicon Savannah, has been delayed repeatedly, illustrating the gap between aspiration and execution. ## The Path Forward Kenya's position as East Africa's digital hub is real but not guaranteed. Maintaining and strengthening it requires continued investment in data centre capacity (as the [licensing framework](/articles/kenya-data-centre-licensing-framework) provides regulatory clarity), expansion of connectivity to underserved areas, development of specialised skills, and policy coherence that supports rather than impedes digital infrastructure investment. The data centres being built today (iXAfrica's expansion, Africa Data Centres' growth, and the next wave of facilities currently in planning) are the physical manifestation of Kenya's digital hub positioning. They are the buildings that make the digital economy possible. And they are the reason that, when a company anywhere in East Africa needs to decide where to host its servers, Nairobi remains the first and most logical choice. The clearest statement of that ambition now sits at the top of government. In a September 2026 essay, the Principal Secretary for ICT and the Digital Economy, Eng. John Kipchumba Tanui, argued that Kenya has already built the cables and that the next national assignment is compute: cloud regions, AI-ready capacity and the services built on top of them. We unpacked the full plan and fact-checked its claims against our directory in [Kenya's data centre and compute hub plan, explained](/articles/kenya-data-centre-compute-hub-plan-explained). If the state executes even part of it, the hub positioning this guide describes stops being a description and becomes a strategy. ## Kenya ICT Policy and Data Centre Industry Impact URL: https://data-centers-254.vercel.app/articles/kenya-ict-policy-framework-data-centres Cluster: Policy & Regulation Published: 2026-08-28 Updated: 2026-08-28 Summary: Kenya's ICT policy environment and its impact on data centres: the Digital Masterplan, licensing, cybersecurity strategy, and African comparisons. ![A hand holds up the pocket edition of the Constitution of Kenya](/images/constitution-kenya-pocket-hand.webp) Kenya's data centre industry does not operate in a vacuum. Every rack installed, every fibre optic cable landed, and every megawatt of power allocated is shaped by a complex web of policies, statutes, and regulatory directives. For investors, operators, and technology companies evaluating Kenya as a data centre destination, understanding the **ICT policy framework** is not optional, it is the foundation upon which every business case is built. This article provides an in-depth analysis of the policies that define Kenya's data centre landscape, examining how they create both opportunities and constraints for the industry. ## The Kenya Digital Masterplan 2022-2032: A Decade of Digital Infrastructure The **Kenya Digital Masterplan 2022-2032**, developed by the ICT Authority and the Ministry of Information, Communications and the Digital Economy, is the most consequential policy document for the data centre sector. It sets out an ambitious roadmap covering five key pillars: digital infrastructure, digital government, digital skills, digital business, and innovation. For the data centre industry specifically, the Masterplan contains several critical provisions. It envisions the establishment of **national data centres** to host government systems and promote data localisation. The plan allocates significant budgetary resources toward building tier-certified government data facilities, with the ICT Authority mandated to oversee their development and operation. This directly stimulates demand for colocation services from private sector operators who may partner with or supplement government infrastructure. The Masterplan also calls for the expansion of broadband connectivity to achieve over 90% population coverage, which is fundamentally linked to data centre demand. More connected users mean more data generation, more cloud adoption, and greater need for local processing and storage capacity. According to the [Kenya Digital Masterplan 2022-2032](https://www.icta.go.ke/kenya-digital-masterplan/), the government targets the establishment of at least 25,000 kilometres of additional fibre optic network, creating the backbone connectivity that makes edge and core data centres viable across the country. ## Vision 2030 and the ICT Pillar Kenya's **Vision 2030** development blueprint identified ICT as a key enabler of national transformation. Under the **Macro-Pillar on Infrastructure**, the government committed to developing a robust ICT infrastructure including data centres, fibre networks, and digital service delivery platforms. While Vision 2030 predates the current data centre boom, it established the policy foundation that subsequent strategies have built upon. The Vision 2030 ICT pillar specifically targeted Kenya becoming a **regional ICT hub**, a goal that has directly benefited data centre operators. The positioning of Nairobi as East Africa's technology capital has attracted cloud service providers, content delivery networks, and hyperscale operators, all of whom require local data centre capacity. Companies like **Africa Data Centres** have cited Kenya's strategic positioning under Vision 2030 as a factor in their expansion decisions. ## The Kenya Information and Communications (Amendment) Act 2013 The **Kenya Information and Communications Act (KICA)**, as amended in 2013, is the primary legislation governing the ICT sector. It establishes the [Communications Authority of Kenya](https://www.ca.go.ke/licensing/) (CA) as the sector regulator with broad powers over licensing, spectrum management, and infrastructure standards. For data centre operators, KICA is significant in several ways. First, it defines the regulatory perimeter within which data centres operate. While the Act does not create a standalone "data centre licence," it classifies data centres as **ICT infrastructure** and subjects them to regulatory oversight. Operators who provide connectivity services from their facilities typically require a **telecommunications licence** or a **broadcasting licence**, while those offering purely colocation services may operate under an application service provider (ASP) licence category. The Act also empowers the CA to set **quality of service standards** and to regulate the sharing of ICT infrastructure. This has practical implications for data centre operators who may wish to share tower infrastructure, duct space, or fibre routes with telecommunications providers. ## Communications Authority of Kenya: The Licensing Regulator The **Communications Authority of Kenya (CA)** is the primary regulatory body that data centre operators must engage with. The CA's mandate extends to licensing all communications services and facilities, including data centres that provide connectivity or host telecommunications equipment. Under the [Kenya data centre licensing framework](/articles/kenya-data-centre-licensing-framework), operators must navigate several licence categories depending on their service model. A pure-play colocation provider may require a Network Facilities Provider (NFP) licence if they own and operate transmission infrastructure, while a full-service data centre offering cloud, connectivity, and managed services may need multiple licence categories. The CA has progressively refined its approach to data centre regulation. In recent consultations, the authority has signalled its intention to develop **more specific guidelines** for data centre licensing, recognising that the current framework (designed primarily for telecommunications operators) does not perfectly fit the data centre business model. This regulatory evolution is critical for the industry, and the [Kenya data centre licensing framework](/articles/kenya-data-centre-licensing-framework) article covers the current requirements in detail. ## ICT Authority: Government Data Centres and e-Government Strategy While the CA regulates the broader ICT sector, the **ICT Authority** has a specific mandate over government digital infrastructure. Established under the ICT Act 2013, the ICT Authority is responsible for the **e-Government strategy** and oversees all government data centres. The ICT Authority manages the **Government Common Core**, a shared services platform that includes government data centres, cloud services, and digital payment infrastructure. This mandate means that the Authority is both a **regulator and a customer** of the data centre industry. Government agencies are required to host their systems on approved infrastructure, and the ICT Authority determines whether that infrastructure is government-owned, outsourced to private operators, or a hybrid of both. For private data centre operators, the ICT Authority's role creates significant demand. As the government digitises services, from tax collection (iTax) to health records (Afya Ke) to land management (ARDHI), the volume of government data requiring secure, compliant hosting grows substantially. Operators who meet the ICT Authority's standards and obtain the necessary approvals can access a substantial and growing revenue stream. ## National Cybersecurity Strategy and Data Centre Security Requirements Kenya's **National Cybersecurity Strategy**, developed under the coordination of the National Cybersecurity Steering Committee, has direct implications for data centre design and operations. The strategy, aligned with the **Computer Misuse and Cybercrimes Act 2018**, establishes a framework for protecting the country's critical information infrastructure. Data centres designated as **Critical Information Infrastructure (CII)** are subject to enhanced security requirements. These include mandatory incident reporting to the **KE-CIRT/CC** (Kenya Computer Incident Response Team), compliance with national security standards, and regular security audits. The strategy also promotes the adoption of international standards such as ISO/IEC 27001 for information security management. For operators, this means that **security compliance is not merely a commercial differentiator** but a regulatory requirement. Data centre security investments, including physical security systems, access controls, surveillance, fire suppression, and cybersecurity operations centres, must align with both the National Cybersecurity Strategy and the [Kenya data sovereignty and localisation requirements](/articles/kenya-data-sovereignty-localisation). The cost implications are significant but also create barriers to entry that benefit established, compliant operators. ![KICC tower and the Nairobi skyline at night](/images/nairobi-skyline-night-kicc.webp) ## Konza Technopolis Development Authority and Its Policy Framework **Konza Technopolis**, also known as "Silicon Savannah," is Kenya's flagship technology city project located approximately 60 kilometres south of Nairobi in Machakos County. Governed by the **Konza Technopolis Development Authority (KoTDA)** under the Konza Technopolis Development Authority Order 2012, the project has a dedicated policy framework designed to attract technology investments, including data centres. KoTDA operates under a **one-stop-shop model** that streamlines regulatory approvals for technology businesses. For data centre developers, this means reduced bureaucratic friction in obtaining construction permits, environmental clearances, and utility connections. The Authority has developed **building codes and standards** specific to technology facilities, which are discussed in our article on [building codes for data centres in Kenya](/articles/building-codes-data-centres-kenya). Konza's policy framework includes provisions for **special economic zone benefits**, including tax incentives, streamlined immigration processes for technology workers, and dedicated infrastructure. The Technopolis has allocated specific zones for data centre development, with pre-built power and cooling infrastructure designed to meet tier-rated facility requirements. While Konza's development has faced delays, the policy framework it establishes represents an important precedent for how Kenya can create **policy-enabled investment zones** for the data centre industry. As construction progresses, Konza is expected to host both government and private data centre facilities. ## The Proposed Data Protection (Amendment) Bill Kenya enacted the **Data Protection Act 2019**, establishing the Office of the Data Protection Commissioner (now the Data Protection Office under the Office of the Attorney General). The Act created obligations around data processing, consent, and cross-border data transfers that directly affect data centre operators. The **proposed Data Protection (Amendment) Bill** is expected to introduce several changes relevant to the data centre industry. Key provisions under discussion include stricter **data localisation requirements** for certain categories of personal data, enhanced obligations for data processors (including data centre operators), and increased penalties for non-compliance. The bill may also introduce sector-specific data protection regulations that would affect data centres hosting financial, health, or government data. For investors, the [Kenya data sovereignty and localisation requirements](/articles/kenya-data-sovereignty-localisation) are a critical policy consideration. Stricter localisation rules would drive additional demand for local data centre capacity but could also increase compliance costs for operators who must demonstrate that specific data categories remain within Kenyan borders. ## Policy Certainty and Investment Decisions **Regulatory clarity is consistently cited by data centre investors as a primary factor** in investment decisions across Africa. In Kenya, the policy environment presents a mixed picture, strong institutional frameworks exist, but uncertainties in specific areas can delay or deter investment. **Africa Data Centres**, a pan-African data centre operator and a subsidiary of the REMRO Group, has expanded its Nairobi footprint significantly. The company has invested in multiple facility expansions in Nairobi, citing Kenya's strategic position, growing digital economy, and improving regulatory environment. However, industry sources indicate that **licensing complexity** and overlapping regulatory mandates between the CA, the ICT Authority, and county governments remain a concern. **PAIX Nairobi** (Pan-African Internet Exchange), operated by the Teraco/DP World group, has also invested in Kenyan infrastructure. PAIX's presence in Nairobi reflects confidence in the market but also highlights the importance of **predictable, transparent regulation**, the company operates across multiple African markets and actively compares regulatory environments when making investment decisions. Investors consistently emphasise that **policy uncertainty**, whether from pending legislation, shifting regulatory interpretations, or inconsistent enforcement across national and county government levels, increases the perceived risk of Kenyan data centre investments. Addressing these uncertainties through clearer, more coordinated policy frameworks would significantly enhance Kenya's attractiveness as a regional data centre hub. ### The Role of Industry Associations in Shaping Policy Kenya benefits from active industry associations that play a crucial role in bridging the gap between the public and private sectors on ICT policy matters. **TESPOK** (Telecommunications Service Providers Association of Kenya) represents the interests of telecommunications and infrastructure operators, including data centre companies. TESPOK engages in policy advocacy, provides industry input on regulatory consultations, and has been instrumental in pushing for more data-centre-specific regulatory frameworks. The association also manages the **Kenya Internet Exchange Point (KIXP)**, which is directly relevant to data centre interconnection and peering economics. **KICTANet** (Kenya ICT Action Network) is a multi-stakeholder platform that brings together government, private sector, civil society, and academia to discuss ICT policy issues. KICTANet has been particularly active on data protection, cybersecurity, and digital rights, all of which affect the data centre operating environment. The network's policy briefs and public consultations have influenced the direction of data protection regulation and cybersecurity strategy. Both organisations demonstrate that **industry-led policy engagement** is essential for creating a regulatory environment that supports data centre growth while protecting public interest. ## Comparative Analysis: Kenya, Rwanda, Nigeria, and South Africa ![Transmission tower beside a commercial high-rise](/images/dc-challenges.webp) Understanding Kenya's policy position requires examining it alongside comparable African markets. Each country offers distinct advantages and challenges for data centre investors. ### Rwanda: Highly Proactive and Streamlined Rwanda has adopted a deliberately **proactive and simplified** approach to ICT regulation. The Rwanda ICT Chamber works closely with the Rwanda Development Board to create a fast-tracked regulatory environment. The country offers a **corporate tax holiday of up to 8 years** for ICT companies, VAT exemptions on imported equipment, and a single regulatory clearance process. Rwanda's **National Data Centre** was developed as a government priority with strong policy backing. The regulatory environment is streamlined, predictable, and designed specifically to attract technology investment, though the smaller market size limits the scale of data centre demand. ### Nigeria: Growing Framework with Scale Nigeria's data centre policy environment is **growing but complex**. The Nigerian Data Protection Act 2023 introduced data protection requirements similar to Kenya's, while the Nigerian Communications Commission (NCC) regulates data centres as part of ICT infrastructure. Nigeria's **special economic zones** offer tax holidays and incentives, and the country's massive population creates enormous data centre demand. However, regulatory complexity across federal and state levels, inconsistent power supply, and foreign exchange challenges create a more difficult operating environment. The [Africa data centre regulation compared](/articles/africa-data-centre-regulation-compared) analysis covers these dynamics in detail. ### South Africa: Mature but Complex South Africa has the **most mature data centre market** in Africa, with the most developed regulatory framework. The **Electronic Communications Act** and the **Protection of Personal Information Act (POPIA)** create a comprehensive compliance environment. However, South Africa's regulatory framework is also the most **complex and compliance-intensive**, with multiple regulators (ICASA, CIPC, Information Regulator) and stringent requirements across licensing, data protection, and labour law. While this complexity provides certainty, it also increases the cost and time required to establish and operate data centres. ### Kenya's Position Kenya occupies a **middle ground**, more mature and institutionally stronger than Rwanda and Nigeria, but less complex than South Africa. Its policy environment is characterised by strong institutions (CA, ICT Authority) with established track records, a growing but sometimes fragmented regulatory framework, and active industry associations that facilitate policy dialogue. The **key opportunity** for Kenya lies in reducing regulatory fragmentation and creating clearer, more coordinated policies that specifically address the data centre sector's unique needs. ## Conclusion: The Policy Path Forward for Kenya's Data Centre Industry Kenya's ICT policy framework has been instrumental in establishing the country as East Africa's leading data centre market. The Digital Masterplan 2022-2032, Vision 2030, and supporting legislation provide a strong strategic foundation. However, the industry faces challenges from **regulatory overlap**, **pending legislative changes** (particularly the Data Protection Amendment Bill), and the need for **data-centre-specific regulations** that better reflect the unique characteristics of the business. The most successful policy environments for data centre investment (as demonstrated by Rwanda's streamlined approach) combine **regulatory clarity, targeted incentives, and efficient approval processes**. Kenya has the institutional capacity to achieve this, and with continued engagement between government regulators, industry associations like TESPOK and KICTANet, and the private sector, the policy framework can evolve to support the next wave of data centre investment that the country's digital economy demands. ## Frequently Asked Questions **What is the Kenya Digital Masterplan 2022-2032 and how does it affect data centres?** The Kenya Digital Masterplan is the government's blueprint for digital transformation over a decade. It includes specific provisions for building government data centres, expanding broadband infrastructure, and establishing a national data centre ecosystem. It directly drives demand for colocation and hyperscale facilities by mandating digital government services and data localisation. **Does the Communications Authority of Kenya license data centres?** Yes. The CA classifies data centres as part of ICT infrastructure under the Kenya Information and Communications (Amendment) Act 2013. Operators typically require a telecommunications licence or an application service provider licence, depending on the services offered. The [Kenya data centre licensing framework](/articles/kenya-data-centre-licensing-framework) provides detailed guidance on compliance. **How does the National Cybersecurity Strategy affect data centre operators in Kenya?** The National Cybersecurity Strategy mandates that critical information infrastructure, including major data centres, implement specific security controls. Operators must comply with the Computer Misuse and Cybercrimes Act 2018, report incidents to the National Computer Incident Response Team Coordination Centre (KE-CIRT/CC), and adhere to minimum security standards that affect facility design, access controls, and data handling practices. **What role does Konza Technopolis play in Kenya's data centre policy?** Konza Technopolis, governed by the Konza Technopolis Development Authority (KoTDA), is a flagship Vision 2030 project designed as a technology city with dedicated data centre zones. It offers a streamlined regulatory environment, pre-approved building standards for technology facilities, and integrated infrastructure including dedicated power and fibre, making it a policy-enabled investment zone for data centre developers. **How does Kenya's ICT policy compare with Rwanda and Nigeria for data centre investment?** Kenya offers a more mature but complex policy environment with established institutions like CA and the ICT Authority. Rwanda is highly proactive with simplified, fast-tracked regulations and aggressive incentives. Nigeria has a growing framework with strong market size but regulatory complexity across federal and state levels. South Africa has the most mature framework but with significant compliance burden. Each presents distinct trade-offs between regulatory certainty and ease of market entry. ## Kenya Internet Speeds: What They Mean for Data Centres URL: https://data-centers-254.vercel.app/articles/kenya-internet-speeds-data-centres Cluster: Connectivity Published: 2026-08-28 Updated: 2026-08-28 Summary: Kenya's average mobile internet speed tops 25 Mbps. How speeds shape data centre demand, user experience, and digital competitiveness. When you open a banking app on your phone and your account balance loads instantly, when a YouTube video starts playing within seconds, or when a video call connects in high definition, you are experiencing the result of internet speed. The speed at which data travels between your device, the network, and the data centre where the service is hosted determines the quality of your digital experience. For Kenya's data centre industry, internet speeds are both an enabler and a measure of the country's digital maturity. ![Diagram of subsea cable architecture from landing station to POP](/images/submarine-cable-diagram.webp) Internet speed in Kenya has improved dramatically over the past decade, driven by submarine cable landings, fibre network expansion, mobile network upgrades, and the growth of local data centre capacity that keeps content and services close to users. But speed is not uniform, it varies by location, by technology, and by time of day. Understanding these variations, and what they mean for data centre demand and strategy, is important for anyone involved in Kenya's digital infrastructure. ## Kenya's Internet Speed in Context ### The Numbers Kenya's internet speeds, as measured by platforms like Ookla's Speedtest Global Index, place the country among the better-connected in Africa while still well below global averages. As of 2025, typical performance includes: **Mobile internet** (4G/LTE, the dominant mobile technology): average download speeds of 25-30 Mbps, with peak speeds of 50-100 Mbps on good connections in well-served areas. Safaricom's 4G network covers most urban and peri-urban areas and delivers the fastest mobile speeds, with Airtel also providing competitive 4G coverage. **Fixed broadband** (fibre connections to homes and businesses): average download speeds of 40-60 Mbps in Nairobi, with premium plans from providers like Safaricom Home Fibre, Zuku, and Faiba offering 100-500 Mbps. These speeds are available primarily in Nairobi, Mombasa, and other major towns where fibre-to-the-home (FTTH) has been deployed. **5G**: Safaricom has launched 5G in select areas of Nairobi, Mombasa, Kisumu, and other towns. 5G speeds can reach 200-1000 Mbps in ideal conditions, though coverage is still limited and device adoption is in early stages. 5G's impact on data centre demand will grow as coverage expands. ![Map of submarine cables landing on the East African coast](/images/submarine-cables-map.webp) ### Global and Regional Comparison Globally, the fastest mobile internet speeds are found in the UAE (300+ Mbps), South Korea (200+ Mbps), and Norway (180+ Mbps). The global average is approximately 50-60 Mbps for mobile and 100+ Mbps for fixed broadband. Kenya's speeds, while above the African average, are roughly half the global average. Within Africa, Kenya typically ranks 3rd-5th for both mobile and fixed speeds. Egypt often leads due to its investments in mobile infrastructure and fibre. South Africa's speeds are comparable to Kenya's but vary more widely due to the geographic spread of its population. Morocco and Tunisia also rank highly, particularly for fixed broadband. ## The Speed Chain: Why Every Link Matters Internet speed experienced by the end user is determined by the slowest link in a chain that connects the user's device to the server in a data centre. Understanding this chain is essential for understanding where speed improvements can be made. ![Chart of satellite and RF connector types](/images/diagram-satellite-connector-types.webp) ### The Last Mile (User to Network) ![Workers installing fibre ducting in a Nairobi street trench](/images/nairobi-street-fibre-trench.webp) For mobile users, this is the radio connection between the phone and the cell tower. For fixed broadband users, this is the fibre or copper connection between the home/business and the ISP's network. The last mile is often the bottleneck because it is the most expensive part of the network to upgrade and the part most affected by physical constraints (distance from the cell tower, quality of the in-building wiring). ### The Backhaul (Network to Data Centre) Once data reaches the ISP's network, it travels over backhaul links (fibre or microwave) to the ISP's core network, and from there to the data centre where the service is hosted. If the service is hosted in a Kenyan data centre, this journey is entirely within Kenya and typically adds 5-20ms of latency. If the service is hosted overseas (in South Africa, Europe, or the US), the journey includes submarine cable segments that add 50-200ms. This is where data centres make a direct impact on perceived internet speed. Services hosted in Kenyan data centres are accessed with 5-20ms latency, while the same services hosted overseas are accessed with 50-200ms latency. For most web browsing and email, this difference is imperceptible. For real-time applications (video calls, online gaming, financial trading), the difference matters. ### The Server (Data Centre Processing) ![Fibre optic termination equipment in a data centre](/images/dc-fibre-optics.webp) The server itself contributes to the speed equation. A slow or overloaded server takes longer to respond to requests, regardless of how fast the network is. This is why data centre operators invest in high-performance servers, fast storage (SSDs), and efficient software. A well-provisioned data centre with modern servers can serve responses in milliseconds; an under-provisioned or poorly managed one can add hundreds of milliseconds. ## How Internet Speeds Drive Data Centre Demand ### The Bandwidth Demand Cycle Faster internet speeds create a virtuous cycle for data centre demand. When users get faster connections, they consume more data, they stream higher-quality video, use more cloud applications, upload and download larger files, and spend more time online. This increased consumption requires more server capacity in data centres to handle the traffic. The evidence is clear globally: as internet speeds increase, data traffic grows exponentially. In Kenya, mobile data consumption has grown from approximately 1 exabyte (1 billion gigabytes) per year in 2018 to over 10 exabytes in 2025. This growth is driven by both the increasing number of internet users (from 40 million to 55 million) and the increasing speed of connections (which enables more data-intensive usage). ### Content Localisation When internet speeds are high enough to support data-intensive applications like video streaming, the location of the server matters more, not less. A user streaming a 4K video on a fast fibre connection consumes 25-50 Mbps, and if the video server is in South Africa or Europe, every second of video requires a round trip of 100-400ms across submarine cables, adding to buffering and startup time. This creates demand for content localisation, placing content servers in Kenyan data centres so that popular content is served locally rather than fetched from overseas every time. Content delivery networks (CDNs) like Cloudflare, Akamai, and Google have established caching nodes in Kenya for precisely this reason, and their presence in Kenyan data centres is a direct result of increasing internet speeds. ### Cloud and SaaS Adoption ![Cargo ship at the Port of Mombasa](/images/mombasa-cargo-ship.webp) Cloud computing and Software-as-a-Service applications require reliable, reasonably fast internet connections. As Kenya's internet speeds have improved, cloud and SaaS adoption has accelerated, more organisations are moving their applications to the cloud, which requires server capacity in data centres (whether local or overseas). The [cloud services comparison guide](/articles/cloud-services-kenya-compared) covers the providers serving Kenya. ## The Urban-Rural Speed Divide ![Technician joining aerial cables on a utility pole from a bucket truck](/images/aerial-fibre-technician-utility-pole.webp) One of Kenya's most significant digital infrastructure challenges is the gap between internet speeds in urban and rural areas. In Nairobi's wealthy neighbourhoods, fibre connections deliver 100-500 Mbps. In rural areas of North Eastern or parts of the Rift Valley, mobile connectivity may be limited to 3G with speeds of 2-5 Mbps, or may be unavailable entirely. This divide affects data centre demand unevenly. Urban users, with fast connections, generate the most data centre demand, they use cloud applications, stream video, and consume data-intensive services. Rural users, with slower connections, generate less demand per user but represent a large and growing market as mobile coverage expands. For data centre operators, the urban-rural divide reinforces the concentration of demand in Nairobi. As internet speeds in rural areas improve (through 4G expansion, satellite internet like Starlink, and government connectivity programmes), demand for data centre services from rural-connected users will grow, but this demand will be served by the same Nairobi-based facilities, because the economics of building data centres in rural areas do not work at current demand levels. ## 5G: The Next Speed Frontier Safaricom's 5G deployment, while still in its early stages, represents the next step in Kenya's internet speed evolution. 5G's promised speeds of 100-1000 Mbps, combined with sub-10ms latency and the ability to connect far more devices per cell tower, will enable new categories of data-intensive applications that drive data centre demand. These applications include augmented and virtual reality (which require both high speed and low latency), connected vehicles, remote healthcare (telemedicine, remote surgery), smart city infrastructure, and massive IoT deployments. Each of these applications generates data that must be processed somewhere, and increasingly, that somewhere will include edge computing facilities in Kenyan data centres. ## The Outlook Kenya's internet speeds will continue to improve. 4G coverage is expanding, 5G is rolling out, fibre-to-the-home deployments are extending beyond Nairobi, and satellite internet services (Starlink) are providing high-speed options for previously unserved areas. Each speed improvement drives data consumption growth, which drives data centre demand. The data centres being built today are preparing for the traffic that tomorrow's faster connections will generate. For Kenya's data centre industry, the message is straightforward: faster internet speeds are the fuel that powers demand. Every improvement in speed (whether from 4G expansion, 5G rollout, fibre deployment, or submarine cable upgrades) translates directly into more traffic that needs to be processed, stored, and served from data centres. The relationship between internet speed and data centre demand is not just correlated, it is causal. ## Kenya's KSh 13 Billion Fibre Backbone Tender, Explained URL: https://data-centers-254.vercel.app/articles/kenya-kdeap-fibre-backbone-tender Cluster: Connectivity Published: 2026-10-14 Updated: 2026-10-14 Summary: Kenya has opened a KDEAP-funded tender for at least 27,000 km of backbone and cross-border fibre, backed by KSh 13 billion. What is verified, what it means. On Tuesday 13 October 2026, the government of Kenya put a number on one of its biggest connectivity bets. At a pre-bid session in Nairobi for the framework agreement tender on national backbone and cross-border links, Principal Secretary for ICT and the Digital Economy Eng. John Kipchumba Tanui announced that the state is making **KSh 13 billion** available to catalyse matching private investment and deliver **at least 27,000 km of backbone, middle-mile and cross-border fibre**. The session brought together internet service providers, telecommunications operators, investors, financiers and development partners, with the CEO of the ICT Authority, Jesse Maruti, also present according to the PS's account. The tender sits under the Kenya Digital Economy Acceleration Project, better known as KDEAP, the World Bank-supported programme that funds much of Kenya's current digital infrastructure build-out. It is not a small procurement: the framework covers two lots, national backbone links and cross-border links, and will run for an initial three-year term with the option to extend by up to two further years. For anyone tracking Kenya's digital infrastructure, from fibre networks to the data centres they feed, this is one of the most consequential procurements of the year. ![Principal Secretary Eng. John Tanui addressing the KDEAP backbone pre-bid session in Nairobi](/images/ps-tanui-kdeap-prebid.webp) ## What the tender actually buys The ICT Authority's advert, published on 15 September 2026 on its open tenders page, describes a "framework agreement for extending national backbone and cross border links" in two lots: Lot 1 for national backbone and Lot 2 for cross-border links. Trade press coverage from September, including ITWeb Africa and Telecompaper, describes the structure as closed framework agreements, which means the authority pre-qualifies a pool of suppliers first, and specific build-out contracts are then awarded against that framework as volumes are confirmed. The Principal Secretary's September announcement spelled out the scope in the same terms: national backbone links, cross-border and metropolitan links, under a three-year framework. Bids close at 10:00 a.m. EAT on 29 October 2026. Both local and international infrastructure providers are invited to participate in what the PS called a competitive procurement, and the stated goals are resilience and redundancy for national infrastructure, more connected public institutions, affordable access, and the extension of last-mile connectivity by private operators to homes and businesses. ## Where the money comes from KDEAP is the financing engine behind this tender. The project is implemented by the ICT Authority under the Ministry of Information, Communications and the Digital Economy and supported by a US$390 million World Bank IDA credit running from 2023 to 2028, according to the project's official site. Its first phase concentrates on three priority areas: digital infrastructure and access, digital government and services, and digital skills and markets. The backbone, last-mile and cross-border connectivity expansion sits squarely in the first of those, and the project has a stated aim of mobilising an estimated US$100 million in private capital for broadband infrastructure. The KSh 13 billion, roughly US$100 million at recent exchange rates as reported by trade outlet W.Media, is the public side of that public-private equation. The mechanism for allocating it is a **Multi-Round Reverse Auction (MRRA)**, in which the missing links have been jointly identified and verified, according to the PS, and public support goes to the operators that require the least subsidy to build them. The approach is deliberately different from a conventional government build: instead of the state laying fibre itself, it subsidises private operators to build routes that would otherwise not be commercially viable, and the auction format is designed to drive the required support down round by round. ![Fibre optic termination equipment in a data centre](/images/dc-fibre-optics.webp) ## The market the tender is betting on The government's case for the investment rests on sector statistics from the Communications Authority's April to June 2026 report, as cited in the PS's post. Mobile subscriptions stand at 88 million, mobile broadband at 54.9 million and mobile money subscriptions at 54 million. The mobile figures describe SIM subscriptions rather than unique individuals, and independent reporting for June 2026 puts SIM subscriptions at 87.99 million, consistent with the government's figure. On the fixed side, the numbers are smaller but growing faster: fixed internet connections reached 2.84 million by June 2026, up 32.4 percent year on year, including 1.57 million fibre connections, up 29.7 percent, figures corroborated by the Communications Authority's own public communications and independent reporting. The fibre base has been expanding quickly. Since 2022, the private sector has added 24,000 km of fibre and the public sector 17,000 km, according to the Principal Secretary, figures reported by People Daily. International capacity stands at approximately 28.95 Tbps across Kenya's submarine cables and cross-border links, the enabler behind the country's cloud services, digital trade and BPO ambitions. Against that base, the government's ambition of more than eight million fixed broadband connections implies nearly a threefold expansion of today's fixed market, and that is the demand the 27,000 km of new fibre is meant to serve. ![Fibre optic cables in a terrestrial handhole and duct network](/images/fibre-handhole-duct-cables.webp) ## Why backbone economics matter for data centres For Kenya's data centre market, this tender touches the input costs that everything else sits on. Backbone and middle-mile routes carry traffic between submarine cable landings at Mombasa, inland population centres and the facilities listed in [the DC254 tracker](/tracker), which last verified 27 tracked facilities with 20 operational in October 2026. Better-resourced routes mean operators can buy transit and backhaul on tighter terms, diversify away from single-path risk, and serve customers outside Nairobi without sacrificing latency guarantees. [Kenya's fibre optic networks and data centres](/articles/fibre-optic-networks-kenya-data-centres) explains how that transport layer works in practice. The cross-border lot matters just as much. Landlocked neighbours in the region currently depend on routes that transit Kenya, and [submarine cables landing at Mombasa](/articles/submarine-cables-landing-mombasa) have historically made the country the region's natural gateway. New cross-border fibre under a Kenyan-led framework strengthens that position and deepens the case for regional traffic exchange and hosting in Nairobi, a dynamic we examined in [East African data centres compared](/articles/east-africa-data-centres-compared). It also complements domestic initiatives such as [Kenya's compute hub plan](/articles/kenya-data-centre-compute-hub-plan-explained), which depends on affordable, redundant connectivity to be credible. ![Submarine cable landing beach on the Kenyan coast](/images/amaco-cable-landing-beach.webp) ## What is confirmed, what is attributed and what is still unknown Following the verification standard this site applies to every claim, the picture divides cleanly into three tiers. **Verified against official or independent sources:** - The tender exists, in two lots (national backbone, cross-border), published on the ICT Authority's open tenders page on 15 September 2026. - The framework structure is a closed framework agreement with a three-year initial term, extendable by up to two years (ITWeb Africa, Telecompaper, AllBusiness Africa). - Bids close at 10:00 a.m. EAT on 29 October 2026, stated in the Principal Secretary's September 2026 post. - KDEAP is funded by a US$390 million World Bank IDA credit for 2023 to 2028, per the project's official site. - Fixed internet reached 2.84 million connections by June 2026, up 32.4 percent, with 1.57 million fibre connections, per the Communications Authority's own communications and independent reporting. **Attributed to the Principal Secretary's post (single-source until further evidence):** - The KSh 13 billion catalytic funding envelope. - The at least 27,000 km fibre target. - The Multi-Round Reverse Auction as the allocation mechanism. - The 24,000 km private and 17,000 km public fibre additions since 2022. - The 54.9 million mobile broadband and 54 million mobile money subscription figures. - The ambition of more than eight million fixed broadband connections. **Still unknown:** - Which specific missing links the MRRA will target, and their individual lengths and budgets. - How much matching private capital the auction will require per route. - The award timeline after the 29 October close, and when the first call-offs against the framework would happen. - Any delivery schedule connecting the framework to actual lit routes. ![Students learning ICT skills in a Kenyan classroom](/images/classroom-ict-training-kenya.webp) ## What to watch The immediate date is 29 October 2026, when bids close at 10:00 a.m. EAT. What follows matters more: the composition of the framework pool, the size of the subsidies the reverse auction settles on, and how quickly the first routes move from award to construction to lit fibre. Each of those steps converts a headline number into verifiable infrastructure, and each is checkable in public procurement records and operator disclosures. DC254 will keep tracking the tender's progression alongside the facilities, capacity and pipeline in [the DC254 tracker](/tracker), and updates to this article will be dated and evidenced. For the underlying transport-layer context, see [the Kenya Internet Exchange Point](/articles/kixp-internet-exchange-point-kenya), [Kenya's digital economy and data centre demand](/articles/kenya-digital-economy-data-centre-demand), and the quarterly [State of the Market 2026-Q3](/research/state-of-the-market-2026-q3) report. ## Frequently asked questions ### What is the KSh 13 billion for? According to Principal Secretary Eng. John Tanui, the government is making KSh 13 billion available under the Kenya Digital Economy Acceleration Project to catalyse matching private investment and deliver at least 27,000 km of backbone, middle-mile and cross-border fibre. The money is not a single construction contract: it is public support allocated competitively through a Multi-Round Reverse Auction to operators who win places on the framework agreements. ### What is KDEAP? KDEAP, the Kenya Digital Economy Acceleration Project, is a government programme implemented by the ICT Authority under the Ministry of Information, Communications and the Digital Economy and supported by a US$390 million World Bank IDA credit running 2023 to 2028. Its first phase covers digital infrastructure and access, digital government and services, and digital skills and markets. The backbone and cross-border connectivity tender falls under its infrastructure component. ### Who can bid and when does the tender close? The ICT Authority's advert, published 15 September 2026, covers two lots: Lot 1 for national backbone links and Lot 2 for cross-border links, structured as closed framework agreements with an initial three-year term extendable by up to two further years. Both local and international infrastructure providers are invited, according to the Principal Secretary's announcement. Bids close at 10:00 a.m. EAT on 29 October 2026, per the PS's September 2026 post. ### What is a Multi-Round Reverse Auction? A reverse auction reverses the usual roles: many suppliers compete to offer a service at the lowest required public support, rather than a buyer choosing among rising bids. In a multi-round design, successive bidding rounds drive the subsidy down until only the most efficient operators remain. Kenya is applying this model so that the KSh 13 billion stretches further, with winners expected to mobilise matching private capital. ### What does the tender mean for Kenya's data centres? Backbone and cross-border fibre is the transport layer that connects data centres to submarine cable landings, internet exchange points and customers across the region. More routes and better redundancy can lower bandwidth costs and improve resilience for the 20 operating facilities in the DC254 tracker, and strengthen the case for regional traffic to be hosted in Kenya rather than abroad. The effect depends on delivery: routes must actually be built, lit and commissioned. ## LuLu Cable: Kenya's 500km Coastal Route Beyond Mombasa URL: https://data-centers-254.vercel.app/articles/kenya-lulu-coastal-cable-system Cluster: Infrastructure Published: 2026-09-10 Updated: 2026-09-16 Summary: The LuLu sea cable: a 500km coastal route from Mombasa to Lamu with five landing points and 144 fibre pairs. What is confirmed, and why it matters. Every one of Kenya's international submarine cables converges on a short stretch of shoreline around Mombasa. That concentration built the country's digital economy, and it is also a single- geography risk the industry has lived with for seventeen years. At ITW Africa 2026, a consortium led by INDOI Ltd. (IOX) and Blue Trade Investments Limited announced LuLu, a planned coastal cable system designed to spread that risk along roughly 500 kilometres of Kenyan shoreline, from Mombasa all the way to Lamu. LuLu is not another international system competing with TEAMS, SEACOM, EASSy or 2Africa. It is something Kenya has never had: a dedicated coastal backbone that connects five Kenyan locations to each other and to the international landing stations, pairing a subsea route with a protected terrestrial path the whole way. The announcement is real, the consortium is named, and the technical specification is unusually detailed for a first disclosure, but so are the caveats, because as of September 2026 no supplier has been signed and no construction has started. ## What Was Actually Announced The system was announced at ITW Africa 2026 with the route, partners and technical architecture all disclosed. LuLu is designed to cover about 500km between Mombasa and Lamu, with five landing points identified along the corridor: Mombasa, Vipingo Special Economic Zone, Kilifi, Malindi and Lamu. The consortium behind it splits responsibilities cleanly: INDOI Ltd. (IOX) is identified as the developer, owner and commercial operator, while Blue Trade Investments Limited is listed as the Kenyan landing party, regulatory interface and delivery partner. They are joined by ARISE IIP (the industrial platform developer behind Vipingo SEZ) along with AfriTrade Consulting Group and Kingdom Bank. The inclusion of ARISE IIP tells you what this project is really about. LuLu is not just a cable; it is infrastructure for the industrial and logistics corridors emerging along Kenya's coast, with the Vipingo SEZ landing explicitly positioned to support data centre, cloud, business process outsourcing, financial technology, cybersecurity and smart manufacturing use cases inside the zone. Kilifi, Malindi and Lamu gain landing stations of their own, towns that today depend entirely on terrestrial fibre backhaul from Mombasa or Nairobi. ## The Build: Dual Paths, 144 Fibre Pairs, and a 25-Year Design Life Technically, LuLu is specified as a dual-path system, pairing a subsea route with a protected terrestrial path running parallel along the corridor. The subsea span is expected to use DWDM technology in the 1550nm window, with 144 fibre pairs and a design capacity of up to 60 Tbps per fibre pair, on paper, a theoretical maximum north of 8,600 Tbps, though real lit capacity depends on the line systems and economics at deployment time. The design life is a planned 25 years, with centralised network management specified across both paths. The dual-path architecture is the interesting engineering choice. Submarine cable faults near shore are usually caused by anchoring and trawling; terrestrial routes fail from roadworks, theft and flooding. By pairing each, LuLu is designed so that a single incident on either path does not take the corridor down, availability through diversity rather than through spare capacity alone. For a cable whose entire premise is resilience, that is the right instinct, and it mirrors the terrestrial protection rings that national fibre operators like Liquid have built inland. ## Why Coastal Diversity Matters for Kenya The case for LuLu starts with geography. Kenya's seven live international cable systems land in the Mombasa area, which means every gigabit flowing into and out of the country (and much of East Africa's) passes through one metropolitan shoreline. A ship anchor drag off Mombasa has repeatedly degraded regional connectivity, and each incident restarts the same conversation about redundancy. Inland fibre routes to Nairobi and the landlocked neighbours mitigate part of the exposure, but the landing concentration itself has never had a coastal alternative. LuLu would give carriers and content platforms a second physical geometry: traffic can enter at one landing point, transit the coastal corridor, and exit at another, with the terrestrial path as backup. The second case is economic development. Vipingo SEZ is the anchor tenant story (an industrial zone with power, logistics and now a planned fibre landing positioned for data centre and fintech tenants) but Kilifi, Malindi and Lamu all stand to gain direct access to high-capacity infrastructure that currently terminates hundreds of kilometres away. Coastal tourism economies are also digitising fast, and local landing points reduce the cost of serving them. ## Status Check: Announced, Not Built This is the section to re-read before repeating any headline. The confirmed status is that LuLu has been announced: route, partners and intended technical architecture disclosed. The next stated milestone is progression toward readiness for service in Q2 2028, and the consortium itself frames that target as subject to contract execution. As of the September 2026 announcement, no system supplier has been named, no construction start or route-survey schedule has been disclosed, and no supply contracts have been completed. That is not an indictment (most cable systems announce before they contract) but it is the difference between a project and a plan. The markers to watch are concrete: a system supplier award, marine route survey activity off the Kenyan coast, landing party agreements for the five sites, and the consortium's financing structure firming up. Until those land, the Q2 2028 date should be treated as a target rather than a committed in-service date. ## What It Means for Data Centres For Kenya's data centre industry, LuLu matters in two specific ways. First, route diversity: Nairobi facilities now reach international capacity through terrestrial fibre to a concentrated set of Mombasa landing stations. A coastal ring with five additional landing points creates new physical paths for protection circuits and adds resilience for the carriers, cloud on-ramps and enterprise WANs that data centre tenants sell. Second, edge locations: a landing station at Vipingo SEZ makes a small data centre or edge node inside the zone instantly viable, because the expensive part (international-grade fibre access) arrives as infrastructure. Facilities inside the zone could serve BPO campuses, fintech workloads and smart manufacturing with local breakouts. What LuLu does not do is add international capacity. Kenya's trans-ocean connectivity still rides on the eight systems at Mombasa, and the country's bandwidth economics will keep being set by those systems and their owners. For the authoritative count and history of the international cables, see our full explainer on the [submarine cables landing in Mombasa](/articles/submarine-cables-landing-mombasa), and for how landing stations connect to facilities inland, read about [data centre interconnection and peering in Kenya](/articles/data-centre-interconnection-peering-kenya). Sources: SubTel Forum (9 September 2026); African Media Agency release via ITW Africa 2026; Developing Telecoms (9 September 2026). Status facts (consortium, route, specification, and the contract-dependent Q2 2028 target) verified against SubTel Forum's detailed report on 10 September 2026. ## PHP2Shell: What Kenya's Government Website Checks Show URL: https://data-centers-254.vercel.app/articles/kenya-php-advisory-php2shell Cluster: Security Published: 2026-10-22 Updated: 2026-10-22 Summary: Two critical PHP flaws prompted urgent patching worldwide. DC254 ran a public check of Kenya's government websites: what the signals show, and what they cannot. Two flaws in the PHP runtime, both rated CVSS v4.0 9.2, put server operators on four continents into patch mode this week. CISA urged immediate updates in Washington. Kenya's Ministry of Information, Communications and the Digital Economy issued a critical advisory to ministries, departments and agencies on 21 October, in collaboration with the ICT Authority. We read the advisory, checked the public surface of 15 government websites the next morning, and wrote down what the signals show and where they stop. The signals show a PHP-based estate. Nine of the ten government websites that answered us carried signatures consistent with PHP-based content platforms. One response header disclosed an actual runtime version, and it was PHP 7.4.30, a branch that left security support in 2022. What the signals cannot show is patch status on the other nine, because a CMS marker is not a version number. That gap between what an outsider can see and what an operator must verify is the story of this week, and it is the reason the advisory exists. ## What happened On 21 October 2026 the PHP project shipped security releases for two vulnerabilities in the language core: CVE-2026-109618 and CVE-2026-109619. CISA, the US cyber defence agency, named the fixed releases the same day: **PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35**. Until a server runs one of those versions, CISA's alert describes the exposure plainly: unauthenticated remote code execution on PHP installations serving web requests. Wiz Research found the bugs. Google Cloud, which owns Wiz, published the disclosure jointly with the PHP project and credited a mix of autonomous AI agents with locating the flaws and helping draft the remediation. Wiz estimates the bugs had been sitting in PHP's codebase for over a decade. The release also came a day early: technical details leaked on social media before the planned 22 October disclosure, and the PHP project's pre-release notice had reserved the right to publish early in exactly that situation. watchTowr documented the leak and the original schedule. The PHP advisory describes both bugs as exploitable before any application code runs. In the first, a memory-exhaustion event leaves a request half-initialised and the cleanup routines treat attacker-controlled input as functions to execute. In the second, request state outlives its own request, and the next request handled by the same worker inherits it. Deployments using mod_php carry the highest risk, because the runtime shares the web server's process. CISA had received no reports of exploitation when it published. The response was broad. France's interior ministry took systems offline as a precaution. The Netherlands' NCSC and the Port of Rotterdam published urgent notices, Finland's state IT agency Valtori told its agencies to patch, and Wikipedia applied preventive measures to its own infrastructure. Cloudflare shipped managed WAF rules across all its plans, including the free tier, and CERT-EU issued advisory 2026-018 for EU institutions. Two footnotes matter for anyone planning a response. First, workarounds circulated within hours, and Datadog's researchers tested the community configurations: they blocked some attack paths and left others open, so the vendor fix remains the only complete remedy. Second, PHP 8.1 and older branches receive no regular upstream releases. The PHP project published minimal source backports so that OS vendors can build patched packages of their own, and CISA's advice for systems that cannot be updated is to take them offline until they can be. ## What the advisory requires The Kenyan advisory is dated 21 October 2026 and went out under the Ministry of Information, Communications and the Digital Economy in collaboration with the ICT Authority. The document itself circulated to ministries, departments and agencies; Kenyan outlets including The Star and Tuko reproduced its contents, and we cite those reports as the source for what it says. It describes the two flaws, their CVSS 9.2 rating, and the remote code execution conditions, then instructs agencies to act. The measures, as reported: identify every PHP-based platform in use, patch internet-facing systems first, limit the exposure of systems that cannot be patched immediately, step up monitoring, apply web application firewall protection where available, review third-party applications for PHP exposure, and report suspected incidents to KE-CIRT/CC, the national computer incident response team at the Communications Authority. Response handling of any personal data involved falls under the Data Protection Act, which we have covered for data centre operators [elsewhere](/articles/kenya-data-protection-act-data-centres). An advisory is an instruction to act. It is also not evidence that any Kenyan system was breached, and readers should be careful not to read one into it. We found no public confirmation of exploitation of these specific flaws in Kenya as of publication. ## What we checked, and how This is a research note, not a scan report, and the distinction is the point. On 22 October 2026 we sent one plain HTTPS GET request to the homepage of each of 15 well-known .go.ke hostnames and recorded the status code, the Server and X-Powered-By headers, and page body markers such as wp-content paths, Drupal and Joomla identifiers, and generator meta tags. No path scanning, no subdomain enumeration, no exploitation attempts, nothing beyond what a visitor's browser receives on a normal page load. The probe is about ninety lines of Python and ships in this site's open repository (scripts/govke-php-probe-20261022.py), so any technical reader can re-run it against any domain list in minutes. | Hostname | Status | What the response showed | | --- | --- | --- | | president.go.ke | 200 | WordPress (generator tag), behind Cloudflare | | immigration.go.ke | 200 | WordPress (generator tag, 6.6.2) | | konza.go.ke | 200 | WordPress (generator tag) | | judiciary.go.ke | 200 | WordPress with Elementor; headers disclose PHP/7.4.30 | | ict.go.ke | 200 | Drupal (generator tag: Drupal 10) | | icta.go.ke | 200 | Drupal markers | | treasury.go.ke | 200 | Drupal (generator tag: Drupal 10) | | health.go.ke | 200 | Drupal (generator tag: Drupal 9) | | kra.go.ke | 200 | Joomla markers | | mombasa.go.ke | 200 | No CMS markers (openresty server header) | | ecitizen.go.ke | 403 | No body served; Server header names cloudflare | | statehouse.go.ke | no response | Connection failed during our window | | education.go.ke | no response | Connection failed during our window | | police.go.ke | no response | Connection failed during our window | | kenya.go.ke | no response | Connection failed during our window | Ten of fifteen hostnames answered. Nine showed signatures consistent with PHP-based CMS platforms. One response in the set disclosed a runtime version: judiciary.go.ke serves X-Powered-By: PHP/7.4.30 alongside a matching Apache server header. PHP 7.4 left security support in November 2022, so that branch has no upstream fix for these CVEs; the backport route, if one is available, runs through the OS vendor. A header can be stale or set by a proxy, so it needs verification on the ground. Our claim is narrower: a public response advertises a runtime branch that the official fix does not cover, and that is worth a phone call by whoever owns the server. Four limitations bound everything above, and we would rather list them than have a reader discover them. - A CMS marker is not an inventory. WordPress, Drupal and Joomla all run on PHP, so the markers tell us the platform class, and they cannot tell us the patch level. Only the one disclosed header pins a version. - One homepage, one request, one day. Anything behind a login, an API path, or a subdomain is invisible to this check, and most real exposure lives in exactly those places. - A 403 or a dropped connection tells an outsider almost nothing. It could be a firewall, a blocklist, a restart, or a routing fault, and we cannot distinguish those from outside. We make no claim about eCitizen beyond what the table shows. - The estate we could not see is larger than the estate we could. Counties, universities, hospitals and thousands of small businesses run PHP too. They sit outside this check entirely and belong to the wider risk picture, not to our findings. ## The data centre angle Every PHP server in Kenya sits somewhere physical. The facility's responsibility ends at power, cooling, connectivity and physical access; the runtime is the tenant's problem, or the managed host's, depending on the contract. Biometric doors and CCTV do not parse HTTP requests, and a Tier III certification speaks to the availability of the plant, not to what a tenant's application does with a malformed request. Readers who want the physical layer mapped out can start with our guide to [data centre physical security in Kenya](/articles/data-centre-physical-security-kenya). Concentration changes the arithmetic. Kenya's commercial hosting estate clusters in a small number of facilities, most of them in Nairobi, which you can verify facility by facility in our [DC Directory](/directory). When many customers share one building, one unpatched runtime in that building raises the potential blast radius for every network it touches. The actual risk in any specific facility depends on segmentation, isolation, permissions and architecture, which is why every advisory, Kenya's included, puts inventory before everything else. It is also why we track the threat volume against Kenyan networks separately from the market numbers in [our data centre coverage](/articles/kenya-data-centre-market-numbers). Kenya has already paid the reputational tuition this year. In July 2026 the presidency's website was defaced and taken offline for a period, an incident that made national headlines. The operational cost was modest by data centre standards. The trust cost was not. A runtime flaw that executes before the application loads is a broader version of the same exposure, because it does not depend on a weak password or a missed CMS update. If you buy hosting rather than operate it, three questions to your provider are worth more than any checklist this week: 1. Which PHP version runs my stack, and when does your fixed release land? 2. Is my application on mod_php, on FPM behind a proxy, or in a container image I never see? 3. Where does your responsibility end and mine begin, in writing? These are procurement questions. Asking them before an incident costs nothing. Asking them after one costs a post-mortem. One more thing follows from how this bug was found. Wiz located a decade-old flaw with autonomous agents, and the same class of tooling will keep finding others. Kenya's data centre market is courting AI workloads, from [GPU cloud buildouts](/articles/gpu-cloud-infrastructure-kenya) to AI-ready halls. AI-ready should mean more than GPU power and cooling; it should mean patch discipline that acts at the speed AI-assisted research now moves. ## What to do now The advisory's measures and CISA's actions overlap almost exactly, so the list below merges both. It takes an afternoon on a small estate and a proper change window on a large one. 1. Inventory every PHP deployment, including the ones inside appliances and vendor packages. CISA notes PHP often ships as a component nobody counts. 2. Patch internet-facing systems first to PHP 8.5.12, 8.4.27, 8.3.36 or 8.2.35. 3. If you run PHP 8.1 or older and cannot patch this week, CISA's guidance is to take the system offline until you can. Ask your OS vendor for a backport timeline in parallel. 4. Treat WAF rules as breathing room while you patch, and nothing more. CISA says a WAF may provide some protection; Datadog found the community workarounds incomplete. 5. Watch logs for PHP worker processes launching shells or downloading tools, per CISA's monitoring guidance. 6. Check third-party and vendor PHP applications against the same fix list, per the advisory's third-party measure. 7. Route suspected incidents to KE-CIRT/CC, and handle any personal data surfaced in a response within the Data Protection Act. One date belongs in procurement documents. Under php.net's published schedule, PHP 8.2 leaves security support on **31 December 2026**, two months from this week. Contracts that specify supported runtimes should carry that date, and any clause naming a version should name the security end date of that version too. The inventory, the patch and the monitoring are small jobs on their own. The habit they build is the part that outlasts this specific bug, because the next runtime disclosure will arrive with the same urgency and, if this week is a guide, with less warning than the calendar promised. ## Kenya on the Continental Pipeline: The Next Decade URL: https://data-centers-254.vercel.app/articles/kenya-pida-infrastructure-pipeline Cluster: Policy Published: 2026-09-08 Updated: 2026-09-16 Summary: Kenya has 186 projects worth about $172bn in AUDA-NEPAD's infrastructure database, including a $395m Mombasa data centre hub. What the registry shows. There is a database that knows what Africa intends to build over the next decade, and almost nobody in the Kenyan data centre conversation has ever queried it. The **African Infrastructure Database (AID)**, maintained by AUDA-NEPAD, the African Union's development agency, is the continental registry behind the Programme for Infrastructure Development in Africa (PIDA). It tracks every serious cross-border infrastructure project on the continent: ports, railways, power pools, pipelines, water systems and, crucially for our beat, the digital layer. Its public API is open, keyed, and (as of 8 September 2026, when we pulled Kenya's records) browsable from right here on DC254. What we found reframes Kenya's infrastructure story in two directions at once. The registry confirms Kenya as the corridor economy of East Africa: the Mombasa–Nairobi spine appears again and again in continental-scale projects. And it contains a project most Kenyan data centre coverage has never mentioned: a **$395 million regional smart hub with data centre capacity, registered at Mombasa**, being developed as a public-private partnership under PIDA's second Priority Action Plan (PAP 2). ![Continental infrastructure registry](/images/dc-policy-regulation.webp) ## The headline numbers Kenya's registry holds **186 projects with roughly US$172 billion in registered capex across four sectors** (transport, energy, water and ICT) as of our 8 September 2026 fetch. The public endpoint serves the first 100 records per request; within that slice, transport dominates (42 projects, including the continental flagship: the $19.2 billion Mombasa–Nairobi–Malaba–Kampala–Kigali standard-gauge railway programme), followed by water (41) and energy (12), with **five ICT-sector projects**. Status labels are registry-speak: 89 of the 100 carried "Active", nine "Completed", two unlabelled. The completed list is a reminder of what "infrastructure" means at continental scale: the **$5 billion Mombasa–Nairobi SGR section**, the Nairobi Southern Bypass, the Mombasa Port new container terminal, the Malaba one-stop border post. These are the corridors and gateways that every Kenyan data centre's diesel, turbines, and fibre drums travel on, and they are the context investors mean when they call Kenya "the logistics hub of East Africa". ## The ICT projects: Kenya's registered digital pipeline Five ICT projects appear in the served slice, and they sketch a coherent strategy: Kenya as the digital gateway for its landlocked and coastal neighbours. **The flagship: "Transborder Submarine Fiber PoPs, Regional Smart Hub Facility and Data Centre" (PIDA I.02.30.05, US$395 million, Active).** Registered at Mombasa: the coordinates land in the Miritini area, home to Digital Realty's MBA1, the most interconnected building on the Kenyan coast, the project is coordinated by the EAC and IGAD under the IGAD Regional Infrastructure Masterplan and is being advanced as a PPP under PIDA PAP 2, serving six countries (Kenya, Uganda, Tanzania, South Sudan, Somalia, Ethiopia). Read carefully: this is a *project preparation*, at Project Structuring stage as of the 2024 milestone, not a construction site. But a continental registry explicitly reserving data centre capacity at Mombasa, in a PPP wrapper with six-country backing, is a signal of institutional intent that no local market report captures. If it proceeds, it would stack regional data centre demand directly onto the cable-landing geography where seven live submarine cable systems surface (nine in total). **The Juba–Nairobi Fiber Optic Link (PIDA I.02.30.06, US$45 million).** The Kenyan segment (roughly 700 km, part of the national NOFBI fibre backbone) is already operational. The South Sudanese side has been delayed by security and financing; the registry notes the August 2023 Ruto–Kiir memorandum that revived it. For Nairobi's data centre operators, this is landlocked-demand plumbing: every kilometre of completed South Sudanese fibre is a new customer for Kenyan colocation and transit. **Two Somalia cross-border links: Garissa–Kismayo and Nairobi–Mogadishu**, each with a point of presence at the Somali end, registered without capex figures. These are the northern extension of the same strategy, Kenyan fibre and Kenyan hubs anchoring neighbouring markets' connectivity. **The DARE1 registry record**, the Djibouti Africa Regional Express submarine cable (4,763 km, landings including Mombasa), which carries a useful lesson in data quality: the registry's summary cites a **60 Tbit/s system design**, while the conservative per-landing figure we carry on our cable layer is 0.96 Tbit/s. Both derive from the cable's documentation at different moments and scopes; we publish the divergence rather than silently picking one, and cross-check against TeleGeography's live map before making capacity claims. ## The power entry that matters to every data centre One energy registry entry deserves special attention from our readers: the **ZTK Transmission Interconnector, Kenya section (PIDA E.02.03.02, US$50 million, Active)**, the Kenyan segment of the Zambia–Tanzania–Kenya 400 kV interconnector, Kenya's bridge into the Southern African Power Pool via the Eastern Africa Power Pool. Why does a transmission line belong in data centre coverage? Because the binding constraint on Kenyan digital infrastructure is not demand, land, or capital, it is the grid's ability to deliver tens of megawatts to a single site, the exact problem that has the Microsoft–G42 Olkaria project reported paused since May 2026. Interconnectors turn a national grid with a ~2.3% reserve margin into a regional one that can trade surpluses and shortfalls; every megawatt of traded capacity is a megawatt of data centre demand that becomes servable. Registry entries like ZTK are the unglamorous plumbing behind every "Kenya is open for AI" headline. The same logic runs in reverse: the **LAPSSET corridor projects** (the $3 billion crude pipeline entry, the $13.5 billion railway) and the SGR programme define where coastal power, fibre and logistics will thicken over the next decade. Data centre siting in Kenya is corridor-adjacent by history, Mombasa Road exists because the highway and the fibre do, and the registry shows where the next thickening happens. ## How to use a registry responsibly The PIDA layer is now live on [our infrastructure map](/infrastructure/map): violet diamonds, separate from the facility markers, with the full project table beneath. Three rules govern how we use this data, and how you should: **First, a registry entry is not a facility.** "Active" in registry-speak means the project is in the continental pipeline, it may be at feasibility, structuring, or financing stage. The Mombasa smart hub's 2024 milestone was *Project Structuring*. Counting registry entries as built infrastructure is how continental "Africa's data centre boom" listicles end up doubling capacity that exists only in project documents. Our [DC Directory](/directory) and the PIDA layer are deliberately separate datasets that never share a sum. **Second, registry numbers are the registrant's own.** Capex, capacity, and coordinates are published as registered, without independent verification, the DARE1 divergence above is the live example. We republish them dated and attributed, not confirmed. Where a registry claim can be checked against an independent source (PeeringDB, operator pages, cable maps), we note the comparison. **Third, the gaps are informative.** Kenya's registry holds 186 projects, but only five in ICT, a ratio that itself tells a story about where continental attention sits versus where Kenya's private digital investment actually flows. The market briefs and private capital documents we cover elsewhere (the Xalam/D4D Hub briefing, the EU–Kenya Digital Dialogue) describe a digital economy that the continental registry barely sketches. Reading both together, the public pipeline and the verified market, is the closest thing this sector has to a full picture. The API is public and keyed; our fetches are dated and repeatable. Expect this layer to be refreshed with each quarterly directory pass, and expect the same discipline here that governs everything else on this site: named sources, fetch dates, divergences published. ## Kenya Power and Data Centres: The Electricity Constraint URL: https://data-centers-254.vercel.app/articles/kenya-power-infrastructure-data-centres Cluster: Energy Published: 2026-08-26 Updated: 2026-09-28 Summary: Kenya's geothermal advantage attracts data centres, but grid reliability and transmission bottlenecks remain the biggest obstacle to growth. Kenya's data centre industry faces a paradox. The country has one of Africa's most attractive energy profiles (nearly 90% renewable electricity, dominated by geothermal from the Rift Valley) yet power supply is the single largest constraint on data centre growth. The stalled Microsoft-G42 one-billion-dollar AI data centre project made this painfully clear: you can have the best regulatory framework and the most strategic location, but if the grid cannot deliver the watts, the data centre cannot be built. This is not a theoretical problem. Kenya's total installed generation capacity stands at approximately 3,500 megawatts, of which about 45% comes from geothermal sources. That sounds substantial until you consider that a single large-scale AI training facility can require 200-500 megawatts, a demand that would represent 5-15% of the entire national grid's firm capacity. The question is not whether Kenya has enough electricity in aggregate, but whether it can deliver sufficient, reliable power to the specific locations where data centres operate. ![Data centre power distribution and backup systems](/images/dc-power-systems.webp) ## Kenya's Energy Mix: The Geothermal Advantage Kenya's electricity generation is one of the greenest in Africa. The energy mix is dominated by renewable sources: geothermal accounts for approximately 45% of installed capacity, hydroelectric for about 30%, wind for 5-8%, and solar a growing but still small share. Thermal (fossil fuel) generation fills the gaps during peak demand or when renewable output dips. ![Lake Turkana Wind Power turbines in northern Kenya](/images/kenya-wind-turbines-2.webp) Geothermal is Kenya's strategic asset for data centres. Unlike solar and wind, geothermal power is available 24 hours a day, 365 days a year, with minimal seasonal variation. The Olkaria geothermal field in the Rift Valley produces over 800 megawatts, and the Geothermal Development Company (GDC) continues to develop new wells. For data centre operators evaluating African locations, Kenya's geothermal baseload offers something few other African nations can match: clean, consistent power without the intermittency that makes solar and wind unsuitable as primary data centre energy sources. However, the geothermal resource is geographically fixed in the Rift Valley, approximately 100-150 km from Nairobi where data centre demand is concentrated. This creates a transmission bottleneck. The power exists, but moving it from generation source to load centre requires sufficient transmission infrastructure, and this is where the constraint becomes apparent. ## Understanding Data Centre Power Requirements Data centres are extraordinary electricity consumers. To understand why power is the binding constraint, it helps to understand the scale of demand: ![Diagram of the electrical chain from grid to GPU rack](/images/diagram-power-chain-grid-to-gpu.webp) - A single server rack typically draws 5-10 kilowatts - A small colocation facility with 100 racks needs 0.5-1 megawatt of IT power - A mid-size facility like Africa Data Centres Nairobi 1 (200 racks) needs 1-2 megawatts - A hyperscale facility like iXAfrica NBOX1.1 (500 racks) needs 4-5 megawatts - An AI training cluster can draw 50-100 megawatts in a single building - The largest planned facilities globally exceed 1 gigawatt (1,000 megawatts) But IT load is only part of the story. Data centres need additional power for cooling (removing the heat servers generate), lighting, security systems, and office space. The total facility power is measured by **PUE (Power Usage Effectiveness)**, defined as the ratio of total facility power to IT power. A PUE of 1.5 means for every 1 watt consumed by servers, 0.5 watts go to everything else. So a 5-megawatt IT load with a PUE of 1.5 requires 7.5 megawatts from the grid. How PUE is calculated, what Kenyan facilities typically achieve, and which improvements pay back fastest are covered in [our data centre PUE explainer](/articles/data-centre-pue-power-usage-effectiveness). ![Power hall inside a Kenyan data centre during a working site visit](/images/site-visit-power-hall-kenya.webp) ## The Transmission Bottleneck Kenya's national grid is managed by Kenya Power and Lighting Company (KPLC), which operates the transmission and distribution network. While generation capacity has grown substantially, transmission infrastructure has not kept pace. Key challenges include: **Insufficient transmission capacity to Nairobi.** The corridor from the Rift Valley geothermal fields to Nairobi carries the bulk of the country's power to its largest load centre. During peak demand periods, this corridor operates near its limits, and any disruption (a line fault, maintenance shutdown, or generation shortfall) can cause voltage instability or load shedding in Nairobi. ![High-voltage transmission towers crossing Kenyan rangeland toward distant mountains](/images/kenya-transmission-towers-riftvalley.webp) **Limited redundancy on critical paths.** A single transmission line failure can disconnect significant generation capacity from the grid. The system is resilient against small disruptions but vulnerable to simultaneous failures or large-scale events. **Distribution constraints in data centre zones.** Even when the national grid has adequate power, delivering it to specific data centre locations along Mombasa Road and other Nairobi corridors requires sufficient local distribution infrastructure. Building new substations and feeders for large data centre loads can take 2-4 years. ![Yard of a high-voltage substation with busbars, insulators and gantries under a wide sky](/images/kenya-hv-substation-yard.webp) The June 2026 Huduma Centre data centre outage in Nakuru demonstrated the practical consequences. A single power supply failure at one facility disrupted government services nationwide. While this was attributed to inadequate backup power at the facility, it highlighted the broader fragility of power supply to critical digital infrastructure. ![High-voltage pylons on the backbone grid](/images/kenya-transmission-pylons.webp) ## Why the Microsoft-G42 Project Stalled The most prominent illustration of Kenya's power constraint is the stalled Microsoft and G42 AI data centre project. Announced in early 2026 with a projected investment of one billion dollars, the facility would have been the largest single foreign direct investment in Kenya's digital infrastructure and a powerful signal to the global tech industry that Kenya was open for serious AI infrastructure. The project stalled in May 2026 when it became clear that the national grid could not reliably deliver the power the facility required. Reports indicated that meeting the project's energy demands would require rationing power to other consumers, a politically untenable proposition for any government. The project has not been formally cancelled, but there is no clear timeline for resolution. This was a missed opportunity at multiple levels. The investment would have created thousands of construction and operational jobs, positioned Kenya as an AI hub for East Africa, and catalysed further investment in the power infrastructure that would benefit the entire economy. Instead, it became a cautionary tale about the gap between regulatory readiness and infrastructure readiness. ## Backup Power: The Last Line of Defence Data centres do not rely solely on the grid. Every serious facility has a layered backup power architecture: **Uninterruptible Power Supply (UPS)**, Battery systems that provide instantaneous power during the seconds between a grid outage and generator startup. UPS systems typically bridge 10-30 seconds, though larger installations can sustain several minutes. They also condition the power, smoothing voltage fluctuations and cleaning harmonic distortion. **Diesel generators**, For extended outages, diesel generators can run for days or weeks, limited only by fuel supply. A typical large data centre has enough fuel on-site for 24-48 hours of operation, with contractual arrangements for refuelling during prolonged outages. **Dual utility feeds**, Facilities connected to two independent power feeds from the grid can continue operating even if one feed fails. This is standard for Tier III and Tier IV data centres and is typically a licensing requirement. The combination of these systems means that a well-designed data centre should never experience downtime due to power issues. The Huduma Centre outage suggests that not all facilities in Kenya meet this standard. For new operators entering the market, backup power is a significant capital cost, diesel generators, UPS systems, and automatic transfer switches can represent 10-15% of total facility cost. ![UPS and switchgear line-up in a data centre power room](/images/ups-switchgear-power-room.webp) ## The Path Forward ![Field technician working on overhead power lines](/images/grid-technician-pole-work.webp) Solving Kenya's data centre power constraint requires action on multiple fronts: **Transmission investment.** KPLC and the government need to accelerate investment in high-voltage transmission corridors, particularly from the Rift Valley to Nairobi and from Nairobi to data centre zones. This requires capital, planning approvals, and construction time, typically 3-5 years for major transmission lines. **On-site generation.** Some data centre operators are exploring on-site power generation, including solar arrays and potentially small geothermal plants. While solar can offset some daytime load, it cannot provide 24/7 baseload power without enormous battery storage, which adds cost and complexity. **Grid-scale storage.** Battery energy storage systems (BESS) can help manage peak demand and provide backup without diesel generators. Kenya's renewable-heavy grid makes storage particularly valuable for smoothing the output of variable sources. ![Diagram showing a solar farm, battery storage, wind farm and a data centre connecting to a central grid hub](/images/diagram-grid-connections-overview.webp) **Dedicated power infrastructure.** For very large facilities (50+ megawatts), building a dedicated power connection (essentially a private transmission line from a generation source to the data centre) may be the most practical solution. This is what hyperscale operators typically do in other markets. ![A large high-voltage power transformer on a plinth inside a substation bay](/images/hv-power-transformer-bay.webp) The policy environment is supportive. The [data centre licensing framework](/articles/kenya-data-centre-licensing-framework) provides regulatory clarity. The National Digital Superhighway Programme signals government commitment. But until the power constraint is addressed through tangible infrastructure investment, Kenya's data centre ambitions will remain capped by the available watts. ![Olkaria geothermal plant with cooling towers and steam](/images/kenya-geothermal-plant-olkaria.webp) ## Why Power Reliability Shapes Kenya's Data Centres URL: https://data-centers-254.vercel.app/articles/kenya-power-reliability-data-centres Cluster: Power & Energy Published: 2026-09-17 Updated: 2026-09-28 Summary: Africa's problem is grid reliability, not generation. How uptime engineering, geothermal baseload and on-site power shape Kenya's data centres. A widely shared analysis published on 16 September 2026 put a familiar frustration into sharp words: "Africa does not have an electricity problem. It has a reliability problem" (Engineering News, 16 September 2026). The numbers behind the headline are sobering. About 53 percent of people in sub-Saharan Africa have access to electricity, and more than 560 million still live without it. Kenya sits well above the regional average, with national access rising from roughly 30 percent in 2014 to over 75 percent (World Bank National Energy Compact 2025-2030), but access is not the same as dependability. For most readers that argument is about households and factories. For anyone buying colocation, cloud or connectivity in Kenya, it lands on something much more specific: a data centre does not sell electricity, it sells uptime, and uptime is engineered from the assumption that the grid will fail. Understanding that mindset, and how Kenya's unusually green grid fits into it, explains most of what you will see and hear when you tour a Kenyan facility or compare prices. This article walks through the difference between generation and reliability, what data centres actually build to survive outages, and where Kenya stands. ## Generation is not reliability The two words get used interchangeably in policy debates, and that is exactly where the confusion starts. Generation is how much electricity a country can produce: megawatts from geothermal fields, hydro dams, wind farms and solar arrays. Reliability is whether that electricity arrives at your building, within voltage and frequency limits, every second of every day. A country can add generation capacity and still fail on reliability, because between the power plant and your rack sit hundreds of kilometres of transmission lines, substations, transformers and switching points, and any one of them can trip. ![High-voltage transmission pylons carrying power across Kenyan rangeland](/images/kenya-transmission-pylons-5.webp) The Engineering News argument (16 September 2026) is that this middle layer is where sub-Saharan Africa loses: the constraint is not how much power is generated but how dependably it moves. That matches what infrastructure buyers experience. A factory can tolerate an interruption; machines restart and the shift continues. A data centre cannot, because its product is continuous availability. Every unplanned interruption, however brief, is a breach of the very thing customers pay for, which is why reliability engineering is not a feature of a data centre, it is the product. This distinction also explains why national statistics can mislead buyers in both directions. A country with a high renewable share can still have a fragile transmission backbone, and a country with an imperfect grid can host world-class facilities, because the facility engineers around the grid rather than trusting it. The correct question is never "is Kenya's power good?" It is "what happens in this building when Kenya's power misbehaves?" ## What uptime engineering actually looks like Inside every serious data centre, grid power passes through a gauntlet designed by engineers who assume failure at each stage. Uninterruptible power supply (UPS) systems, usually large banks of batteries, hold the load instantly when input power strays, bridging seconds to minutes. Diesel generators start within that window and carry the load for as long as fuel lasts. Between them sits automatic transfer switching that decides, in milliseconds, which source feeds your rack. ![Electrical switchgear cabinets inside a data centre power room](/images/dc-switchgear-2.webp) Beyond the backup chain, layout matters as much as equipment. Dual power paths (A and B feeds) route electricity to your equipment over completely separate cabling, switchgear and UPS systems, so a single fault cannot take both paths down. Designs are graded N+1 (one spare component per group) or 2N (two fully independent paths, each able to carry the whole load alone), and tier ratings formalise this: our explainer on [what Tier III and Tier IV ratings actually mean](/articles/data-centre-tier-ratings-explained) covers how concurrent maintainability and fault tolerance translate into expected uptime. ![Rows of UPS battery cabinets in a data centre electrical room](/images/dc-ups-batteries-wide.webp) Two practical truths follow from all of this. First, the grid is just the first and cheapest source, not the only one; a well-built facility treats utility supply as an input it can lose. Second, the quality of the engineering shows up in boring details: when the fuel contract was last tested, how often generators run under load bank tests, how quickly the operator can maintain one path without dropping the other. Buyers who ask those questions get better service than buyers who ask about megawatts on a brochure. ## Kenya's grid is greener than it is perfect Kenya starts from a stronger base than most of the region, and the difference is one word: baseload. Roughly 90 percent of the country's electricity comes from renewable sources (US International Trade Administration, 5 July 2024), and the anchor is geothermal, at about 45 percent of generation, running around the clock at capacity factors of 90 to 95 percent (Climate Investment Funds, 5 September 2024; African Leadership Magazine, 29 May 2026). Kenya has around 950 megawatts of geothermal installed in the Rift Valley, anchored by Olkaria, the largest geothermal complex in Africa, plus Lake Turkana Wind's 310 megawatts and seasonal hydro. Geothermal's value for reliability is precisely that it does not care about weather. Hydro sag in dry seasons and solar's nightly disappearance are the classic renewable weak points, but steam fields run through droughts and dark evenings alike, as our guide to [geothermal energy for Kenyan data centres](/articles/geothermal-energy-kenya-data-centres) explains in depth. This is why Nairobi colocation can credibly claim both low-carbon supply and a stable baseload, a combination our analysis of [Kenya's renewables and industrial power](/articles/kenya-renewables-industrial-power-data-centres) argues is a structural siting advantage. Stable baseload, however, is not the same as unlimited capacity. Kenya's record peak demand hit 2,439 megawatts in December 2025 (TechCabal, 20 August 2026), and large new requests now strain against that ceiling. The clearest example is the Microsoft and G42 project near Olkaria, announced in May 2024, whose power request reportedly grew from 60 megawatts to 1,000 megawatts, roughly a third of the country's total installed generation, and the project has stalled partly over that power question (Kenyans.co.ke, 22 August 2026; Business Daily, 10 September 2026). The lesson generalises: generation and transmission planning move at utility pace, while digital demand moves at investment pace, and the gap between those speeds is the reliability problem wearing a business suit. ## Bring your own power: the Amaco lesson The most instructive response to the reliability gap in Kenya right now is being built on the coast. The Amaco HERCULES project in Mombasa, a $1.5 billion AI data centre plan at Dongo Kundu, is designed to generate its own electricity from a floating power barge fed by LNG, rather than waiting for grid allocation (Business Daily, 17 August and 10 September 2026). Reported initial demand at the site runs 75 to 100 megawatts, which would add roughly 4 percent to Kenya's national peak, and the developers chose to sidestep the queue entirely. ![A floating power barge moored at a coastal terminal](/images/amaco-power-barge.webp) That is the extreme version of what every data centre does in miniature. A facility with generators and UPS systems is already a small self-supplying island; Amaco simply scales the island up until it barely needs the mainland. The pattern is worth naming because it will spread: as AI-scale loads arrive in African markets, the projects that move fastest will be the ones that stop treating the grid as a dependency and start treating it as one input among several. Buyers evaluating large deployments in Kenya should expect more "bring your own generation" proposals, and should evaluate them on fuel logistics, conversion efficiency and emissions as much as on headline capacity. None of this makes Kenya's grid a liability to apologise for. It makes the grid a design input, and Kenyan operators who plan that way are following the same logic that made colocation an industry anywhere in the world. The facilities that will win enterprise trust are the ones that can show you their outage history, not just their renewable percentage. ## Questions to ask before you sign When you evaluate a Kenyan facility, power deserves its own hour of the tour. Start with the contracted power SLA: how many kilowatts per rack are guaranteed, and how is downtime compensated? Then walk the path: how many independent power routes reach your hall, and are they truly separate from the utility feed to the last cable? Ask for UPS autonomy at your actual load, generator runtime on site tanks, and the depth of the refuelling contract behind them. Ask when the transfer from UPS to generator was last tested under real load, because untested redundancy is a rumour, not a design. Close with the supply mix and the meter: what share of the facility's supply is geothermal baseload versus seasonal hydro, what PUE (power usage effectiveness) the operator targets, and how it is actually metered. [Our data centre PUE guide](/articles/data-centre-pue-power-usage-effectiveness) explains what typical values mean and how operators improve them. Operators with good answers volunteer these numbers; our [Nairobi colocation buyer's guide](/articles/colocation-data-centres-nairobi-buyers-guide) collects the full question set. An hour of dated, specific questions buys you years of fewer surprises, which is the entire trade a data centre asks you to make. ## Frequently asked questions ### Is Kenya's power supply reliable enough for data centres? For well-engineered facilities, yes. Kenya's structural advantage is geothermal baseload, which supplies roughly 45 percent of generation around the clock (Climate Investment Funds, 5 September 2024), and no credible Kenyan data centre depends on the grid alone anyway: UPS systems, generators and dual paths carry the uptime promise. What a buyer should judge is the facility's tier design, power SLA and outage history, not the national average. ### Why is geothermal power so valuable for data centres? Because it is baseload: geothermal plants run 24 hours a day at capacity factors of 90 to 95 percent, independent of rainfall and daylight (see our geothermal energy guide). Hydro output can sag in dry seasons and solar disappears at night, but geothermal steam fields in the Rift Valley deliver steady power, and Kenya has about 950 megawatts of it installed, anchored by Olkaria, the largest geothermal complex in Africa. ### What does 2N power mean in a data centre? 2N means two complete, independent power paths from the utility or generators all the way to your equipment, each capable of carrying the full load on its own. If one path fails, the other carries on without interruption. N+1, by comparison, means one path with a spare component. 2N costs more to build; it removes single points of failure that N+1 can still contain. ### How long can a data centre run without the grid? Designed properly, indefinitely. UPS batteries carry the load for seconds to minutes while generators start, and generators run for days on tank fuel. What makes it indefinite is fuel logistics: serious operators hold on-site reserves plus contracted refuelling, and they test the full handover under load bank conditions. Ask any facility you pay for how long their autonomy lasts at full load and when it was last tested. ### Why did the Microsoft and G42 data centre project stall? Multiple reasons were reported, but power was central: the project's electricity request grew from 60 megawatts to 1,000 megawatts, roughly a third of Kenya's total installed generation (Kenyans.co.ke, 22 August 2026; Business Daily, 10 September 2026). Kenya's record peak demand is 2,439 megawatts (TechCabal, 20 August 2026), so a request of that scale cannot be absorbed by the grid quickly. It shows why large projects now design their own generation. ### What is SAIDI and should I ask a data centre for it? SAIDI (System Average Interruption Duration Index) is the average number of outage minutes a customer experiences per year. It is a utility metric, so it describes the grid, not any specific facility. Asking a utility's SAIDI is useful context; asking your data centre operator for their own recorded outage history and how their SLA compensates for downtime is far more useful, because their UPS, generators and dual paths stand between you and the grid's statistics. ## Kenya Power Tariffs for Data Centres: Rates & Benchmarks URL: https://data-centers-254.vercel.app/articles/kenya-power-tariffs-data-centres Cluster: Power & Energy Published: 2026-08-28 Updated: 2026-09-28 Summary: Kenya Power's industrial tariffs shape data centre operating costs. The tariff structure, demand charges, negotiation, and regional comparison. Electricity is the single largest operating cost for a data centre. In a well-run facility, power accounts for 30–40% of total operating expenditure, more than staff, more than connectivity, more than insurance, maintenance, and property costs combined. For a 2-megawatt data centre running 24 hours a day, the annual electricity bill can exceed KES 250 million ($1.6 million). Understanding how Kenya Power structures its tariffs, what drives costs up or down, and how data centre operators can negotiate and optimise their power costs is therefore not an optional exercise, it is a core business competency. ![Electrical switchgear and UPS cabinets](/images/dc-ups-batteries-wide.webp) Kenya's power sector is regulated by the Energy and Petroleum Regulatory Authority (EPRA), which approves the tariff structure that Kenya Power applies to different customer categories. Data centres, as large industrial consumers, fall into specific tariff categories that include both energy charges (per kilowatt-hour consumed) and demand charges (per kilovolt-ampere of maximum demand). The interaction between these two charge components, plus various levies and pass-through costs, determines the final bill. ## The Tariff Structure Explained ### Energy Charges The energy charge is the cost per unit of electricity consumed, measured in kilowatt-hours (kWh). For large commercial and industrial customers, Kenya Power applies time-of-use (TOU) pricing, which means the rate varies depending on when the electricity is consumed. TOU pricing reflects the reality that electricity is cheaper to generate during off-peak hours (when demand is low and base-load generators like geothermal plants can meet demand) and more expensive during peak hours (when demand is high and more expensive generation sources may be needed). The typical TOU structure for large commercial customers in Kenya includes three periods. Peak hours (typically 18:00 to 22:00) carry the highest rate, reflecting the evening surge in demand as residential and commercial consumers use electricity simultaneously. Shoulder hours (typically 06:00 to 18:00) carry a mid-range rate. Off-peak hours (typically 22:00 to 06:00) carry the lowest rate, as demand drops and surplus geothermal capacity is available. The specific rates vary by tariff category and are adjusted periodically by EPRA. As of 2025, indicative energy charges for large commercial customers range from approximately KES 10 per kWh during off-peak hours to KES 18 per kWh during peak hours. For a data centre that runs at constant load 24 hours a day (which is the normal operating pattern), the blended average energy charge (weighted across peak, shoulder, and off-peak consumption) is typically KES 13–16 per kWh. ### Demand Charges The demand charge is based on the maximum power demand recorded during the billing period, measured in kilovolt-amperes (kVA). This is different from energy charges in a crucial way: the demand charge reflects the peak power draw, not the total energy consumed. A data centre that draws a steady 1,000 kVA all month pays a lower demand charge (per kVA) than one that normally draws 800 kVA but occasionally spikes to 1,500 kVA, even if both consume the same total energy. ![Engineers inspect a solar installation in Kenya](/images/kenya-solar-farm.webp) Demand charges for large commercial customers typically range from KES 300 to KES 500 per kVA per month. For a data centre with a contracted demand of 2,000 kVA (2 MVA), the monthly demand charge alone is KES 600,000 to KES 1,000,000. For a large facility with 5,000 kVA of contracted demand, the demand charge can exceed KES 2 million per month, a significant cost that must be managed. ### Pass-Through Costs and Levies Beyond the basic energy and demand charges, the electricity bill includes several additional components. The Fuel Energy Cost Charge (FECC) covers the cost of thermal generation fuel. The Inflation Adjustment Charge accounts for inflation in operating costs. The Forex Adjustment Charge reflects the impact of currency fluctuations on equipment and fuel costs (since many components are imported). The Renewable Energy Levy funds the development of renewable energy sources. Each of these is a small per-kWh addition, but collectively they can add 15–25% to the base energy charge. ## Power Usage Effectiveness and the Bill Data centres consume power at two levels: the IT load (the power used by servers, storage, and network equipment to do actual computing work) and the facility load (the power used by cooling systems, lighting, and supporting infrastructure). The ratio of total facility power to IT load is expressed as the Power Usage Effectiveness (PUE). A PUE of 1.0 would mean every watt of facility power goes to IT equipment (impossible in practice). A PUE of 1.5 means for every 1 watt of IT power, 0.5 watts are used for cooling and overhead. Kenyan data centres typically achieve PUE values of 1.4–1.7, depending on the facility's age, cooling technology, and ambient conditions. Nairobi's moderate climate (average temperature 18–25°C, altitude 1,795m) is favourable for cooling, allowing many facilities to use free cooling (outside air) for significant portions of the year, which reduces the energy consumed by mechanical chillers. ![CRAC cooling units behind server racks](/images/dc-cooling-crac.webp) The PUE directly affects the electricity bill. A 1MW IT load at a PUE of 1.5 requires 1.5MW of total facility power, meaning 500kW is consumed by cooling and overhead. At a blended rate of KES 15/kWh, the monthly energy cost for the IT load alone is KES 10.8 million, while the cooling and overhead adds KES 5.4 million, a 50% increase. Improving PUE from 1.5 to 1.3 would save KES 2.16 million per month for a 1MW IT load, KES 26 million per year. How that ratio is measured, what typical Kenyan facilities achieve, and which improvements pay back fastest are covered in [our data centre PUE guide](/articles/data-centre-pue-power-usage-effectiveness). ## Negotiation Strategies Large data centre operators do not simply pay the published tariff. Kenya Power, like most utilities, is willing to negotiate bespoke agreements for large, creditworthy customers who can commit to significant consumption volumes over multi-year periods. The key elements of a data centre power negotiation include the following. ### Committed Volume The most important negotiating lever is committed consumption volume. A data centre operator that can commit to a minimum monthly consumption (say 500,000 kWh or 1,000,000 kWh) provides Kenya Power with predictable revenue, which the utility values for its own planning and procurement. In exchange, the operator can negotiate lower energy rates, reduced demand charges, or both. ### Contract Duration Longer contract durations (5, 10, or even 15 years) provide greater value to Kenya Power and justify better rates. Data centres, unlike most businesses, are inherently long-term, a facility is designed to operate for 20+ years, and the power connection is a fundamental part of the infrastructure. Aligning the power contract duration with the facility's operational life makes sense for both parties. ### Dedicated Feeders and Power Quality ![Kenya Power crews installing power cables along a street trench](/images/kenya-power-feeder-works.webp) Data centres require high-quality power with minimal voltage fluctuations, frequency deviations, or momentary interruptions. Standard Kenya Power distribution feeders serve multiple customers and may experience voltage sags, momentary outages, and frequency variations that are acceptable for residential or commercial customers but problematic for sensitive data centre equipment. As part of a negotiated agreement, a data centre operator can request a dedicated feeder (a power line serving only the data centre) which provides better power quality and isolation from other customers' load patterns. ### Demand Charge Management The demand charge structure can be negotiated in several ways. Operators can request a lower per-kVA demand charge in exchange for a higher committed demand level. They can negotiate "ratcheted" demand charges that average peak demand over multiple months rather than billing based on a single month's peak. And they can include provisions for demand response, agreeing to reduce their load during grid emergencies in exchange for reduced demand charges. ## Regional Cost Comparison Kenya's power costs for data centres are competitive within the African context, though not the lowest on the continent. Here is how Kenya compares to other major African data centre markets. Nigeria's power costs are the highest among major African markets. Grid power costs KES 20–35 per kWh, and because the grid is unreliable, data centres must run diesel generators for significant portions of each day, adding another KES 30–50 per kWh in fuel costs. The effective all-in power cost for a Nigerian data centre can exceed KES 50 per kWh, roughly three times the Kenyan cost. South Africa's power costs are broadly similar to Kenya's at KES 15–22 per kWh, but load shedding (scheduled power outages implemented by Eskom to manage supply shortages) adds significant costs in generator fuel, battery wear, and operational complexity. When these additional costs are included, South African data centre power is 20–40% more expensive than the published tariff would suggest. ![Solar panels rows under a clear sky](/images/kenya-solar-farm-2.webp) Egypt benefits from subsidised natural gas for power generation, resulting in some of the lowest industrial electricity prices in Africa at KES 8–12 per kWh. Ethiopia, with abundant hydroelectric power, offers rates as low as KES 6–10 per kWh, though grid reliability and connectivity limit its appeal for international operators. Kenya's competitive position is strengthened by its renewable generation mix. With over 90% of grid power coming from geothermal, hydro, wind, and solar, Kenya's electricity is both low-cost and low-carbon. As environmental sustainability becomes a more important factor in data centre site selection, this combination of cost and greenness gives Kenya a genuine competitive advantage. ## The Path to Lower Costs ![Officials at a REREC electricity pole handover in rural Kenya](/images/rerec-pole-handover-kenya.webp) For data centre operators in Kenya, the path to lower power costs runs through three channels: negotiation with Kenya Power for better rates and terms, operational efficiency improvements (lower PUE, better demand management), and on-site generation (particularly solar PV as discussed in [our solar power guide](/articles/solar-power-data-centres-kenya)). The operators who combine all three (negotiate well, operate efficiently, and generate their own clean power) will achieve the lowest all-in power costs and the strongest competitive position in Kenya's growing data centre market. Power costs are not going to decrease. As Kenya's economy grows and electrification expands, demand for grid power will increase, and tariffs will adjust accordingly. Data centre operators who invest now in efficiency, negotiation capability, and on-site generation will be the ones who maintain their cost advantage as the market matures and competition intensifies. ## Kenya's Renewable Energy Meets Data Centre Demand URL: https://data-centers-254.vercel.app/articles/kenya-renewables-industrial-power-data-centres Cluster: Market Analysis Published: 2026-08-31 Updated: 2026-09-28 Summary: Renewables overtake coal globally in 2026 as data centres drive demand. Kenya enters that moment with a 90%-renewable grid, a green data centre edge. A quiet milestone is arriving in global electricity. The International Energy Agency expects renewable generation to overtake coal as the world's largest source of electricity in 2026, after reaching near parity in 2025, with renewables forecast to grow more than eight percent this year and lift their share of global generation from 33 percent in 2025 to 37 percent by 2027. The same IEA outlook expects global electricity demand to grow 3.6 percent in 2026, and it names the drivers explicitly: industry, electric vehicles, and expanding data centres. For the first time, the world's most talked-about new load (the data hall) is growing into an electricity system that is itself going green. Kenya enters that moment with an unusual hand. More than 90 percent of the country's electricity already comes from renewable sources (geothermal, hydro, wind, and solar) and the government targets a fully renewable system by 2030. A commentary published in The Standard on 31 August 2026 argues the country must now convert that clean-power advantage into factories, exports, and jobs before someone else captures the value. For Kenya's data centre community, the argument lands squarely on home turf, because data centres are simultaneously named among the demand drivers of this new era and positioned to be among its most discriminating buyers of green power. This article unpacks what the global crossover means for siting decisions, audits the opinion column with appropriate editorial distance, maps the industrial pipeline now forming around Kenya's renewable zones, and asks the question our readers actually care about: where do data centres fit in Kenya's green industrial push? ## The Global Crossover: Clean Power Meets Compute Demand The IEA's numbers describe a collision of two curves that data centre operators should track closely. On one side, renewable generation keeps compounding at high single-digit rates and is set to become the world's dominant source of electricity within the year. On the other, electricity demand is accelerating after a long plateau, with data centres contributing a visible share of new load for the first time in the agency's demand outlooks alongside industry electrification and EV charging. The intersection of those curves changes how sites get chosen. In a coal-heavy grid, a hyperscaler that wants clean operations must assemble its own solution (renewable PPAs, on-site solar, batteries, and certificate accounting) and each layer adds cost, complexity, and reputational risk if the accounting is challenged. In a grid that is already over 90 percent renewable, the carbon story is structural rather than assembled. A colocation operator in Nairobi can point to the national generation mix and credibly market low-carbon capacity without buying a single extra certificate, an advantage our [geothermal energy explainer](/articles/geothermal-energy-kenya-data-centres) examines in detail. That advantage compounds as AI workloads grow. Training clusters draw sustained, gigawatt-scale power, and every megawatt-hour consumed carries a carbon number that someone upstream will eventually scrutinise. Markets that can supply round-the-clock clean electrons (which Kenya can, because geothermal provides baseload rather than daylight-only generation) are making a claim that solar-heavy markets cannot yet match without storage. Kenya's grid, long framed as a reliability liability, is quietly becoming a siting asset for exactly the workloads now driving global demand. ## Kenya's Renewable Foundation, in Numbers The foundation is real. Geothermal supplies roughly 45 percent of Kenya's electricity generation, with around 950 megawatts installed across the Rift Valley, anchored by the Olkaria complex in Naivasha, the largest geothermal installation in Africa. Lake Turkana Wind Power adds East Africa's largest wind farm at 310 megawatts, expanding solar installations keep scaling across the arid belts, and hydro continues to provide seasonal flexibility. Together they push the renewable share of generation above 90 percent, among the highest of any significant economy in the world. ![Wind turbines turning across a Kenyan landscape](/images/kenya-wind-turbines-wide.webp) Two properties of that mix matter for digital infrastructure. First, geothermal's baseload character means the grid's cleanest source is also its most dependable, which is why our analysis of [Kenya Power tariffs for data centres](/articles/kenya-power-tariffs-data-centres) keeps returning to generation economics as the sector's structural advantage. Second, the resource is nowhere near exhausted, KenGen and the Geothermal Development Company continue to prove additional steam fields, meaning Kenya can grow its clean supply as demand arrives rather than choosing between growth and decarbonisation. ## An Opinion Column, Read Carefully The immediate trigger for this analysis is a commentary in The Standard titled "Kenya must turn renewable energy advantage into industrial power", authored by a global communications strategist and former business and technology editor. Readers should know what that piece is: advocacy, written persuasively, with a clear pro-partnership framing, Dubai features as the story's hero, and the column reads like it was placed to support a commercial narrative. That does not make it wrong, but it does mean its deal values deserve the same scrutiny we apply to data centre announcements on this site. Separating the persuasive from the verifiable, the verifiable scaffolding is solid. The IEA figures check out against the agency's published outlooks. The named projects are real and independently reportable: the Sleeping Warrior Special Economic Zone at Elementaita is being developed around renewable power; Dubai-based Aquilastor announced a 150-million-dollar electric and hybrid vehicle assembly plant at the Olkaria Green Energy Park targeting 50,000 vehicles a year and more than 3,000 direct jobs; DP World has agreed with GulfCap Africa to develop a 222-hectare Mombasa Industrial Park near the port; and at COP28 Kenya announced an 800-million-dollar partnership with AMEA Power for a proposed 200-megawatt development at Paka, alongside the UAE's 4.5-billion-dollar Africa green energy package. The Kenya Green Investment Fund, seeded with 40 million dollars through the Kenya Development Corporation and seeking another 160 million from institutional investors, is genuine policy architecture. What remains unproven is the thesis itself, that these threads will weave into industrial transformation, and that foreign partnerships will deliver local value rather than extracted returns. That is precisely the test our market applies to data centre announcements, so the column's central question is one we are well equipped to track. ## The Industrial Pipeline Taking Shape Around Kenya's Green Power ![High-voltage transmission pylons crossing Kenyan terrain](/images/kenya-transmission-pylons-2.webp) Zoom out and a pattern emerges: Kenya's green electrons are starting to attract industrial tenants, not just residential meters. The Sleeping Warrior zone near Elementaita is the clearest example, a special economic zone designed around renewable power from the ground up, courting tenants from agro-processing to clean manufacturing. The Aquilastor plant at Olkaria goes further, siting manufacturing directly inside a geothermal park, essentially co-locating a factory with its power plant. The Mombasa Industrial Park pairs cheap coastal land with port logistics and, though the column does not say it, with the submarine cables that land on the coast, a detail that matters enormously for any digital tenant. For data centres, this pipeline is both opportunity and competition. Opportunity, because green industrial parks normalise the idea that power-hungry facilities should locate where clean energy is cheapest, and because each successful tenant proves the delivery model (land, power agreements, incentives, workforce) that a data centre development would follow. Competition, because every megawatt committed to an EV assembly line or a processing plant is a megawatt the grid does not have headroom to hand to a compute campus, and because industrial policy prioritises export jobs when allocating scarce infrastructure. Efficiency sharpens that competition: a facility that improves its [data centre PUE](/articles/data-centre-pue-power-usage-effectiveness) converts more of its allocation into billable compute instead of overhead. The 200-megawatt scale of the AMEA Power project is instructive: it is roughly the size of a mid-sized data centre campus, and exactly the kind of generation addition that compute-hungry projects wait for. ## Where Data Centres Fit in Kenya's Green Industrial Push The honest answer is that data centres fit in three places, each with a different value proposition. The first is Nairobi, where the existing cluster, the operators catalogued in [our facility directory](/directory), sits on dense fibre, the KIXP peering point, and enterprise demand. The green-grid story strengthens Nairobi's hand in regional comparisons, giving Kenyan operators a marketing claim their counterparts in diesel-dependent markets cannot match, and it aligns neatly with the national ambition our [market outlook](/articles/kenya-data-centre-market-outlook-2025-2030) tracks. The second is field-adjacent compute. As AI training decouples workloads from cities, the economics of building near Olkaria or Menengai (buying power near generation cost, dropping transmission exposure, accepting a latency hit) keep improving. The vehicle plant at the Olkaria Green Energy Park proves the concept; a training-focused AI data centre is the same trade with different tenants. The obstacle is not power but connectivity and operations, which is why field-adjacent projects will most likely arrive as purpose-built AI campuses rather than general colocation. The third is Mombasa, where the new industrial park adds a logistics anchor to the coast's existing advantage: the submarine cable landing stations that carry most of East Africa's international bandwidth. A coastal green industrial zone with subsea connectivity is a credible edge-computing and content-delivery location, and if the park's masterplan reserves space for digital infrastructure, Mombasa's long second-fiddle role in [Kenya's data centre geography](/articles/why-data-centres-cluster-nairobi-mombasa) could begin to change. Placing data centres inside special economic zones would also raise interesting tax questions our [incentives analysis](/articles/tax-incentives-data-centre-investment-kenya) will be watching. ![Rows of photovoltaic panels at a Kenyan solar farm](/images/kenya-solar-farm-2.webp) ## The Value-Capture Test, Applied to Data Centres The opinion column's sharpest line is its warning that Kenya "cannot afford to remain an exporter of raw potential while others capture the investment, technology and industrial value generated from it". Strip away the Dubai framing and that warning applies verbatim to data centres, because a data hall can be just another form of exporting raw potential: electrons out, compute services in, with the high-value engineering done elsewhere. Whether Kenya's green power produces industrial value or simply hosts imported value is a policy choice, not a market inevitability. The practical test is jobs and skills per megawatt. A hyperscale campus can be built with a handful of long-term staff and flown-in specialists, or it can be anchored in local operations, security, mechanical and electrical trades, and a talent pipeline that feeds the region, the difference our careers coverage documents facility by facility. It shows up in procurement too: whether local contractors win the fit-out and maintenance scopes, whether training programmes attach to project approvals, and whether operators publish the energy and employment data that lets the public hold them to their promises. Kenya's emerging green industrial policy gives the government leverage to ask for those things at approval stage; data centre developers should expect the questions, and the serious ones will welcome them. There is also a defensive reason to get this right. If Kenya positions data centres as part of its industrial future (rather than as mere power consumers) the sector earns a seat in the conversations that allocate grid headroom, SEZ slots, and incentives. If it does not, compute projects will find themselves queued behind factories, and rightly so in political terms. ## What to Watch Next Four markers will tell us whether this narrative converts. Watch Kenya Power and the regulator for commercial mechanisms (green tariffs or corporate PPA frameworks) that would let large digital buyers contract renewable supply directly; the grid's carbon mix is already an asset, but contracting structures decide who can monetise it. Watch tenant announcements at Sleeping Warrior and the Mombasa Industrial Park, and whether any digital-infrastructure name appears among them. Watch execution on the Olkaria Green Energy Park, where the vehicle plant will demonstrate whether power-park industrialisation actually delivers in Kenya. And watch whether the Microsoft and G42 flagship project re-emerges with a green-power story attached, because a one-gigawatt commitment wrapped in the world's newest renewable-crossover narrative would recalibrate the entire region's positioning, the scenario our [Kenya digital hub analysis](/articles/kenya-east-africa-digital-hub) laid out. The global context is now formally favourable: the world's electricity system is going green in the same decade its demand growth went digital. Kenya holds an early, real advantage at that exact intersection. Turning it into energised data halls, skilled jobs, and retained value is the same test the opinion column sets for factories, and it is the test this site exists to score honestly. ## Keppel's US$1.2B Tokyo Deal: Signal for African Capital URL: https://data-centers-254.vercel.app/articles/keppel-dc-reit-tokyo-acquisition-signals Cluster: Market Analysis Published: 2026-09-04 Updated: 2026-09-04 Summary: Keppel DC REIT is paying 190 billion yen for two Tokyo hyperscale facilities. The deal metrics are the benchmarks African assets must meet. While most of East Africa's data centre conversation is about construction (who is building, where, and when the grid can keep up) the money side of the industry just posted a masterclass. Keppel DC REIT and its sponsor Keppel announced on 1 September 2026 that they will collectively acquire nearly all of two Tokyo data centres for 190 billion yen, roughly US$1.2 billion, in one of the year's defining statements about what institutional capital will pay for mature digital infrastructure. The facilities are Tokyo Data Centre 4 and Tokyo Data Centre 5 in Inzai City, Greater Tokyo: freehold, hyperscale, fully-fitted co-location buildings in one of the world's deepest data centre markets. Nothing about the transaction touches Africa directly. Everything about its structure should. The deal is a public, audited catalogue of the metrics that global capital underwrites at scale, and the distance between those metrics and what most African facilities can evidence today is, in one number each, exactly why the continent's data centre story remains a development story rather than a REIT story. ## The Deal Mechanics, Briefly The structure is precise. Keppel DC REIT and Keppel, through its stake in Keppel Japan KK, will jointly hold 90 percent of the two facilities, with the Reit taking an 88.62 percent effective interest in each and the current operator (described only as an established global data centre owner) retaining the remaining 10 percent. The aggregate price of 190 billion yen lands about 2.1 percent below the assets' combined valuation of 194 billion yen, with the Reit's share of the payment totalling roughly 168.4 billion yen. Completion is targeted for the fourth quarter of 2026. Funding follows the classic infrastructure playbook: match the equity raise to the asset's income currency. The Reit's manager launched a private placement targeting at least S$600 million, 280.1 million new units offered at S$2.096 to S$2.142, a 2.5 to 4.6 percent discount to the pre-announcement volume-weighted average price, and demand proved strong enough to upsize it within a day to S$625 million, fixing the issue price at S$2.10 (about 4.5 percent below the S$2.20 close before the announcement) for 297.6 million new units that begin trading on 10 September. Alongside the placement sits an advanced distribution of S$0.02241 to S$0.02281 per unit for the period between 10 July and 9 September. The balance comes from yen-denominated debt and internal sources, so the facility's revenues and the debt servicing its purchase sit in the same currency. African developers hammering out project finance would do well to internalise that discipline: currency-mismatched data centre debt has quietly killed more regional business cases than land or fibre ever have. ![Exterior view of a large-scale data centre campus](/images/worlds-largest-data-centers-2026.webp) ## The Metrics That Made It Accretive The reason the acquisition is distributable-income accretive rather than dilutive is a short list of contractual features, and it doubles as the cleanest checklist African operators will find anywhere for what "institutional grade" actually means. Pro forma, had the deal closed on 1 January 2025, full-year distribution per unit for FY2025 would have risen 2.6 percent, from S$0.10381 to S$0.10649. Underneath that: contracted average annual rent escalation of about 2.8 percent; in-place rents estimated to be at least 30 percent below prevailing market rates, embedding reversion upside at renewal; weighted average lease expiry of 4.5 years for Tokyo Data Centre 4 and 10.6 years for Tokyo Data Centre 5; and both facilities fully occupied by four investment-grade internet enterprise and IT services clients, three of them new to the portfolio. The manager also highlighted portfolio effects, top-client concentration falling from 43.5 percent to about 38.2 percent of rental income, and contracted power capacity edging up from 95 to 96 percent. ![Fully fitted server racks in an operational data hall](/images/hero-server-hall.webp) Each line item answers a specific investor fear. Escalators answer inflation. Below-market in-place rents answer valuation risk, you are not buying the top of the market. Long WALE answers vacancy risk. Investment-grade covenants answer counterparty risk. And contracted power capacity answers the question that has become the industry's universal constraint: does the asset control the electrons its revenue depends on? Kenya's operators will recognise the last one immediately, because grid access is precisely where [Kenya's power infrastructure](/articles/kenya-power-infrastructure-data-centres) shapes local deal terms. Strategically, the acquisition lifts the Reit's assets under management from S$6.3 billion to approximately S$7.6 billion across 27 data centres in 10 countries, and raises Japan's share of portfolio rental income from roughly 9 percent to about 23 percent, with Singapore anchoring at around 60 percent. Even a S$7.6 billion vehicle manages geographic concentration deliberately, a discipline that reads as caution to Tokyo and as aspiration to African markets still waiting on their first REIT-grade portfolio. ## Why a Tokyo Deal Matters in Nairobi The connection to Kenya is not rhetorical. Global capital is finite, fungible, and comparative: the same pension, infrastructure, and sovereign funds evaluating Tokyo assets are the LPs and co-investors African developers pitch. A US$1.2 billion cheque for fully-leased, freehold hyperscale stock in Greater Tokyo sets the opportunity cost against which any Nairobi project competes. When a Nairobi developer promises a 15 percent development yield, the investor's internal question is why they should take construction and lease-up risk in a frontier market when a 2.8-percent-escalator, investment-grade-leased Japanese asset is available at a 2.1 percent discount to valuation. The answer, of course, is growth, African data centre demand is compounding faster than Tokyo's, and our [Kenya market outlook](/articles/kenya-data-centre-market-outlook-2025-2030) makes that case in detail. But growth only converts into capital when it arrives wrapped in the contractual features this deal displays. That is the practical translation for Kenyan founders and facility owners: the valuation conversation is won in the lease, not in the pitch deck. Lease length, tenant covenant quality, escalators, power contracts, and concentration limits are the difference between pricing an asset as infrastructure and pricing it as a bet. ![Nairobi's skyline at dusk](/images/nairobi-skyline.webp) There is a local-capital angle too. Kenya has no listed data centre REIT, and Kenya's capital markets have yet to see a securitised digital infrastructure vehicle of the kind Singapore, Japan, and increasingly Europe take for granted. The [tax incentives](/articles/tax-incentives-data-centre-investment-kenya) available to Kenyan data centre investors address the build phase; a REIT framework would address the exit phase, giving today's private backers a route to recycle capital into the next facility. Until that market deepens, the Keppels of the world will keep finding African exposure through private development capital rather than portfolio acquisitions, because there is not yet an institutional-grade portfolio to buy. ## The Playbook, distilled Strip the deal to its transferable lessons and four stand out for African operators, investors, and anyone building a data centre business on the continent, points we also stress in our [due diligence checklist](/articles/data-centre-due-diligence-kenya) and [market entry guide](/articles/starting-data-centre-business-kenya). First, contract for the long sale from day one. Every feature that made the Tokyo assets institutional-grade (escalators, WALE, covenants) was written into leases years before any acquisition was contemplated. Operators who sign three-year leases at flat rents with single-tenant concentration are not building an asset; they are renting out a building. Second, treat power as a contracted revenue input, not an operating detail: 96 percent contracted capacity is a headline investors read. Third, match funding currency to income currency, the yen-debt structure is not cleverness, it is survival. Fourth, concentration is a managed metric, not an accident: the Reit celebrated reducing top-client dependence from 43.5 to 38.2 percent, and African facilities planning hyperscale anchor tenants should model the same drift and its exit-day consequences. None of this requires a Singapore listing to be useful. It requires founders who understand that the eventual buyer (whether a REIT, a pension fund, or a strategic operator) will underwrite the leases, not the ambition. Tokyo just published the syllabus. ## KIXP: How Kenya's Internet Exchange Keeps Traffic Local URL: https://data-centers-254.vercel.app/articles/kixp-internet-exchange-point-kenya Cluster: Connectivity Published: 2026-08-26 Updated: 2026-10-06 Summary: KIXP is the Kenya Internet Exchange Point. It keeps local internet traffic inside Kenya, reducing latency and costs. Here is how it works and why it matters. ![Submarine cable coming ashore at a Kenyan beach landing](/images/mombasa-cable-landing.webp) Every time a Safaricom subscriber in Nairobi opens the Kenya Revenue Authority iTax portal, and every time an Airtel user in Mombasa streams a video hosted on Google's local cache, that traffic flows through KIXP, the Kenya Internet Exchange Point. KIXP is the single most important piece of internet infrastructure that most Kenyans have never heard of, and it fundamentally shapes how fast, how cheap, and how reliable the internet feels in Kenya. KIXP is an Internet Exchange Point (IXP): a physical facility where different networks meet to swap traffic directly. Instead of every local data request detouring through an undersea cable to Europe and back, KIXP keeps that traffic inside Kenya. The result is lower latency, lower costs, and a more resilient internet for the entire country. ## How Does KIXP Actually Work? At its core, KIXP is a high-speed Ethernet switch (technically, a set of redundant switching platforms) housed in [carrier-neutral data centres](/articles/what-is-a-data-centre) in Nairobi. Each member network runs a physical fibre connection from their own equipment to the KIXP switch. Using the Border Gateway Protocol (BGP), members advertise which IP addresses they can reach. When one member wants to send data to an IP address belonging to another member, the traffic goes straight across the switch rather than traveling through an upstream transit provider. The mechanics are straightforward, but the economics are profound. Without KIXP, a Safaricom customer accessing a website hosted on Liquid Intelligent Technologies would send traffic from Safaricom's network, through a transit provider, often routed via the [submarine cables landing in Mombasa](/articles/submarine-cables-landing-mombasa) to a point of presence in Europe or the Middle East, and then back down to Liquid's network in Nairobi. That round trip adds 150 to 250 milliseconds of latency and consumes expensive international bandwidth on both ends. With KIXP, the same traffic crosses a single Ethernet switch in Nairobi. Latency drops to under 5 milliseconds. No international bandwidth is consumed. Both networks save money, and the user gets a faster experience. ![Map of submarine cables landing on the East African coast](/images/submarine-cables-map.webp) ## The History of KIXP KIXP was established in 2000 by the [Telecommunications Service Providers Association of Kenya (TESPOK)](https://www.tespok.co.ke/), making it one of the first IXPs in sub-Saharan Africa. At the time, Kenya's internet was served almost entirely through satellite links. International bandwidth cost upwards of USD 7,000 per megabit per second per month, a price that made any meaningful local internet ecosystem impossible. The early years were difficult. KIXP's founding members had to overcome regulatory resistance from the then-monopoly operator, Kenya Posts and Telecommunications Corporation (KPTC), and later Telkom Kenya. There was no legal framework for internet peering, and the dominant incumbent had little incentive to share traffic locally when it was earning revenue carrying that same traffic internationally. The breakthrough came with the liberalisation of Kenya's telecommunications sector in the mid-2000s. The Communications Commission of Kenya (now the Communications Authority of Kenya) issued new licensing frameworks that encouraged competition and infrastructure sharing. When TEAMS and SEACOM submarine cables landed in Mombasa in 2009–2010, international bandwidth prices collapsed. But paradoxically, this made KIXP *more* important, not less: cheaper international bandwidth meant more Kenyans came online, which meant more local content and services, which meant more traffic that benefited from local peering. By 2012, KIXP was handling over 1 Gbps of peak traffic, and the Internet Society measured 19 Gbps by 2020, with the exchange's cost savings to members quadrupling over the previous decade. The curve then steepened dramatically: TESPOK announced a historic peak of 1.3 Tbps in June 2024, and by October 2026 PeeringDB listed KIXP Nairobi with 144 peering networks and roughly 2,985 Gbps of connected port capacity. Each local cache a network installs compounds the effect, because cached content pulls more traffic onto the exchange. ## Who Connects to KIXP? KIXP's membership has expanded far beyond traditional ISPs. The current member roster reflects the maturity of Kenya's internet ecosystem: - **Mobile network operators**: Safaricom, Airtel Kenya, and Telkom Kenya all peer at KIXP. This means that a Safaricom user calling an Airtel number via WhatsApp stays entirely local. - **Fixed and fibre ISPs**: Liquid Intelligent Technologies, Wananchi Group (Zuku), Poeple (PoA Internet!), AccessKenya, and dozens of smaller regional ISPs. - **Content providers and CDNs**: Google, Netflix, Meta (Facebook and Instagram), and Cloudflare all maintain a presence at KIXP or connect through their local caching infrastructure. This is why YouTube and Netflix stream in high definition without buffering for most Kenyan users. - **Government and education**: The Kenya Education Network (KENET), which connects universities and research institutions across the country, peers at KIXP. Government services hosted by the ICT Authority also exchange traffic locally. - **Cloud and hosting providers**: Local and regional cloud providers connect at KIXP so that their customers' workloads communicate efficiently with other Kenyan networks. ![Blue-lit server racks in a data hall](/images/hero-server-hall.webp) ## Peering vs Transit: What Is the Difference? Understanding KIXP requires understanding the distinction between peering and transit, because these two arrangements determine how most internet traffic flows, and how much it costs. | Characteristic | Peering (at KIXP) | Transit (upstream provider) | |---|---|---| | **Cost** | Typically free (settlement-free) | Paid per Mbps or Gbps | | **Scope** | Only reaches other peers at the same IX | Reaches the full global internet | | **Traffic flow** | Direct, often single switch hop | Routed through one or more intermediate networks | | **Latency** | Very low (1–5 ms within Nairobi) | Higher (depends on transit provider's path) | | **Who uses it** | Networks exchanging local traffic | Networks needing global reach | | **Example** | Safaricom ↔ Google cache at KIXP | Small ISP → Liquid → London → destination | Most Kenyan networks use a combination. They peer at KIXP for local traffic (which, thanks to content caching, can represent 60–80% of total volume) and buy transit from an upstream provider for everything else. The more traffic a network can peer locally, the less transit it needs to purchase, and the lower its operating costs. ## Why KIXP Matters for Internet Costs in Kenya Kenya has some of the most affordable mobile data prices in Africa. According to the Alliance for Affordable Internet (A4AI), 1 GB of mobile data in Kenya costs roughly 2–3% of monthly income at the national poverty line, well below the A4AI's affordability threshold of 2% for low-income groups (as measured at the median income level). KIXP is not the only reason, but it is a significant contributing factor. Consider the arithmetic. If Safaricom handles 500 Gbps of total traffic during peak evening hours, and 70% of that traffic can be peered locally at KIXP (350 Gbps), then Safaricom only needs to purchase transit capacity for the remaining 150 Gbps. Without KIXP, that full 500 Gbps would need international transit. At current wholesale prices, the difference is tens of thousands of dollars per month, savings that directly affect consumer pricing. KIXP also reduces the load on Kenya's international links. This is important for national resilience. The [submarine cables landing in Mombasa](/articles/submarine-cables-landing-mombasa) have finite capacity. During cable cuts (which happen several times a year) KIXP-local traffic is completely unaffected. Kenyans can still access local banking, government services, and locally cached content even when the country's connection to the global internet is degraded. ![Aerial view of the Port of Mombasa and its container terminal](/images/mombasa-port-wide.webp) ## KIXP's Technical Infrastructure KIXP operates multiple switching platforms across carrier-neutral data centre facilities in Nairobi to ensure redundancy, and in December 2024 it announced a peering node partnership with iXAfrica Data Centers, extending the exchange into one of the facilities where cloud and AI workloads are now hosted. The exchange uses a Layer 2 Ethernet fabric, meaning it operates at the data link layer. Members establish BGP sessions (typically over 10 Gigabit or 100 Gigabit Ethernet ports) to exchange routing information. The technical specifications matter for reliability. KIXP's switching infrastructure is designed to deliver: - **Redundancy**: Dual switches in separate data centres ensure that a single hardware failure does not disconnect members. Each member is encouraged to connect to both sites. - **Low latency**: The entire switching fabric adds less than 1 millisecond of delay. Traffic between two members in the same data centre facility experiences sub-millisecond latency. - **Scalability**: The current platform supports 100 Gigabit Ethernet ports, with the ability to upgrade to 400 Gigabit as demand grows. - **Neutrality**: KIXP is operated by TESPOK on a non-profit basis. No member receives preferential treatment. The exchange does not sell transit, it only facilitates peering. The physical locations of KIXP's switching equipment are within commercial data centres in Nairobi. You can find these and similar facilities in the [Kenya data centre directory](/directory). Members collocate their own routers and servers in these same facilities, or they run dedicated fibre from their own premises to the KIXP point of presence. ![Chart of common network protocols including HTTP, DNS, SSH and TCP](/images/diagram-network-protocols.webp) ## The Role of Content Caching at KIXP One of the most impactful developments at KIXP in the last decade has been the arrival of large content providers. Google installed a Global Cache node accessible via KIXP in the early 2010s. Netflix followed with its own Open Connect appliance. Meta, Cloudflare, and Akamai have all established local presence. Content caching works by storing popular content (YouTube videos, Netflix shows, Facebook images, web assets) on servers physically located in or near KIXP. When a Kenyan user requests that content, it is served from the local cache rather than from a data centre in Europe or the US. Because the cache connects to KIXP, every member network can reach it with minimal latency and zero international bandwidth cost. The effect on traffic patterns has been dramatic. Before content caching, the vast majority of traffic on Kenyan networks was outbound, Kenyan users pulling content from overseas. Today, thanks to caching and the growth of local digital services like M-Pesa, KENET, and government portals, a substantial and growing fraction of traffic is genuinely local in both directions. ## How Does KIXP Compare to Other African IXPs? KIXP is frequently cited as a model for internet exchange development in Africa. According to the [Internet Society](https://www.internetsociety.org/resources/deploying-internet-exchange-points/), Kenya was among the first countries on the continent to establish a functional IXP, and its experience has informed IXP deployment strategies across the region. South Africa's NAPAfrica, based in Johannesburg and Cape Town, handles several times KIXP's volume due to the country's larger economy and more developed hosting sector, though KIXP's 1.3 Tbps peak now puts it in the same terabit class as the biggest African exchanges. Nigeria's IXPN, based in Lagos and Abuja, has grown rapidly on similar dynamics. But on a per-capita basis, Kenya's peering ecosystem punches above its weight. The combination of high mobile penetration, strong local content growth, and a competitive ISP market has driven KIXP's traffic to levels that countries with larger populations have struggled to match. Rwanda's RINEX, Tanzania's TIX, and Uganda's UIXP have all benefited from lessons learned at KIXP. Regional interconnection between East African IXPs is an ongoing effort, with the goal of keeping intra-African traffic on the continent rather than routing through Europe. ## The Policy and Regulatory Environment Kenya's regulatory framework has been broadly supportive of internet peering, though the relationship has not always been smooth. The Communications Authority of Kenya (CA) has not mandated peering, KIXP operates on a voluntary, bilateral peering model. Some policymakers have argued that mandatory peering would improve service quality, while the industry has generally preferred the current voluntary approach, arguing that it encourages investment and avoids the complexity of regulatory intervention. Taxation of digital services is a related policy issue that indirectly affects KIXP. Kenya's Digital Services Tax, introduced in 2021, imposed a levy on income from services delivered over the internet. While the tax does not directly target peering, it affects the economics of content providers operating in Kenya and, by extension, the incentives for maintaining local caching infrastructure. The Kenya Information and Communications (Amendment) Act, 2013 also has implications for internet infrastructure. The Act provides a framework for the sharing of telecommunications infrastructure, including ducts, towers, and fibre routes. This infrastructure sharing has lowered the cost for new ISPs to connect to KIXP by reducing the capital expenditure needed to build or lease last-mile fibre to the exchange point. ## What Are the Challenges Facing KIXP? Despite its success, KIXP faces several challenges: **Limited regional interconnection**. Traffic between Kenya and its neighbours (Uganda, Tanzania, Rwanda) often still routes through Europe. While there are efforts to establish cross-border peering, regulatory differences, commercial disputes, and the lack of direct fibre links between national IXPs remain obstacles. **Concentration in Nairobi**. KIXP's infrastructure is concentrated in Nairobi. ISPs serving Mombasa, Kisumu, or coastal regions must backhaul traffic to Nairobi to peer, adding cost and latency. There have been discussions about establishing remote peering points or a secondary KIXP node in Mombasa, but no concrete plans have been implemented. **Power costs**. Kenya's electricity tariffs, while lower than many African countries, have been rising. Data centres (including those hosting KIXP) are significant power consumers, and these costs are ultimately borne by members. **Security concerns**. Like any critical internet infrastructure, KIXP is a potential target for cyberattacks. While the exchange itself is a Layer 2 facility and does not inspect application-layer traffic, BGP hijacking and route leaks from misconfigured members are ongoing risks that require constant monitoring. ## Frequently Asked Questions **What is an Internet Exchange Point?** An IX is a facility where different networks connect directly to exchange traffic. Without an IX, traffic between two networks in the same city might route through Europe or South Africa. With an IX, it stays local, faster and cheaper. **Who can connect to KIXP?** KIXP is open to any organisation that operates its own autonomous system (AS). This includes ISPs, mobile operators, content providers like Google and Netflix, cloud platforms, government networks, and educational institutions. **How much traffic does KIXP handle?** KIXP's peak traffic reached a historic 1.3 Tbps in June 2024 (TESPOK), and by October 2026 PeeringDB listed 144 peering networks with roughly 2,985 Gbps of connected port capacity, making KIXP one of the largest internet exchange points in East Africa. Growth has accelerated as content providers, cloud platforms and now AI workloads establish local presence. **Does KIXP reduce internet costs for users?** Indirectly, yes. By keeping traffic local, ISPs avoid paying for international bandwidth on every local connection. This cost saving can be passed on to consumers. KIXP has been cited as a factor in Kenya's relatively affordable internet prices. **What is peering vs transit?** Peering is a direct, typically free, exchange of traffic between two networks at an IX. Transit is when you pay another network to carry your traffic to destinations you cannot reach directly. Networks use a mix of both. ## Why KIXP Will Become Even More Important Kenya's internet user base continues to grow. The Communications Authority of Kenya counted 84.1 million active mobile subscriptions and 62.6 million mobile broadband subscriptions in the third quarter of the 2025/26 financial year, with mobile broadband uptake still climbing quarter on quarter. The government's digital transformation agenda (which includes moving services online, expanding e-commerce, and developing smart city infrastructure) will generate enormous volumes of local traffic. Cloud computing adoption among Kenyan enterprises is accelerating. As more businesses migrate workloads to cloud platforms with local points of presence, the traffic between those platforms and end users will flow through KIXP. The same applies to edge computing, 5G networks, and the Internet of Things, all of which generate traffic that benefits from local peering. KIXP is not a glamorous technology. It does not make headlines the way a new submarine cable or a data centre grand opening does. But it is the quiet infrastructure that makes everything else work efficiently. Without KIXP, Kenya's internet would be slower, more expensive, and more fragile. With it, the country has built one of the most effective peering ecosystems on the African continent, and that foundation will support the next generation of Kenya's digital economy. ## Konza Data Centre Expansion: PPP Preparation Funding URL: https://data-centers-254.vercel.app/articles/konza-ppp-data-centre-funding Cluster: Kenya Market Published: 2026-10-24 Updated: 2026-10-24 Summary: Kenya's PPP fund will pay for advisory and feasibility work on Konza's data centre expansion. What the money buys, the tender trail behind it, and what comes next. ## What Was Signed Kenya's National Treasury, through the Public-Private Partnerships Project Facilitation Fund, has signed a funding agreement with the Konza Technopolis Development Authority (KoTDA) to move a data centre expansion into active project preparation. The agreement, reported on 23 October 2026, was signed by KoTDA chief executive John Paul Okwiri. It covers the project known in Kenya's PPP pipeline as the Konza Cloud Expansion and Smart City Facilities. The money does not build anything. It pays for the work that comes before a shovel moves: transaction advisory services, feasibility assessments, and the development of an investment and implementation framework. The PPP Directorate's own announcement describes the funding as support for project preparation, including the engagement of a transaction advisor to structure the project for market engagement. ![Group photo of KoTDA and National Treasury officials at the signing](/images/konza-ppp-signing-group.webp) That distinction matters more than it may look. Data centre announcements in this region routinely blur the line between a signed construction contract and an early preparation step. This is firmly the second kind, and reading it correctly is the difference between a market signal and wishful thinking. ## What Preparation Funding Actually Buys Kenya's PPP framework moves a project through defined stages: identification, preparation, procurement, and finally financial close followed by construction. The Project Facilitation Fund exists because the early stages cost real money before any investor is attached. Advisory fees, feasibility studies and legal structuring all fall due while the project still belongs entirely to the state. For the Konza expansion, the agreement covers three things: - **Transaction advisory**: engaging a qualified advisor to design the commercial structure, draft tender documents, and run the procurement that brings private bidders to the table - **Feasibility assessment**: testing whether the expansion makes sense technically and financially, at what scale, and on what commercial terms - **Investment and implementation framework**: the document set that tells prospective bidders how the project would be contracted, paid and governed Only after that work lands does the project go to market. If it reaches financial close, private capital funds the build under a PPP contract. If the feasibility or the market engagement fails, the project stays a pipeline entry, which is where several Kenyan PPP projects have historically stalled. ![Two officials shaking hands while holding the signed agreement](/images/konza-ppp-signing-handshake.webp) ## The Paper Trail Behind This Project The funding agreement is the latest entry in a procurement history that goes back years. The public record: | Date | Milestone | Where it is recorded | |------|-----------|---------------------| | Jun 2017 | Commercial contract with Huawei signed for the original Konza data centre and smart city scope | AidData loan record | | Apr 2019 | China Eximbank approves an RMB 1.225 billion (about KSh 17 billion) concessional loan | AidData; DataCenterDynamics | | Feb 2020 | Uptime Institute Tier III Design certification for the national data centre | Uptime Institute directory | | 2020 | Transaction advisory tender for the Konza Cloud Expansion and Smart City Facilities already in circulation | National Treasury PPP Directorate documents | | Aug 2022 | Tier III Constructed Facility certification; the national data centre goes live | Uptime Institute directory | | Nov 2025 | Feasibility study report for the cloud expansion completed | Executive summary published on konza.go.ke in April 2026 | | 17 Mar 2026 | Pre-bid meeting held for the transaction advisor tender | Tender notice | | Oct 2026 | PPP Project Facilitation Fund funding agreement signed | TechAfrica News; PPP Directorate | Two things stand out. First, the expansion has been in the PPP pipeline since at least 2020, and the tender documents were updated several times before the funding agreement finally landed. That history is a fair warning about pace. Second, the feasibility work is not theoretical: a 21-page executive summary on konza.go.ke describes a facility offering SaaS, PaaS, IaaS and colocation services, with a sustainability and operations plan attached. ## How the First Konza Data Centre Was Financed The operating facility at Konza, tracked by DC254 as the [Konza National Data Centre](/directory/konza-national-dc), was not built through a PPP. It was financed by the RMB 1.225 billion concessional loan from China Eximbank, signed in April 2019 with a 19.5-year maturity, a 2 percent fixed rate and a seven-year grace period. Huawei held the underlying commercial contract, signed in June 2017, and delivered the facility as contractor. ![The Konza National Data Centre building](/images/konza-national-dc-render.webp) That facility carries a Uptime Institute Tier III Constructed Facility certification dated 23 August 2022, has around 140 racks, and primarily serves government workloads. The loan's grace period ended in March 2026, so repayment obligations now sit on the government's books for the facility the loan built. The expansion takes a different route. Rather than a bilateral loan with a named contractor, KoTDA is structuring the expansion so private bidders compete to finance and deliver it. The state brings the site, the existing facility and the anchor demand. The market brings the capital. ## What the PPP Route Changes Three differences are worth watching. **Competition instead of appointment.** The Huawei build was a single-contractor arrangement inside a government-to-government loan. A PPP procurement, by design, puts the project in front of multiple bidders. If the transaction advisor's process works as intended, operators and investors already active in Nairobi get to price the expansion competitively. **Commercial discipline on the demand case.** A feasibility study that has to survive market engagement is a harder test than an internal plan. The executive summary's colocation ambition will be tested against a real question: does government demand plus Konza's SEZ incentives justify private capital 60 kilometres from Nairobi's cluster? [Konza's wider offer](/articles/konza-technopolis-data-centres-kenya) is covered separately on this site. **A published reference point.** Because the PPP process runs through documented stages, the expansion now has public milestones to be held against: the advisor award, the market engagement documents, and eventually the contract and financial close terms. Facilities built inside bilateral loan agreements rarely publish that much. ![Aerial view of the Konza Technopolis Complex with solar panels on the roof](/images/konza-technopolis-aerial.webp) ## How DC254's Tracker Reads It DC254 has added the expansion to the [directory](/directory) as its own record: Konza Cloud Expansion and Smart City Facilities (PPP), status Early Stage, verified October 2026. The operating facility keeps its existing record unchanged. The stage language is deliberate. Early Stage means a project exists in documents, with a sponsor and preparation funding, but has no operator, no published capacity and no financial close. The record moves to Committed when a PPP contract is signed and financed, to Under Construction on groundbreaking, and to Operational on certified delivery. Capacity fields stay empty until a figure is published; the tracker does not guess. ## What Would Move the Needle Four markers to watch over the next 12 to 18 months: 1. **Transaction advisor award.** The tender has been live since before March 2026. An award would signal the preparation funding is being spent as intended. 2. **Published feasibility detail.** The executive summary is public; the full report's scale assumptions are not. Any disclosed MW figure would be the first capacity number for the expansion. 3. **Market engagement documents.** The structure KoTDA and its advisor take to bidders will show whether the project is scoped around government cloud, commercial colocation, or both. 4. **Financial close.** The single biggest gate. Until a PPP contract is signed and financed, the expansion remains a pipeline project. ## What the Public Record Still Does Not Say Three gaps matter for anyone assessing this announcement. No contract value for the funding agreement has been published. No capacity figure, in MW or racks, exists in the public documents DC254 reviewed. And no construction timeline has been given, which is unsurprising at this stage but worth stating plainly. DC254 has reviewed the feasibility executive summary, not the full report. The summary describes services and a sustainability plan; it does not settle the demand or scale questions. This article describes the project's stage as the documents support it: ready to move into procurement, with everything after that unproven. Readers weighing Kenya's wider compute ambitions should pair it with the [compute hub plan explainer](/articles/kenya-data-centre-compute-hub-plan-explained) and the [data sovereignty context](/articles/kenya-data-sovereignty-localisation) that underpins government demand. ## Frequently Asked Questions ### What is the PPP Project Facilitation Fund? It is a National Treasury fund that pays for project preparation under Kenya's PPP Act 2021. It covers transaction advisory, feasibility studies and structuring costs that fall due before any private investor attaches to a project. The fund exists so government can move pipeline projects to a market-ready state at public cost rather than stalling at the idea stage. ### Does the funding agreement mean construction is starting at Konza? No. The agreement pays for preparation: transaction advisory services, feasibility assessments and the development of an investment and implementation framework. Construction would only follow after the project goes to market, a private partner is selected through procurement, and the PPP contract reaches financial close. ### What is the Konza Cloud Expansion and Smart City Facilities project? It is the PPP project to expand the existing national data centre at Konza Technopolis. Its feasibility documentation scopes services including SaaS, PaaS, IaaS and colocation. The project has been in Kenya's PPP pipeline since at least 2020, with a feasibility study report completed in November 2025 and a transaction advisor tender in progress since before March 2026. ### How was the existing Konza National Data Centre financed? Through a concessional loan of RMB 1.225 billion (about KSh 17 billion) from China Eximbank signed in April 2019, with Huawei as contractor under a commercial contract signed in June 2017. The facility holds a Uptime Institute Tier III Constructed Facility certification from August 2022, has around 140 racks, and primarily serves government workloads. ### How big will the Konza expansion be? No capacity figure has been published. Neither the funding agreement announcement nor the feasibility executive summary discloses a MW or rack target for the expansion. DC254's tracker keeps the record's capacity fields empty until a figure appears in official documents. ## Konza Technopolis and Kenya's Data Centre Future URL: https://data-centers-254.vercel.app/articles/konza-technopolis-data-centres-kenya Cluster: Kenya Market Published: 2026-08-28 Updated: 2026-08-28 Summary: Inside Konza Technopolis, Kenya's 5,000-acre smart city: purpose-built infrastructure, SEZ incentives, and what it means for data centres. ## Kenya's Most Ambitious Infrastructure Project Sixty kilometres south of Nairobi, on the arid plains of Machakos County, one of Africa's most ambitious smart city projects is taking shape. **Konza Technopolis** (a 5,000-acre development conceived as the cornerstone of Kenya's Vision 2030 strategy) aims to create a world-class technology hub that the government has dubbed **"Silicon Savannah."** ![Nairobi's skyline at dusk](/images/nairobi-skyline.webp) For Kenya's data centre industry, Konza represents both an enormous opportunity and a complex question. Can a purpose-built smart city, still in its early phases of development, attract the investment and demand needed to become a meaningful data centre location? Or will the established infrastructure and ecosystem of Nairobi continue to dominate? The answer will have significant implications for [Kenya's position as East Africa's digital hub](/articles/kenya-east-africa-digital-hub) and the broader trajectory of the continent's data centre market. ## The Vision 2030 Genesis Konza Technopolis was conceived as part of **Kenya Vision 2030**, the country's long-term development blueprint launched in 2008. The vision identified information and communications technology (ICT) as a key pillar of economic transformation, and Konza was designed as the physical manifestation of this ambition, a dedicated city where technology companies, research institutions, and government agencies would co-locate to create an innovation ecosystem. The [Konza Technopolis Development Authority (KoTDA)](https://konza.go.ke) was established by an Act of Parliament in 2012 to oversee the planning, development, and management of the project. The **World Bank has provided project appraisal support and financing**, recognizing Konza's potential to catalyze Kenya's knowledge economy. ![Aerial view of the Parliament Buildings of Kenya, Nairobi](/images/kenyan-parliament-buildings.webp) The master plan for Konza is comprehensive: a city designed for approximately 200,000 residents and 100,000 jobs, with dedicated zones for business process outsourcing, life sciences, education, and (critically for the data centre industry) **technology infrastructure including data centres**. ## Current Development Status Konza's development has been progressing in phases, with the [World Bank supporting the project through its Konza Technopolis Development Project](https://documents.worldbank.org/en/publication/documents-reports/documentdetail/konza-technopolis). ### Phase 1 Progress Phase 1 of Konza covers approximately 400 acres of the total 5,000-acre site and focuses on core infrastructure and the first batch of buildings: - **Roads**: Internal road networks within the Phase 1A area have been completed, connecting the core zone to the A104 highway and the Nairobi-Mombasa railway - **Power**: A dedicated power substation has been constructed, with provisions for dual-feed redundancy from the national grid. The substation is designed to support the high power densities required by data centres and other technology facilities - **Water and sewerage**: Modern water supply and wastewater treatment infrastructure has been installed - **Fibre connectivity**: Multiple fibre providers have laid connectivity to the site, including via the **Kenya Electricity Transmission Company (KETRACO)** fibre network that runs along high-voltage transmission lines - **Konza Technopolis Complex**: The first major building (a mixed-use facility for KoTDA operations, a conference centre, and incubation space) is operational ### Timeline Expectations The Konza development timeline has experienced delays that are common in projects of this scale. While the original vision anticipated a faster buildout, the [Kenya data centre market outlook to 2030](/articles/kenya-data-centre-market-outlook-2025-2030) suggests that the most meaningful data centre development at Konza is likely to materialise in the **2027-2030 timeframe**, as Phase 1 infrastructure matures and tenant demand becomes more visible. ## The Data Centre Component ### Designated Land and Zoning Konza's master plan includes **specifically designated zones for data centre development**, with plots sized and zoned to accommodate facilities ranging from 1 MW to 50+ MW. The designated data centre zone is positioned to take advantage of the site's power and connectivity infrastructure while maintaining appropriate buffers from residential and commercial areas. This is a significant advantage over developing a data centre within Nairobi's urban core, where **zoning constraints, land availability, and noise regulations** can complicate or prevent data centre construction. ### Power Infrastructure ![Data centre cooling and power distribution equipment](/images/dc-cooling-crac-2.webp) Power is the single most critical (and expensive) input for data centre operations, and Konza's power infrastructure is designed with this in mind: - **Dedicated substation**: A purpose-built substation within the Technopolis, designed to handle the high power loads required by data centres and other technology operations - **Dual-feed redundancy**: The substation is connected to the national grid via two independent feed lines, providing N+1 redundancy at the utility level, a requirement for Tier III and above data centres - **Expansion capacity**: The power infrastructure is designed with significant headroom for expansion, meaning new data centre tenants will not need to wait years for power upgrades - **Proximity to geothermal energy**: Konza's location provides relatively straightforward access to Kenya's geothermal power resources from the Olkaria and Menengai fields via the national grid, supporting the growing demand for [renewable energy-powered data centres](/articles/geothermal-energy-kenya-data-centres) ### Fibre Connectivity Data centres require multiple, diverse fibre connections for redundancy and performance. Konza is served by: - **Multiple fibre providers**: Several licensed fibre network operators have connectivity to the site - **Proximity to submarine cable routes**: While Konza is inland, its connectivity to Nairobi and Mombasa means it benefits from the multiple submarine cable systems landing at the Kenyan coast - **Interconnection to KIXP**: Although the Kenya Internet Exchange Point (KIXP) is physically located in Nairobi, Konza-based data centres can peer via extended connectivity, ensuring efficient local traffic exchange ### Confirmed and Prospective Investors While no major data centre operator has publicly committed to building at Konza as of 2026, several factors suggest this is likely to change: - **Africa Data Centres** (part of the Remgro group) has expressed interest in expanding its footprint beyond Nairobi, and Konza's infrastructure and incentives make it a logical candidate - **Liquid Intelligent Technologies** (formerly Liquid Telecom) has been expanding its data centre presence across Africa and could view Konza as a strategic location - **International cloud providers** exploring African expansion (including potential AWS, Google Cloud, and Azure local regions) may find Konza's purpose-built infrastructure and SEZ incentives attractive compared to building in congested urban areas - **The Kenyan government** itself is a significant potential anchor tenant, as it digitises government services and seeks sovereign data hosting capabilities ## Konza's Advantages Over Nairobi ### Lower Land Costs Land in Nairobi's commercial zones (particularly areas like Westlands, Upper Hill, and Industrial Area where existing data centres are concentrated) is among the most expensive in East Africa. Konza offers land at **a fraction of Nairobi's commercial rates**, a meaningful advantage for capital-intensive data centre projects where land acquisition can represent 5-15% of total project costs. ### Purpose-Built Infrastructure Nairobi's data centres operate within an urban environment where infrastructure was not designed for their specific requirements. Power capacity may need upgrading, fibre may need to be extended, and access roads may not be designed for heavy equipment. Konza's infrastructure is **designed for the loads and requirements of technology operations**, reducing both capital expenditure and development timelines. ### Planned Redundancy The Konza master plan incorporates redundancy in power, water, and connectivity from the outset. In Nairobi, achieving the same level of redundancy often requires costly and time-consuming infrastructure upgrades. ### SEZ Tax Incentives This is perhaps Konza's most significant competitive advantage. The **Special Economic Zone (SEZ) designation** provides: - **10% corporate tax rate** (compared to Kenya's standard 30%) for the first 10 years, with a potential 15% rate for the next 10 years - **VAT exemption** on imported equipment, servers, and construction materials, a significant benefit given that most data centre equipment is imported - **Stamp duty exemption** on land transactions within the SEZ - **Accelerated depreciation** on capital assets, improving cash flow for capital-intensive operations For a [colocation data centre in Kenya](/articles/colocation-data-centre-kenya) investing KES 5-10 billion in infrastructure, these tax incentives can reduce the effective cost of operations by 15-25% over a decade, a difference that can be the deciding factor in location selection. ### Pre-Handled Environmental Clearances Environmental Impact Assessments (EIAs) and other regulatory clearances have been completed at the zonal level for Konza, meaning new data centre developments within the designated zones face a **simplified approval process** compared to greenfield developments elsewhere. This can shave months off development timelines. ## The Challenges Konza Faces ### Distance from Nairobi's Business District ![Transmission pylons crossing open country](/images/kenya-transmission-pylons-2.webp) At 60 kilometres from Nairobi's Central Business District, Konza is not a trivial commute. For data centre operations that require on-site staff 24/7, the distance creates challenges in talent attraction and retention. Many experienced data centre professionals currently live and work in Nairobi, and convincing them to relocate or commute to Konza requires significant incentives. The **standard-gauge railway** connecting Nairobi to Konza (via the Athi River station) and the improved road network mitigate this to some extent, but the distance remains a psychological and practical barrier. ### Limited Existing Ecosystem Data centres benefit from agglomeration effects, the clustering of complementary businesses, service providers, and skilled workers. Nairobi has this ecosystem: network providers, equipment vendors, managed service providers, and a deep talent pool. Konza is building this ecosystem from scratch, and ecosystem development takes years, even with proactive effort. ### Construction Delays and Execution Risk Large-scale infrastructure projects in developing economies frequently experience delays due to funding constraints, political changes, procurement challenges, and construction logistics. Konza has already experienced delays relative to its original timeline, and potential data centre investors will be evaluating the project's execution track record carefully. ### The Demand Question The most fundamental challenge facing Konza's data centre ambitions is demand. **Is there sufficient demand for additional data centre capacity in Kenya** to justify building in a new, unproven location? Kenya's existing data centre supply in Nairobi is still growing to meet demand, and some analysts question whether the market can support a second significant data centre cluster. Proponents argue that Konza's lower costs and SEZ incentives will **create demand that would not otherwise exist**, attracting international companies that would not have considered Kenya without the specific advantages Konza offers. Detractors counter that the market is not yet large enough and that concentrating investment in Nairobi's existing cluster would be more efficient. ## Comparing Konza to Other African Tech Cities ### Rwanda's Kigali Innovation City **Kigali Innovation City** is a smaller, more focused project than Konza. Located on 61 hectares in Kigali, it aims to create an innovation hub with a strong emphasis on technology education, research, and incubation. Rwanda's stable governance, ease of doing business, and strategic positioning as a conference and tech hub have attracted significant international attention. Compared to Konza, Kigali Innovation City benefits from Rwanda's smaller, more agile government and strong international partnerships, but is constrained by Rwanda's smaller market size and lack of submarine cable landing infrastructure. ### Nigeria's Eko Atlantic City **Eko Atlantic City** in Lagos is a much larger, commercially-driven development that aims to create a new financial and business district on reclaimed land adjacent to Victoria Island. While not specifically a technology city, Eko Atlantic is attracting technology companies and data centre operators due to its modern infrastructure and proximity to Nigeria's massive market. Eko Atlantic's model is more market-driven than Konza's government-led approach, which brings both advantages (faster commercial development) and disadvantages (less strategic coherence). ### Ethiopia's ICT Park **Ethiopia's ICT Park** in Addis Ababa is a government-led industrial park focused on IT outsourcing, business process outsourcing, and technology manufacturing. It has attracted some international BPO operators but has faced challenges related to Ethiopia's regulatory environment, foreign exchange constraints, and political instability. Konza's advantage over the Ethiopian model lies in Kenya's more liberalised telecom sector, stronger private sector participation, and established reputation as a technology leader in East Africa. ## What Konza's Success or Failure Would Mean The implications of Konza's trajectory for Kenya's data centre industry are significant in either scenario: ### If Konza Succeeds A successful Konza Technopolis with multiple operational data centres would: - **Validate Kenya's Vision 2030** strategy and attract further investment in the country's digital infrastructure - **Create a second data centre cluster** that reduces concentration risk and provides geographic diversity for disaster recovery and business continuity - **Attract international operators** who require purpose-built infrastructure and SEZ incentives - **Create a technology ecosystem** that extends beyond data centres to include cloud services, AI development, fintech, and digital government services - **Position Kenya as the clear leader** in East Africa's data centre market, ahead of competing markets in Tanzania, Uganda, and Rwanda ### If Konza Struggles If Konza fails to attract significant data centre investment, the implications would be more nuanced but still important: - **Nairobi's dominance would continue**, with data centre development concentrated in the capital's existing commercial areas - **Investor confidence in government-led tech city projects** across Africa could be affected - **The SEZ model for data centre development** would face questioning, potentially slowing similar initiatives elsewhere - **Kenya would still benefit** from the infrastructure investments already made at Konza, but the project would fall short of its transformative ambitions ## Conclusion Konza Technopolis represents Kenya's most ambitious bet on its digital future. For the data centre industry, it offers a compelling combination of purpose-built infrastructure, significant tax incentives, and a strategic vision that aligns with the country's growth trajectory. The challenges (distance, ecosystem development, execution risk, and demand uncertainty) are real but not insurmountable. The next five years will be decisive. As Phase 1 infrastructure matures and the first major tenants establish operations, the market will determine whether Konza can deliver on its promise as a [colocation and cloud infrastructure hub](/articles/colocation-data-centre-kenya) that complements and extends Kenya's data centre ecosystem beyond Nairobi. For data centre operators, investors, and technology companies watching Kenya's market, Konza is a development that deserves close attention. The project's trajectory will shape not just where data centres are built in Kenya, but the country's ability to maintain its position as East Africa's leading digital economy. ## Frequently Asked Questions ### What is Konza Technopolis and how does it relate to data centres? Konza Technopolis is a 5,000-acre smart city project located 60km south of Nairobi, conceived under Kenya's Vision 2030 to create a world-class technology hub. The development includes designated land for data centre construction, with purpose-built power infrastructure (dedicated substation), redundant fibre connectivity from multiple providers, and Special Economic Zone status offering 10% corporate tax rate. These features are specifically designed to attract data centre operators and cloud service providers. ### What tax incentives does Konza offer for data centre operators? Konza Technopolis has been designated as a Special Economic Zone (SEZ), which provides several significant tax incentives: a reduced corporate tax rate of 10% (compared to Kenya's standard 30%), exemption from VAT on imported equipment and construction materials, exemption from stamp duty, and accelerated depreciation on capital assets. For capital-intensive data centre operations where equipment costs can exceed tens of millions of dollars, these incentives can substantially improve the investment case. ### What infrastructure is already in place at Konza Technopolis? As of the latest Konza Technopolis Development Authority (KoTDA) reports, Phase 1 infrastructure includes: completed road networks within the core area, a dedicated power substation with dual-feed capability, fibre optic connectivity from multiple providers including Kenya Electricity Transmission Company (KETRACO), water and sewerage systems, and the Konza Technopolis Smart City Basic Infrastructure. The Phase 1A area, which includes the Konza Technopolis Complex building, is operational. ### How does Konza compare to Nairobi for data centre development? Konza offers several advantages over Nairobi: significantly cheaper land (a fraction of Nairobi's commercial real estate costs), purpose-built infrastructure designed for technology companies, planned redundancy in power and connectivity, pre-handled environmental clearances, and SEZ tax incentives. However, Nairobi has the advantage of an established business ecosystem, proximity to customers, existing fibre interconnection points like KIXP, and a larger pool of skilled workers. The two locations are likely to serve different market segments. ### How does Konza compare to other African tech city projects? Konza is often compared to Rwanda's Kigali Innovation City (a smaller, more focused project with strong government backing), Nigeria's Eko Atlantic City in Lagos (a larger, more commercially-driven development), and Ethiopia's ICT Park in Addis Ababa (a government-led industrial park focused on IT outsourcing). Konza is distinguished by its scale (5,000 acres), its explicit focus on becoming a technology and science hub, and its integration with Kenya's broader Vision 2030 development strategy. ## Large Language Models in Kenya: Compute and Sovereignty URL: https://data-centers-254.vercel.app/articles/large-language-models-kenya Cluster: AI & Machine Learning Published: 2026-08-28 Updated: 2026-09-28 Summary: Which Kenyan organisations are adopting LLMs, the compute they need, and why data-sovereignty rules under Kenyan law decide where models can run. ![Server racks in a Nairobi data centre hall](/images/dc-gpu-cluster-2.webp) ## The LLM Revolution Reaches Kenya Large language models (LLMs) have moved from research curiosity to enterprise necessity in under three years. Globally, organisations are deploying generative AI for customer service, content creation, data analysis, software development, and decision support. Kenya is no exception, but the country's adoption trajectory is shaped by unique infrastructure constraints, regulatory considerations, and a rapidly maturing local AI ecosystem. The numbers tell a compelling story. According to the Communications Authority of Kenya, internet penetration in the country exceeded 85% in 2025, with mobile data usage growing at over 20% year-on-year. This digital foundation, combined with Kenya's position as East Africa's leading tech hub, has created fertile ground for LLM adoption. Yet the gap between **enthusiasm and infrastructure** remains the defining challenge. This article examines how Kenyan organisations are adopting LLMs today, the compute infrastructure required, the data sovereignty implications under Kenyan law, and what the future holds for local AI deployment. ## How Kenyan Organisations Are Using LLMs Today ### Telecommunications: Safaricom's AI-Powered Customer Service Safaricom, Kenya's largest telecoms operator with over 45 million subscribers, has been at the forefront of AI adoption. The company has integrated LLM-powered chatbots into its customer service operations, handling thousands of routine enquiries daily, from M-Pesa transaction queries to data bundle requests. While Safaricom has not publicly disclosed whether it runs these models locally or via cloud APIs, the volume of personal data processed makes the [Kenya Data Protection Act and its implications for data centres](/articles/kenya-data-protection-act-data-centres) a critical consideration in their architecture decisions. The efficiency gains are substantial. Industry benchmarks suggest that AI chatbots can resolve 60-70% of tier-one customer enquiries without human intervention, reducing call centre costs by 30-40% while improving response times from minutes to seconds. ### Banking and Financial Services: NCBA and Beyond NCBA Group has deployed an AI-powered banking assistant that helps customers with account enquiries, loan information, and financial product recommendations. This builds on the broader trend of [machine learning adoption in Kenyan banking and fintech](/articles/machine-learning-kenyan-banking-fintech), where LLMs are adding a natural language interface to existing ML systems. Equity Bank and KCB Group are reported to be exploring similar deployments, particularly for customer onboarding in multiple Kenyan languages. The ability of LLMs to process and generate text in Kiswahili and vernacular languages represents a significant opportunity for financial inclusion, potentially bringing banking services to millions of underserved Kenyans who are more comfortable in their mother tongue than in English. ### Agriculture and Supply Chain: Twiga Foods Twiga Foods, Kenya's leading agricultural supply chain platform connecting over 14,000 farmers to urban markets, has begun leveraging LLM technology for supply chain optimisation. This includes analysing market data, generating demand forecasts in natural language for farmers, and automating communications between suppliers and buyers. The application of AI in agriculture is particularly impactful in Kenya, where smallholder farmers produce over 70% of the country's food supply. ### Government and Public Services The Kenyan government has shown growing interest in LLM applications. The Kenya Revenue Authority has explored AI tools for taxpayer communication and compliance assistance. The e-Citizen platform, which processes millions of transactions monthly for services ranging from passport applications to business registration, is evaluating LLM-powered interfaces to simplify citizen interactions with government services. However, government adoption faces additional hurdles beyond infrastructure. **Data sovereignty requirements** are particularly stringent for public sector data, and the regulatory framework for government use of AI is still under development. ### Media and Content Creation Kenya's vibrant media sector has embraced generative AI tools for content drafting, translation between English and Kiswahili, and social media management. Nation Media Group, The Standard Group, and several digital-native publishers have adopted AI-assisted workflows. While this raises important questions about journalistic integrity and attribution, the productivity gains in fast-paced newsrooms are driving rapid adoption. ![Fibre optic termination equipment in a data centre](/images/dc-fibre-optics.webp) ## The Infrastructure Challenge: Inference vs Training Understanding the distinction between **LLM training** and **LLM inference** is essential for assessing Kenya's infrastructure needs. ### Training: Beyond Kenya's Current Reach Training a large language model from scratch requires clusters of hundreds or thousands of GPUs (typically NVIDIA H100s or A100s) running continuously for weeks or months. Meta's Llama 3 70B model reportedly required over 6,000 GPUs for training. The capital expenditure for such infrastructure runs into tens of millions of dollars, and the power requirements (often exceeding 10 megawatts for a training cluster) put it well beyond what Kenya's current data centre ecosystem can support. This is a reality Kenya must accept strategically: **large-scale foundation model training will remain offshore for the foreseeable future**. The country's power grid, while improving, cannot yet reliably deliver the dedicated, high-density power that GPU training clusters demand. As explored in our analysis of [GPU computing needs in Kenya's data centres](/articles/gpu-computing-kenya-ai-needs-data-centres), the gap between current capacity and AI training requirements is measured in orders of magnitude. ### Inference: The Achievable Frontier Inference (using a trained model to generate responses) requires far less compute. A 7-billion parameter model like Llama 3 8B or Mistral 7B can run on a single NVIDIA A100 (80GB) GPU server for concurrent user queries. Even a 13-billion parameter model can be served efficiently with quantisation techniques that reduce memory requirements without significant quality loss. This is where Kenya's opportunity lies. A single GPU server, collocated in a Nairobi data centre with adequate power and cooling, can serve LLM inference for dozens of enterprise clients. The capital investment is in the range of USD 100,000–300,000 for the GPU hardware, plus ongoing colocation and connectivity costs, well within the budget of Kenya's larger enterprises and technology companies. ![Technician working with fibre in a data hall](/images/fibre-splicing-technician-data-hall.webp) ## The Emerging Kenyan AI Ecosystem ### Local AI Startups Nairobi's startup ecosystem is producing a growing number of AI-native companies. iCompute AI has focused on building AI solutions tailored to the African market, including natural language processing for Kenyan languages. Several other Nairobi-based startups are working on domain-specific LLM applications in healthcare diagnostics, legal document analysis, and agricultural advisory services. The Microsoft Africa Development Centre (ADC) in Nairobi, which employs over 500 engineers, has made significant contributions to global AI products, including work on Azure AI services, Copilot features, and models optimised for African languages. The [Microsoft Africa Development Centre in Nairobi](https://www.microsoft.com/en-us/africa/development-centre) represents the largest single concentration of AI engineering talent in East Africa. ### Regional AI Research While Kenya's dedicated AI research infrastructure is still developing, the country benefits from regional initiatives. Google's AI research centre in Accra, Ghana (the only Google AI research centre on the African continent) produces research that serves the entire region, including Kenya. Their work on African language models, agricultural AI, and health diagnostics has direct applicability to Kenyan use cases. ### Open-Source Models: Lowering the Barrier The availability of high-quality open-source LLMs has been transformative for Kenyan organisations. [Meta's Llama family of models](https://llama.meta.com/) (particularly Llama 3 8B and 70B) and Mistral AI's models provide capabilities that, two years ago, were only available through expensive proprietary APIs. The advantages of open-source models for Kenya are significant: - **Data sovereignty**: Organisations can run models entirely on-premises or in Kenyan data centres, ensuring personal data never leaves the country. - **Cost predictability**: No per-token API charges means costs are fixed and known in advance. - **Customisation**: Models can be fine-tuned on domain-specific Kenyan data (legal documents, medical records, agricultural data) without sharing that data with external providers. - **Kiswahili and vernacular support**: Open-source models can be fine-tuned on local language datasets, improving their performance for Kenyan users. The Africa-centric fine-tuning of models like Llama for Kiswahili and other African languages is an active area of work, with researchers at the University of Nairobi and Strathmore University's iLabAfrica contributing to multilingual model development. ## Data Sovereignty: The Kenya Data Protection Act and LLMs The Kenya Data Protection Act 2019 (DPA) is the primary legal framework governing how personal data is processed, stored, and transferred. For organisations deploying LLMs, several provisions are directly relevant. **Section 48 of the DPA** restricts cross-border data transfers to countries or territories with comparable data protection standards, or where adequate safeguards are in place. When a Kenyan bank sends customer data to OpenAI's servers in the United States for processing through ChatGPT, this constitutes a cross-border data transfer that must comply with these requirements. The Office of the Data Protection Commissioner (ODPC) has issued guidance indicating that organisations using foreign cloud-based AI services must conduct a Data Protection Impact Assessment (DPIA) and implement appropriate contractual safeguards. This regulatory pressure is a key driver pushing Kenyan enterprises toward local LLM deployment. For sectors handling particularly sensitive data, banking (regulated by the Central Bank of Kenya), healthcare (regulated by the Ministry of Health), and government services, the case for local inference infrastructure is strongest. These organisations cannot easily justify sending citizen or customer data to foreign APIs, creating a clear market opportunity for GPU-equipped data centre facilities in Nairobi. ![Technician terminating fibre optic cable](/images/fibre-splicing-termination.webp) ## What Infrastructure Would Kenya Need? To support meaningful local LLM deployment, Kenya needs investment in several interconnected areas. ### GPU-Equipped Colocation Facilities Current Kenyan data centres are primarily designed for standard enterprise workloads with rack power densities of 3-8 kW. GPU inference servers, even for smaller models, require 10-20 kW per rack. Purpose-built AI inference colocation (with higher power density per rack, liquid cooling or enhanced air cooling, and high-bandwidth, low-latency networking) is the foundational requirement. ### NVIDIA Inference GPUs The NVIDIA A100 and H100 remain the gold standard for LLM inference. The newer L40S and H200 GPUs offer improved performance per watt. For cost-sensitive deployments, NVIDIA's T4 and even consumer-grade RTX 4090 GPUs can serve smaller models, though with limitations on concurrent user capacity. The availability of these GPUs in Kenya is improving through distributors like Spectra Systems and other IT infrastructure providers, but pricing remains 20-40% higher than in the US or Europe due to import duties and logistics. ### Edge AI for Low-Latency Applications Not every LLM deployment requires a data centre. Edge AI (running smaller, optimised models on devices or on-premises servers) has compelling use cases in Kenya. Retail businesses can run customer service bots on local servers. Healthcare facilities can deploy diagnostic AI without internet connectivity. Agricultural cooperatives can use AI advisors in areas with limited connectivity. ### Connectivity and Latency While Kenya has excellent international connectivity through the submarine cables landing in Mombasa, internal latency between data centre facilities and end users varies. A well-connected Nairobi data centre can serve clients across East Africa with 20-50ms latency, adequate for most LLM inference applications, but demanding for real-time voice or video AI features that require sub-100ms end-to-end response times. ## The Skills Gap and Training Programmes Kenya's most significant long-term constraint on LLM adoption is not hardware, it is people. The country needs professionals skilled in prompt engineering, model fine-tuning, MLOps, GPU cluster management, and AI safety. Several programmes are addressing this gap: - **Microsoft ADC Skills Programme**: Microsoft's Africa Development Centre runs technical training programmes for Kenyan developers, including AI and machine learning modules. - **Google Developer Groups (GDG) Nairobi**: Regular meetups and workshops on AI technologies, including LLM fine-tuning and deployment. - **University of Nairobi AI Research Lab**: Academic research and postgraduate training in machine learning and natural language processing. - **Strathmore University iLabAfrica**: Applied research and industry partnerships focused on AI solutions for African challenges. - **DeepLearning.AI**: Online courses, many available for free, covering practical LLM deployment skills. - **Andela and other talent platforms**: Connecting Kenyan AI engineers with global and local opportunities. Despite these programmes, demand for AI skills in Kenya far outstrips supply. The Kenya ICT Board estimates that the country needs at least 10,000 additional AI-skilled professionals by 2028 to meet industry demand, a target that will require significant scaling of current training efforts. ![Power hall inside a Kenyan data centre during a working site visit](/images/site-visit-power-hall-kenya.webp) ## The Road Ahead Kenya's LLM ecosystem is at an inflection point. The demand is real and growing across every major sector. The technology (particularly open-source models) is accessible and increasingly optimised for the African context. The regulatory environment, while still evolving, is creating a clear incentive for local infrastructure investment. What remains is the **infrastructure buildout**, GPU-equipped colocation in Nairobi, skills development at scale, and a policy framework that supports innovation while protecting citizens' data rights. Organisations that invest in these capabilities now will be positioned to capture the enormous value that LLMs can deliver in Kenya's rapidly digitising economy. The question is no longer whether Kenya will develop significant LLM infrastructure, but how quickly and through what combination of public and private investment. The foundations are being laid today. ## Frequently Asked Questions **Can Kenyan organisations run LLMs locally without sending data abroad?** Yes, but with limitations. Small to medium LLMs (7B-13B parameters) such as Llama 3, Mistral 7B, and Phi-3 can run on a single NVIDIA A100 or H100 GPU server for inference. Kenya's data centres are beginning to offer GPU-capable colocation, but options remain limited compared to Europe or the US. Organisations handling sensitive data, such as banks and hospitals, are the primary drivers of local deployment. **What is the difference between LLM training and inference, and why does it matter for Kenya?** Training involves teaching a model on massive datasets and requires clusters of hundreds or thousands of GPUs running for weeks or months. Inference is using a trained model to generate responses and needs far less compute, often a single GPU server for smaller models. Kenya is not positioned for large-scale training due to power and GPU infrastructure gaps, but inference is entirely feasible with modest infrastructure investments. **How does the Kenya Data Protection Act affect LLM usage?** The Kenya Data Protection Act 2019 requires that personal data processed by organisations must have a lawful basis, and data controllers must ensure appropriate security measures. Sending personal data to foreign LLM APIs (such as OpenAI or Google) may constitute cross-border data transfer, which under the Act requires adequacy assessment or appropriate safeguards. This is pushing Kenyan banks and government agencies toward local or self-hosted LLM solutions. **Which Kenyan companies are actively using large language models?** Safaricom uses AI chatbots for customer service, NCBA Group has deployed AI-driven banking assistants, Twiga Foods applies LLM technology for supply chain optimisation, and several media houses use generative AI for content drafting and translation. Government agencies, particularly the Kenya Revenue Authority and the e-Citizen platform, are exploring LLM-powered citizen service tools. **What skills are needed for LLM deployment in Kenya, and where can people learn them?** Key skills include prompt engineering, model fine-tuning, MLOps (model deployment and monitoring), GPU server management, and data engineering. Training programmes are available through the Microsoft ADC skills programme, Google Developer Groups in Nairobi, the University of Nairobi's AI research lab, Strathmore University's iLabAfrica, and online platforms like DeepLearning.AI. The skills gap remains significant, with demand far outstripping supply. ## Liquid Intelligent: The Backbone of Kenya's Data Centres URL: https://data-centers-254.vercel.app/articles/liquid-intelligent-technologies-kenya Cluster: Operators Published: 2026-08-29 Updated: 2026-08-29 Summary: Liquid Intelligent Technologies runs Africa's largest fibre network and Nairobi data centres, connecting Kenya to 13+ countries and subsea cables. ![Landing crew securing a submarine cable buoy on the beach](/images/mombasa-cable-landing-4.webp) Liquid Intelligent Technologies is not a data centre company. That distinction matters, because understanding who Liquid is and what they do is essential to understanding how data works in Kenya and across East Africa. Liquid is a connectivity company, a fibre infrastructure builder, and an enterprise technology services provider. They happen to sit at the centre of Kenya's data centre ecosystem because every rack, every cloud region, and every submarine cable landing in Mombasa ultimately relies on the kind of terrestrial fibre infrastructure that Liquid has spent two decades building across the continent. As a subsidiary of Cassava Technologies, the pan-African technology group founded by Strive Masiyiwa, Liquid operates over 100,000 kilometres of fibre spanning more than 13 African countries. In Kenya, their network forms the critical link between submarine cable landing stations on the coast and data centre clusters in Nairobi, while also extending connectivity to landlocked East African nations that depend on Kenya as their gateway to the global internet. This profile examines Liquid's origins, their infrastructure in Kenya, their relationship to Africa Data Centres, and why their connectivity-first model makes them fundamentally different from pure-play data centre operators competing in Nairobi. ## From Econet to Liquid: The Origins Liquid Intelligent Technologies traces its roots to Econet Wireless, the mobile telecommunications group that Strive Masiyiwa founded in Zimbabwe in 1993. Masiyiwa, one of Africa's most prominent entrepreneurs, spent years battling regulatory barriers before successfully launching Econet as Zimbabwe's first private mobile network. The company later expanded across multiple African markets. The data infrastructure story began to take shape in the early 2000s when Econet started building terrestrial fibre networks to connect its own mobile towers and operations. What started as internal infrastructure gradually evolved into a wholesale connectivity business. By 2013, the operation had been rebranded as Liquid Telecom, reflecting a pivot towards becoming a dedicated pan-African fibre and data services provider. The rebrand to Liquid Intelligent Technologies in 2021 signalled a further evolution. The company was no longer just a telecoms carrier. It was positioning itself as a full-stack technology company offering cloud computing, cybersecurity, collaboration tools, and managed IT services alongside its core fibre infrastructure. This broader ambition led to the creation of Cassava Technologies as the parent group, with Liquid Intelligent Technologies serving as the connectivity and enterprise services arm alongside sister companies like Africa Data Centres and Liquid Datapoint. ## The Fibre Network: 100,000 Kilometres Across Africa ![Technician terminating fibre cabling at network racks](/images/technician-rack-cabling-kenya.webp) Liquid's single most important asset is its fibre network. At over 100,000 kilometres, it is the largest independently owned fibre infrastructure on the African continent. The network spans more than 13 countries, including South Africa, Zimbabwe, Zambia, Kenya, Tanzania, Uganda, Rwanda, Burundi, the Democratic Republic of Congo, Mozambique, Lesotho, Botswana, and Nigeria. The network's architecture is designed around a principle that matters enormously for data centre operators: redundancy through geographic diversity. Rather than relying on a single route between any two points, Liquid's fibre follows multiple paths. In East Africa, this means that traffic between Nairobi and Mombasa can travel along separate physical routes, reducing the risk that a single cable cut will take down connectivity to the submarine cable landing stations. For Kenya's data centre industry, Liquid's network serves several critical functions. First, it provides the terrestrial backhaul that connects Nairobi's data centre cluster along Mombasa Road to the submarine cable systems landing at the coast. Kenya is served by multiple submarine cables including SEACOM, TEAMS, EASSy, DARE1, and others, all of which land in Mombasa. Without robust terrestrial fibre linking Mombasa to Nairobi, those cables would be of limited value to the data centres where the actual compute and storage happens. Second, Liquid's network extends beyond Kenya's borders to connect landlocked countries. Uganda, Rwanda, Burundi, South Sudan, and eastern DRC all rely on transit through Kenya to reach submarine cable bandwidth. Liquid's cross-border fibre routes mean that a data centre in Nairobi can serve customers across the entire East African region, with connectivity delivered over Liquid's own infrastructure rather than a patchwork of third-party providers. ## Liquid's Kenya Operations Liquid's Kenyan operations are headquartered in Nairobi and form a critical part of the company's East African strategy. The company holds licences from the Communications Authority of Kenya to operate as a network facilities provider and telecommunications services operator. Their Kenyan fibre network runs from Mombasa through Nairobi and extends westwards towards Uganda. In Nairobi, Liquid maintains a significant enterprise presence, providing dedicated fibre links, MPLS networks, SD-WAN solutions, and internet transit to banks, government agencies, multinational corporations, and technology companies. Many of these customers house their IT infrastructure in Nairobi data centres, creating a natural link between Liquid's connectivity services and the colocation facilities where their servers reside. ![Aerial view of the Port of Mombasa](/images/mombasa-port.webp) Liquid also provides wholesale bandwidth and transit services to other operators, internet service providers, and mobile network operators in Kenya. This wholesale business is less visible to end users but is foundational to how the internet works in the country. Smaller ISPs that cannot justify building their own submarine cable capacity or long-haul fibre rely on purchasing bandwidth from operators like Liquid who have that infrastructure in place. ## The Nairobi Data Centre and Liquid Datapoint Liquid's own data centre presence in Kenya operates primarily through Liquid Datapoint, a subsidiary focused on providing data centre, cloud, and managed IT services. Liquid Datapoint offers colocation, managed hosting, and cloud computing services from facilities that are connected to Liquid's fibre backbone. Liquid Datapoint occupies a distinct position in the Kenyan market. Unlike Africa Data Centres, which positions itself as a carrier-neutral colocation platform, Liquid Datapoint tends to serve customers already using Liquid's connectivity services who want an integrated solution combining network, compute, and managed services from a single provider. In practice, this means Liquid Datapoint competes less aggressively on pure colocation pricing and more on the value of integrated service delivery. A bank that wants dedicated fibre connectivity, MPLS links to branches, and a managed data centre environment can procure all of this from the Liquid group, simplifying procurement, support, and contract management. ## The Cassava Technologies Family: Liquid vs Africa Data Centres The relationship between Liquid Intelligent Technologies and Africa Data Centres is a source of frequent confusion in the Kenyan market, and clarifying it is important for anyone evaluating data centre options. Both companies are subsidiaries of Cassava Technologies. Before the Cassava rebrand, the entire group operated under the Liquid Intelligent Technologies name, which is why you will still encounter references to "Liquid data centres" in older materials. The corporate restructure separated the business into focused entities. Africa Data Centres is the dedicated data centre infrastructure company. It builds and operates carrier-neutral colocation facilities, including two in Nairobi and one in Mombasa. These facilities are designed to host multiple connectivity providers, and Liquid Datapoint is one of several carriers available within them. Liquid Intelligent Technologies is the connectivity and enterprise technology services company. It builds and operates fibre networks, provides internet and private network services, and offers cloud, cybersecurity, and collaboration tools. Liquid Datapoint is its data centre and managed services subsidiary. The practical implication for Kenyan customers is that you can colocate in an Africa Data Centres facility and connect to Liquid's network from there, or you can colocate directly with Liquid Datapoint and get an integrated connectivity and hosting package. The two paths serve different customer preferences. The carrier-neutral route offers more flexibility and competitive pricing on connectivity. The integrated route offers simplicity and a single point of accountability. ## Connecting Landlocked Countries: Kenya as a Regional Hub One of Liquid's most strategically important roles in East Africa is connecting landlocked countries to the global internet through Kenya's submarine cable infrastructure. This function is central to understanding why Kenya's data centre market is larger than its domestic population and economy would suggest. Uganda, Rwanda, and Burundi have no coastline and no direct access to submarine cables. All their international internet traffic must transit through a neighbouring country with landing stations. Kenya, through Mombasa, is the primary gateway, and Liquid's cross-border fibre network is one of the main arteries carrying that transit traffic. Liquid's fibre routes from Nairobi westwards to Kampala, Kigali, and Bujumbura mean a data centre in Nairobi can serve customers across the entire East African Community with low-latency connectivity. This regional reach makes Nairobi a more attractive location for data centre investment, because a single facility can address a market of over 200 million people across multiple countries. The transit business generates significant revenue for Liquid's Kenyan operations. Every megabit of bandwidth flowing from Mombasa to Kampala or Kigali over Liquid's fibre contributes to the economics that justify continued investment in network expansion and reliability. ## Enterprise Services and Cloud Partnerships Liquid's enterprise business in Kenya extends well beyond basic fibre connectivity. The company has positioned itself as a full-service technology partner for large organisations, offering a portfolio that includes software-defined wide-area networking (SD-WAN), cybersecurity services, unified communications and collaboration tools, cloud migration support, and managed IT services. On the cloud partnership front, Liquid is a Microsoft Cloud Solution Provider and has partnerships with AWS, Google Cloud, and Oracle Cloud. For Kenyan data centre customers, these partnerships affect how hybrid cloud architectures are built. An enterprise running workloads in a Nairobi colocation facility can connect to Microsoft Azure, for example, over Liquid's network with lower latency and more predictable performance than traversing the public internet. Liquid's cybersecurity offerings are also relevant to the data centre conversation. As Kenyan organisations digitise and move critical workloads into data centre environments, the security of connectivity between those environments and the outside world becomes paramount. Liquid provides next-generation firewall services, distributed denial-of-service (DDoS) protection, and secure access service edge (SASE) solutions that complement the physical security measures within data centre facilities. ## How Liquid Differs from Pure-Play Data Centre Operators Understanding Liquid's positioning requires distinguishing between connectivity-first companies and pure-play data centre operators. Companies like Africa Data Centres, PAIX, iXAfrica, and Wingu Africa are fundamentally in the business of building and operating physical data centre facilities. Their core product is space, power, cooling, and physical security. Liquid, by contrast, is fundamentally in the business of moving data. Their core product is bandwidth, connectivity, and network services. The data centre assets they operate through Liquid Datapoint exist primarily to support and extend their connectivity and enterprise services offering. This distinction has practical implications. When a pure-play operator builds a new data centre, the investment thesis is based on filling that facility with colocation customers. When Liquid extends its network, the thesis is broader, encompassing increased network utilisation, new enterprise service contracts, and the strategic benefit of controlling more of the end-to-end connectivity path. For customers, this means Liquid's data centre offering appeals most to organisations wanting an integrated technology partner. Banks, telecoms operators, government agencies, and large enterprises that buy connectivity, cloud, and security services from Liquid will naturally find Liquid Datapoint's integrated data centre services compelling. Startups and cloud-native companies that prefer to mix and match connectivity providers may be better served by carrier-neutral facilities. ## East Africa Strategy and Kenya's Hub Role Liquid's East Africa strategy positions Kenya as the regional hub for both connectivity and data centre services. Kenya has the largest economy in East Africa, the most developed submarine cable infrastructure, the most mature internet exchange point (KIXP), and the deepest pool of technology talent in the region. From Liquid's perspective, Kenya serves as the anchor for their East African network topology. The fibre routes connecting Mombasa's landing stations to Nairobi and onwards to Uganda, Rwanda, and beyond create a hub-and-spoke pattern with Nairobi at the centre. Data centre facilities in Nairobi benefit because they sit at the junction where international bandwidth meets regional distribution networks. Liquid has invested heavily in network reliability and redundancy in Kenya, maintaining diverse fibre routes between Mombasa and Nairobi. This redundancy is a prerequisite for the uptime guarantees that enterprise and hyperscale customers demand, and it is one reason Nairobi has emerged as the preferred data centre location in East Africa. As enterprise customers demand more sophisticated technology solutions, including AI and machine learning workloads, Liquid is positioning itself to provide the high-bandwidth, low-latency connectivity these workloads require, as well as the hybrid cloud infrastructure to support them. ## Future Plans and Outlook Looking ahead, Liquid's trajectory in Kenya and East Africa will be shaped by continued growth in data demand driven by cloud adoption, mobile internet usage, and increasingly by AI workloads. Liquid has signalled ambitions to expand both the geographic reach and capacity of its fibre network, with additional routes for redundancy and higher-capacity links to support growing bandwidth demands. The relationship between Liquid and Africa Data Centres will also continue to evolve. As Cassava Technologies matures as a group, there may be greater coordination between the connectivity and data centre businesses, potentially offering customers more seamless integrated solutions that combine carrier-neutral colocation with Liquid's network and enterprise services. Kenya's regulatory environment, managed by the Communications Authority of Kenya, will shape Liquid's investment decisions. The government's digital transformation agenda, including Konza Technopolis and the national broadband strategy, aligns with Liquid's infrastructure-first approach and could create opportunities for public-private partnerships. The emergence of AI as a driver of data centre demand could further elevate Liquid's role. AI training and inference workloads require massive bandwidth and low-latency connectivity, exactly the kind of infrastructure that Liquid provides. If Kenya becomes a meaningful location for AI infrastructure in Africa, Liquid's fibre network will be the critical enabler connecting that infrastructure to data sources, cloud platforms, and end users across the continent. ## Why Liquid Matters for Kenya's Data Centre Industry Liquid Intelligent Technologies may not be a data centre company in the traditional sense, but no serious analysis of Kenya's data centre landscape can ignore them. They own and operate the terrestrial fibre infrastructure that makes data centres in Nairobi viable. They connect Kenya to the global internet through Mombasa's submarine cables and to the regional market through cross-border fibre links. Through Liquid Datapoint, they operate their own data centre facilities serving customers who want integrated solutions. For organisations evaluating data centre and connectivity options in Kenya, understanding Liquid's role is essential. Whether you end up colocating in an Africa Data Centres facility with Liquid as your connectivity provider, placing your infrastructure directly with Liquid Datapoint, or choosing an entirely different path, Liquid's network is almost certainly part of the infrastructure chain that connects your data to the world. In Kenya's data centre ecosystem, Liquid Intelligent Technologies is not just another operator. They are the backbone. To explore data centre facilities and connectivity providers across Kenya, visit our [data centre directory](/directory). For a deeper look at the terrestrial infrastructure connecting Kenya's data centres, see our guide to [fibre optic networks in Kenya](/articles/fibre-optic-networks-kenya-data-centres). To understand the submarine cable systems that deliver international bandwidth to Kenya's shores, read our overview of [submarine cables in Mombasa](/articles/submarine-cables-landing-mombasa). ## Machine Learning in Kenyan Banking: The Data Centre Link URL: https://data-centers-254.vercel.app/articles/machine-learning-kenyan-banking-fintech Cluster: AI & Finance Published: 2026-08-28 Updated: 2026-08-28 Summary: Kenyan banks and fintechs use machine learning for fraud detection, credit scoring, and analytics. The infrastructure behind the models. When a customer sends KES 50,000 via M-Pesa at 2 AM from a location they have never used before, and the transaction is completed in under two seconds, there is a good chance that a machine learning model has already analysed the transaction in real time, checking the amount against the customer's typical patterns, the location against their known locations, the recipient against their transaction history, and decided that the transaction is legitimate. This analysis happens in milliseconds, on servers in a data centre, before the customer even sees the confirmation message. ![Server room power and battery infrastructure in a Kenyan data centre](/images/technician-ups-battery-room.webp) Machine learning (ML) is transforming Kenya's banking and financial technology sector from the inside. It is not a customer-facing feature that users see or interact with directly. Instead, it operates behind the scenes (in fraud detection systems, credit scoring engines, customer analytics platforms, and operational automation tools) making decisions that affect millions of transactions every day. And every one of these ML systems runs on servers in data centres, creating a growing and often overlooked demand for data centre infrastructure. ## The ML Landscape in Kenyan Financial Services Kenya's financial services sector is one of the most technologically advanced in Africa, and machine learning adoption reflects this. The sector's ML deployment can be categorised into several use cases, each with different infrastructure requirements. ### Fraud Detection and Prevention Fraud detection is the most mature and widely deployed ML application in Kenyan banking. Every bank and mobile money operator in Kenya uses ML models to analyse transactions in real time, looking for patterns that indicate fraud, money laundering, or account takeovers. The ML models used for fraud detection are typically trained on historical transaction data, millions or billions of past transactions labelled as legitimate or fraudulent. The model learns patterns: a customer who normally sends KES 1,000-5,000 to family members suddenly sending KES 500,000 to an unknown account at 3 AM triggers a high-risk score. A new device logging into an account from an unusual location triggers additional authentication. Real-time fraud detection requires ML models to score every transaction in milliseconds, the model must return a risk score before the transaction is approved or declined. This requirement for low-latency inference means the models must run on servers with fast processors, preferably in data centres close to the banking systems that generate the transactions. Hosting fraud detection models on cloud servers in South Africa or Europe adds 50-200ms of latency, which may be acceptable for batch processing but is problematic for real-time transaction scoring. ### Credit Scoring and Lending Credit scoring is the ML application with the most transformative potential for Kenya's financial inclusion. Traditional credit scoring relies on formal financial records (bank statements, loan repayment history, employment verification) that exclude millions of Kenyans who are unbanked or underbanked. ML-based credit scoring uses alternative data sources to assess creditworthiness. M-Pesa transaction history (frequency of transactions, amounts, consistency of income flows), airtime purchase patterns, utility payment records, social media activity (with consent), and device data can all be used to build credit models that serve people who have no formal credit history. Several Kenyan fintechs have built their businesses around this approach. Companies like Tala, Branch, and Silba have disbursed millions of small loans to previously unbanked Kenyans, using ML models that analyse mobile phone data to make lending decisions. These models are trained on large datasets (millions of past loans with known repayment outcomes) and are continuously refined as new data becomes available. ### Customer Analytics and Personalisation Banks use ML to segment their customer base and personalise products and services. By analysing transaction patterns, channel preferences, and product usage, ML models can predict which customers are likely to be interested in a new product, which are at risk of churning, and what products would best serve each customer's needs. These models typically run in batch mode (processing data overnight or periodically rather than in real time) and therefore have less stringent latency requirements than fraud detection models. They can run on standard CPU servers and do not necessarily require GPU infrastructure. However, they do require access to large datasets and significant data processing capacity. ## Data Centre Requirements for ML Workloads ### Training Infrastructure ![Engineer working at server racks in a Kenyan facility](/images/technician-rack-cabling-kenya.webp) Training an ML model is the most computationally intensive phase. It involves processing large datasets (often terabytes of transaction history) through multiple iterations of the model to optimise its parameters. For simple models (like logistic regression for credit scoring), training can be done on standard CPU servers in hours. For complex models (like deep neural networks for fraud detection), training may require GPU servers and take days or weeks. The data centre infrastructure required for training depends on the model complexity. For most Kenyan banking ML applications, a cluster of 4-8 GPU servers (using NVIDIA T4 or A100 GPUs) is sufficient. For frontier models (like large language models), the requirements are much larger, but these are not typically used in banking applications. Training workloads are typically batch jobs, they run for hours or days and then stop. This means the infrastructure can be shared among multiple teams and multiple models. Some organisations train models in the cloud (using AWS, Azure, or Google Cloud GPU instances) and deploy them in local data centres for inference, combining the cloud's flexibility with local hosting's compliance advantages. ### Inference Infrastructure Inference (using a trained model to make predictions) is the ongoing, production workload. Every time a customer makes a transaction, the fraud detection model must score it. Every time a loan applicant submits an application, the credit scoring model must assess it. These inference requests happen continuously, 24 hours a day. For most banking ML models, inference can run on standard CPU servers. A well-optimised credit scoring model can score a loan application in under 100 milliseconds on a modern CPU server. Real-time fraud detection, which must handle thousands of transactions per second, may benefit from GPU acceleration but can also be achieved with CPU-optimised models. The key inference infrastructure requirement is not raw compute power but reliability and low latency. Inference servers must be available 24/7, must respond within tight latency budgets, and must be deployed close to the systems that generate the requests. This is why inference models for Kenyan banking are typically hosted in Kenyan data centres rather than in the cloud. ### Data Infrastructure ML models are only as good as the data they are trained on. Kenyan banks and fintechs must store and process large volumes of transaction data, customer data, and alternative data sources. This data must be stored securely (complying with the [Data Protection Act](/articles/kenya-data-protection-act-data-centres)), organised for efficient access, and available for training pipelines. Data infrastructure for ML includes data lakes (large-scale storage for raw data), data warehouses (organised storage for structured data), feature stores (storage for pre-computed ML features), and data pipelines (automated workflows that move data from source systems to ML training environments). All of this infrastructure runs in data centres. ## The ML Talent Challenge ![Partnership signing in front of Oracle and iXAfrica DataCentre banners](/images/oracle-ixafrica-partnership.webp) Kenya produces many software developers, but the specific skills required for ML engineering, statistical modelling, feature engineering, model optimisation, MLOps (the practice of deploying and managing ML models in production), are in short supply. The Microsoft Africa Development Centre in Nairobi employs ML engineers working on global products, and some of these skills are transferring to the local ecosystem. Strathmore University, the University of Nairobi, and other institutions have introduced data science and ML programmes, but the pipeline is still thin relative to demand. For banks and fintechs, this skills gap means that ML adoption requires either hiring from a limited talent pool (driving up salaries), training existing staff (which takes time), or partnering with specialised ML companies (which adds cost and dependency). Some organisations address the gap by using cloud-based ML platforms (AWS SageMaker, Azure Machine Learning, Google Vertex AI) that reduce the operational complexity of ML deployment. ## The Regulatory Environment The Central Bank of Kenya (CBK) has been cautiously supportive of technology adoption in financial services, including AI and ML. The CBK's guidance on technology in banking emphasises risk management, data protection, and the need for human oversight of automated decisions. Key regulatory considerations for ML in Kenyan banking include: **Explainability**: When an ML model declines a loan application, the customer has a right to know why. The CBK may require that ML-based decisions can be explained in terms that customers and regulators understand. This is straightforward for simple models (like decision trees or logistic regression) but challenging for complex models (like deep neural networks). **Data protection**: ML models trained on personal data must comply with the Data Protection Act. This includes obtaining appropriate consent, ensuring data is used only for stated purposes, and implementing security measures to protect the training data. **Fairness and bias**: ML models can inadvertently discriminate against certain groups if the training data reflects historical biases. Ensuring that credit scoring models do not discriminate based on gender, ethnicity, or other protected characteristics is both a regulatory requirement and an ethical imperative. ## The Future Machine learning in Kenyan banking and fintech is still in its early stages. The most mature applications (fraud detection, basic credit scoring) are well-established. The next wave will include more sophisticated applications: natural language processing for customer service (chatbots that understand Swahili and Sheng), computer vision for document verification (automating KYC processes), and predictive analytics for financial planning. Each of these applications requires data centre infrastructure, for training, inference, and data storage. As ML becomes more deeply embedded in Kenya's financial services, the data centre infrastructure that supports it will become more critical, more specialised, and more valuable. The banks and fintechs that invest in ML today are also, indirectly, investing in the data centre infrastructure that makes it possible. ## MEA Colocation Market Surges to $12.88 Billion by 2031 URL: https://data-centers-254.vercel.app/articles/mea-colocation-market-surge-2031 Cluster: Market Analysis Published: 2026-09-04 Updated: 2026-09-04 Summary: Arizton values the Middle East and Africa colocation market at $3.33bn in 2025, rising to $12.88bn by 2031. Where Nairobi fits in. The Middle East & Africa data centre colocation market was valued at USD 3.33 billion in 2025, and Arizton Advisory & Intelligence expects it to reach USD 12.88 billion by 2031, a compound annual growth rate of 25.32 percent that is hard to ignore from any vantage point on the continent. Behind the headline sits a familiar stack of drivers: enterprises accelerating cloud adoption, the first serious wave of AI deployments, smart city programmes across the Gulf and Africa, and governments treating digital infrastructure as economic policy rather than procurement. Colocation, in this story, is the pragmatic middle path, the way to get compute capacity quickly without writing a nine-figure cheque for a private facility. For Kenya's data centre community, the report is more than a regional cheerleading exercise. It quantifies the construction-cost pressures every operator in Nairobi already feels, ranks the markets Kenya competes against for mobile capital, and confirms something our [facility directory](/directory) has tracked for a while: Nairobi is East Africa's centre of gravity, with eight operational data centres and seven more in the pipeline as of December 2025. This article unpacks the report's key numbers, maps where the investment is concentrating, and asks the question that matters for our readers: when the regional market nearly quadruples in six years, how much of that wave can Kenya realistically catch? ## The Numbers Behind the 25.32% CAGR Arizton's central finding is that colocation in MEA is growing because it solves the region's core infrastructure dilemma. Enterprises and governments want digital capacity now, but building private data centre facilities means confronting land scarcity, elevated labour costs, inflation, and power access, the report explicitly cites these as the reasons construction runs between USD 6 million and USD 12 million per megawatt depending on location. Colocation converts that hostile capex environment into a monthly operating expense, which is why demand keeps compounding even as headline interest rates stay uncomfortable. The growth is underpinned by policy as much as by demand. Economic diversification strategies (Saudi Arabia's Vision 2030 being the loudest, but similar playbooks across the Gulf and in African capitals) treat data centre capacity as a sovereign capability. Rising internet traffic, expanding smart city projects, and favourable regulatory frameworks do the rest. The report also flags the counterweight honestly: geopolitical conflict and regional instability can delay projects through supply chain disruption, material shortages, and labour constraints. Paradoxically, the same instability pushes investment toward the region's more stable markets, which is precisely the bracket Kenya markets itself in. ![Rows of server racks inside a colocation data hall](/images/hero-server-hall.webp) One segment detail deserves attention from buyers. Despite the sustainability conversation, diesel generators remain the backbone of resilience across the region, Africa Data Centres' Lagos facility, for example, maintains 48 hours of diesel backup at full load. The report notes newer UPS chemistries such as Nickel-Zinc and Sodium-Ion batteries gaining traction on power density and safety, and confirms that the 42U rack remains the workhorse while 45U–48U cabinets appear in large environments. For anyone procuring colocation in Nairobi, those are the rack-level realities behind the brochure. ## Where the Money Is Concentrating The regional split inside "MEA" is stark, and Kenya needs to read it with clear eyes. The Middle East (led by large-scale developments in Saudi Arabia and the UAE) is expected to account for the significant majority of regional investment, propelled by government-backed digital transformation, smart city spending, and hyperscale cloud onshoring. Saudi Arabia's market is described as being in a high-growth, investment-heavy phase evolving into an AI-driven digital infrastructure hub; the UAE ranks among the most advanced markets in the region on the strength of connectivity and a well-defined regulatory framework. Africa's pockets of scale are real but narrower. Nigeria hosts around 20 colocation data centres, the majority concentrated in Lagos, and the pipeline keeps deepening, with Open Access Data Centres announcing a USD 240 million, 24MW two-stage Lagos facility in March 2025, the first 12MW targeted for 2026. South Africa anchors the southern end of the continent with roughly 11 operational submarine cables as of November 2025, and drew a strategic marker in July 2025 when NVIDIA and Cassava Technologies signed a partnership to build AI-ready data centres across Egypt, Nigeria, Kenya, and Morocco with around USD 700 million attached. Egypt is flagged as a market where foreign operators are beginning to enter through local partnerships. Even the Gulf's second-tier markets are moving. In August 2025, Ooredoo inaugurated a data centre and cable landing station in Salalah, Oman (starting with 125 racks and planning 500) explicitly designed to host the 2Africa subsea cable. That single project bundles three of the region's repeating patterns into one: colocation capacity, cable-driven connectivity, and state-adjacent capital. ## Nairobi's Position in the Regional Map Against that backdrop, Nairobi's numbers hold up well for an East African market. Arizton identifies Nairobi as the top data centre destination in Kenya, with eight existing and seven upcoming facilities as of December 2025. Our own tracking bears this out: the [DC254 directory](/directory) profiles the operators behind those facilities, from [Africa Data Centres](/articles/africa-data-centres-kenya-operations) and [Liquid Intelligent Technologies](/articles/liquid-intelligent-technologies-kenya) to [iXAfrica](/articles/ixafrica-data-centres-kenya), whose Nairobi campus markets itself as the region's first hyperscale AI-ready facility, and the [comparison tool](/directory/compare) makes the capacity, power, and certification differences between them searchable. ![Nairobi's skyline under bright daylight](/images/nairobi-skyline.webp) What gives Nairobi a differentiated pitch in a 25-percent-growth region is the combination Kenya keeps underselling. The grid is over 90 percent renewable, anchored by geothermal baseload, a structural carbon story that Gulf competitors spending billions on solar and storage cannot yet replicate. The licensing framework is explicit, with NFP-Tier 1 and Tier 2 categories that operators can plan against. And the connectivity case keeps strengthening: submarine cables land in Mombasa, and terrestrial fibre routes the traffic up the Northern Corridor to Nairobi, a geography we examined in detail when explaining [why data centres cluster in Nairobi and Mombasa](/articles/why-data-centres-cluster-nairobi-mombasa). The honest caveats sit in the report's cost line. At USD 6–12 million per megawatt region-wide, and with [Kenya Power industrial tariffs](/articles/kenya-power-tariffs-data-centres) representing the largest recurring operating cost, Nairobi cannot win on cheap construction or cheap power alone. It wins (when it wins) on stability, carbon, talent, and the depth of its interconnection ecosystem via [KIXP](/articles/kixp-internet-exchange-point-kenya). ## What the Vendor Landscape Tells Kenyan Buyers The report's vendor roll-call reads like the tenant list of any Nairobi facility's cross-connect room: Africa Data Centres, Equinix, Digital Realty, Vantage Data Centers, EdgeConneX, Raxio Group, Rack Centre, Orange Business, and carrier-neutral and regional names across the Gulf. New entrants (Agility Logistics Parks, DataVolt, Kasi Cloud, Mega DC, and more than a dozen others) signal that capital still believes there is room at the table. Two data points matter most for East African readers. First, DataVolt's planned Neom development with around 1.5 GW of IT power capacity is a useful reminder of the scale gap: Kenya's entire operational market is a rounding error against single Gulf campuses being scoped today. Second, Sahayeb's two Riyadh and Dammam facilities (about 60 MW combined across 600,000 square feet) show what a mid-sized, state-aligned build looks like. Kenya's realistic strategy is not to out-build the Gulf but to capture the workloads that need Kenya specifically: East African latency, sovereign localisation under the [Kenya Data Protection Act](/articles/kenya-data-protection-act-data-centres), and carbon-conscious capacity. ## The Kenya Takeaway A market growing at 25.32 percent a year does not guarantee any individual country a share of the growth, it guarantees competition for it. The report's own segmentation points to where Kenya should play. Colocation remains the on-ramp for enterprises digitalising without building, which favours Nairobi's established operators and the seven facilities still in the pipeline. AI is named as a demand driver, but the near-term African opportunity is inference and enterprise AI services rather than training campuses, a distinction we unpacked in our [AI data centres in East Africa](/articles/ai-data-centres-east-africa) analysis. And the sustainability trend runs in Kenya's favour: the report highlights renewable procurement across the region, from Kenya's September 2025 clean energy policy targeting universal green power access by 2030 to Egypt's 12 GW renewable target for 2026, context that makes our earlier piece on [Kenya's renewable advantage](/articles/kenya-renewables-industrial-power-data-centres) look conservative rather than promotional. ![Network cabling and switching equipment in a data centre](/images/dc-networking.webp) The watch items for the rest of 2026 are straightforward. Whether the seven upcoming Nairobi facilities hold their timelines against global equipment lead times. Whether Kenya's grid connection queues can keep pace with the pipeline, given that the report ranks power availability as the region's structural constraint. And whether the NVIDIA–Cassava programme materialises its Kenya commitments on the schedule its announcement implied. If those three break well, Nairobi's share of a $12.88 billion regional market will be bigger in 2031 than the pessimists assume today. ## Meta's Petal Cable: 1 Pbps and What It Means for Kenya URL: https://data-centers-254.vercel.app/articles/meta-petal-petabit-subsea-cable-kenya Cluster: Connectivity Published: 2026-09-22 Updated: 2026-09-22 Summary: Meta's Petal cable will carry 1 petabit per second across the Atlantic by 2029. It doesn't land in Africa: here's why it still changes Kenya's connectivity math. On 21 September 2026, Meta announced Petal, the first subsea cable built to deliver petabit capacity (1,000 terabits per second) across an ocean. The system will span roughly 7,000 km between the United States and France, and Meta says it doubles the capacity of today's most advanced transoceanic cables, making it the largest single generational jump in transoceanic capacity ever deployed. Construction brings together NEC as the subsea system specialist, Sumitomo Electric Industries as the fibre supplier, and Orange, which will land the system on France's Atlantic coast. Meta expects Petal to be in service in 2029. So let's deal with the obvious question first, because credibility depends on it: Petal does not land in Kenya, and it does not land anywhere in Africa. It is a transatlantic trunk between two of the world's densest internet hubs. But dismissing it as irrelevant to East Africa would be a mistake, for three reasons that have nothing to do with hype: a transit path that ends in America, a capital pattern Meta has already applied to Kenya once, and a fibre technology that will eventually reshape what African cable systems can carry. ## What Meta actually announced The headline number is capacity. Where traditional transoceanic cables carry terabits per second, Petal is designed to deliver 1 petabit per second, by Meta's own illustration, roughly the network capacity required for 75% of the world's population to stream music at the same time. That is not an incremental upgrade; it is the largest generational increase in transoceanic capacity of any system, ever, according to Meta. For context, 2Africa, the Meta-led consortium system that has been live at Mombasa and Mtwapa since June 2024, carries a design capacity of 180 Tbps. Petal is more than five times that on a single span a fraction of the length. ![Animated chart of Meta's transatlantic cable capacity step-changes](/images/meta-petal-transatlantic-step-changes.gif) The engineering story behind the number matters just as much. Petal will be the first cable to deploy multi-core fibre technology at transoceanic distances, carrying traffic in two separate cores inside the same cable. The practical consequence: twice the data without a proportional increase in fibre, physical materials, or power. Meta points to its engineering blog for the technical deep-dive, but the headline is simple: the industry has found a way to multiply capacity per cable rather than laying more of them. It is also worth reading Petal as what it really is: a story about the stack that carries AI, cloud and digital services, where compute demand fills data centres, data centres fill networks, and networks ride subsea cables between continents. Petal is the bottom of that stack getting a generational upgrade. ![Meta animation of the transition from single-core to two-core fibre](/images/meta-petal-two-core-fibre.mp4) ## Petal does not land in Africa: here's why it still matters to Kenya **First, the transit path.** Nearly all Kenyan traffic headed for the Americas already rides a two-part journey: north from Mombasa across one of the [seven live cable systems](/articles/submarine-cables-landing-mombasa) to Europe, then west across the Atlantic to the United States. Petal upgrades the second half of that journey. The far end of Kenya's route to the Americas is being rebuilt a full capacity generation ahead of demand, which is what keeps end-to-end performance stable as video, cloud and AI workloads grow. Petal is not capacity Kenyan operators can buy (hyperscaler cables largely serve their own internal traffic), but its arrival is the clearest signal of where the trunk layer above East Africa is heading. ![Animated map of Meta's transatlantic cable routes between the United States and France](/images/meta-petal-route-map.gif) **Second, the capital pattern.** Petal is Meta's 20th-plus subsea investment, following Project Waterworth, a planned 50,000 km system announced in 2025 to connect the US, India, Brazil and South Africa. Google made the same transatlantic move in July 2025 with Sol, linking the US, Bermuda, the Azores and Spain. The pattern is consistent: hyperscalers fund cable generations where their compute demand sits. Meta has already applied that logic to Kenya once, by leading the 2Africa consortium to Mombasa. Every announcement like Petal makes the next hyperscaler decision about African routes more likely, not less. ![Diver inspecting a submarine cable on the seabed](/images/submarine-cable-diver.webp) **Third, the technology benchmark.** The two-core fibre that makes Petal possible is the same technology direction that will eventually reach regional systems. Today Kenya's aggregate international bandwidth exceeds 20 Tbps of lit capacity, served by seven live systems at Mombasa. When multi-core designs mature into commercially deployable regional systems, the arithmetic changes: operators could multiply capacity on existing routes without new beach landings, new permits, or new marine surveys. That is a future-looking consequence, not a present fact, but it is exactly the kind of shift that determines whether Mombasa's infrastructure keeps pace with [Kenya's data centre build-out](/directory). ## What this changes at Mombasa: nothing, yet It is worth stating plainly what Petal does not do. It does not add redundancy to East Africa's Indian Ocean corridor, where the live-system count and route diversity through Mombasa remain the region's real resilience story. It does not change Africa-1's status (landed and awaiting full service), which remains the most consequential near-term cable development for Kenya. And it does not alter the terrestrial picture, where the [fibre routes connecting Mombasa's landing stations to Nairobi's data centres](/articles/fibre-optic-networks-kenya-data-centres), and the peering at the [Kenya Internet Exchange Point](/articles/kixp-internet-exchange-point-kenya) that keeps local traffic local, still determine most of the latency Kenyan users actually experience. The [Berbera northern route](/articles/berbera-northern-route-kenya-fibre-redundancy), for operators diversifying beyond the Mombasa corridor, is likewise unaffected. ![Cable landing infrastructure on the Kenyan coast at Mombasa](/images/mombasa-cable-landing-3.webp) That honesty is the point. The global cable market now moves in hyperscaler generations, and African readers deserve coverage that distinguishes between systems that touch their shores and systems that shape their transit prices. Petal is firmly the second category: a benchmark to plan against, not a landing to celebrate. ## What to watch next Three markers will tell Kenyan infrastructure watchers whether Petal's announcement is a turning point or a headline. Watch 2029: if Petal enters service on schedule with its stated capacity, the transatlantic trunk layer doubles, and the per-bit economics of Kenya-to-Americas transit improve at the margin as upstream markets densify. Watch Africa-1's full-service launch at Mombasa, which would take the region from seven live systems to eight and is the closest, most tangible cable development on Kenya's horizon. And watch the fibre technology itself: the moment any consortium announces multi-core designs for an African route, whether a 2Africa successor or a new regional system, the Petal benchmark will have arrived on our shores. --- Meta's Petal cable is a transatlantic story with a Kenyan tail. The system itself will never touch Mombasa's waters, but the capital, technology and capacity trajectory it represents will shape every cable decision that does, and for a region whose digital economy is growing faster than its lit bandwidth, that trajectory is the news. ## Microsoft and G42 in Kenya: The $1B Story So Far URL: https://data-centers-254.vercel.app/articles/microsoft-g42-kenya-data-centre Cluster: AI & Infrastructure Published: 2026-09-16 Updated: 2026-09-16 Summary: Microsoft and G42 promised Kenya a $1 billion data centre and a new cloud region. What was announced, why it stalled in 2026, and what comes next. Two years ago, Kenya looked set to land the biggest technology investment in its history. On 22 May 2024, during the season of President Ruto's Washington state visit, Microsoft and the Abu Dhabi AI company G42 announced a $1 billion digital initiative for Kenya (Microsoft News, 22 May 2024). The headline item was a data centre: a geothermal-powered campus developed with local operator EcoCloud, pitched at 100 megawatts in its first phase and expandable towards 1 gigawatt (Data Center Dynamics, 22 May 2024). The quieter item was arguably bigger: a new East Africa cloud region for Azure, Microsoft's first on this side of the continent. By May 2026, the project had stalled so completely that Kenya formally suspended the plan (ThinkGeoEnergy, 8 May 2026). This article walks through what was actually promised, why the arithmetic failed, and what the wreckage teaches every data centre project now on the table, from EcoCloud's own build at Olkaria to the $1.5 billion off-grid Amaco proposal in Mombasa. ![Rows of GPU compute racks inside a hyperscale data centre hall](/images/dc-gpu-cluster-4-wide.webp) ## The deal, piece by piece The $1 billion was never one building. Read the original announcement carefully and it is an ecosystem play with three connected parts (Microsoft News, 22 May 2024). **Part one: the campus.** G42 and EcoCloud would build a data centre campus running on geothermal power, in the direction of Naivasha and the Olkaria steam fields, about 100 kilometres northwest of Nairobi. The scale was serious: 100 megawatts of first-phase capacity, expandable to 1 gigawatt, with phase one reportedly carrying 24 megawatts of IT load across eight 3-megawatt data halls, each holding 300 to 500 racks (Data Center Dynamics, 22 May 2024). At announcement, the first phase was expected to reach operational status in about two years, which put the target around mid-2026 (Data Center Knowledge, May 2024). **Part two: the cloud region.** Microsoft and G42 would design and operate a new East Africa cloud region for Azure together with the Kenyan government, inside what the announcement called a "trusted data zone". The point of that framing was sovereignty: government workloads and regulated data would stay onshore, under Kenyan law, with governance arrangements both parties could defend. Today, a Kenyan enterprise using Azure writes its traffic to Europe or South Africa. A local region changes latency, cost and compliance at a stroke. **Part three: demand.** The Kenyan government committed to move more of its services onto the cloud (Gulf News, 22 May 2024). That is the piece hyperscalers insist on: a cloud region without anchor tenants is an empty building. The government was to be that anchor. ![Industrial switchgear cabinets feeding a large data centre power room](/images/dc-switchgear-2.webp) ## How the partnership came together The Kenya announcement did not appear in isolation. On 16 April 2024, Microsoft had invested $1.5 billion in G42 itself, taking a minority stake in the Abu Dhabi company and putting Microsoft President Brad Smith on its board (Microsoft News, 16 April 2024). The investment was widely read as part of a technology diplomacy balancing act between Washington and the Gulf, and it made G42 Microsoft's vehicle for expansion in regions where Microsoft wanted a local partner with government relationships. Kenya was the first big test of that model. On the ground, the local pieces were already moving. EcoCloud, a Kenyan developer, had broken ground in 2023 on its own Project Eagle data centre at the KenGen Green Energy Park in Olkaria, inside Africa's largest geothermal complex. When the Microsoft and G42 initiative landed, EcoCloud became the local partner for the flagship campus, with Sudlows Consulting and X2X Group named as lead consultants (Data Center Dynamics, 22 May 2024). For a moment, Kenya had the full stack: a hyperscaler, a Gulf AI champion, a local developer, geothermal power and a government willing to commit demand. ## What went wrong: power and payments The stall came into public view in a single week of May 2026, and it had two strands. **The power arithmetic.** Speaking in Nairobi, President Ruto said the project as designed would have needed about a third of Kenya's roughly 3,000 megawatts of installed generating capacity. "To switch on that one data center, we would need to shut off power for half the country," he said (Semafor, 6 May 2026). The campus itself was a 100 megawatt IT load; the president's arithmetic described the generation and transmission build-out that a campus of that ambition would pull with it. Kenya's grid is genuinely strong by regional standards, but it cannot hand one site the load of a mid-sized city in 2026. **The paperwork trail.** Government officials told Semafor that the technology ministry developed a project concept note and took it to the National Treasury, which never approved the funding. A meeting between Kenyan officials and Microsoft executives in August 2025 had already made clear the original May 2026 online date would be missed (Semafor, 6 May 2026). By 8 May 2026, reports confirmed Kenya had suspended the planned $1 billion data centre (ThinkGeoEnergy, 8 May 2026). **The payment demand.** Days later, Bloomberg and Reuters added the commercial strand: the project had been delayed by disagreements between Microsoft's side and the Kenyan government over guaranteed payments the investors sought, and over the structure of the power arrangements, with a possible scale-back of the project under discussion (Bloomberg, 11 May 2026; Reuters, 10 May 2026). In plain terms: the investors wanted the state to underwrite a baseline of revenue and power, and the state, already strained, would not commit on those terms. ![Map of Kenya's geothermal fields in the Rift Valley](/images/kenya-geothermal-map.webp) ## What survives, and what replaced it The story does not end at the suspension, because the ingredients it tried to combine are still real. EcoCloud's own Project Eagle at Olkaria continues its development path, and The Star reported in September 2026 that construction activity at Olkaria and Konza is moving, even without the Microsoft flag on it (The Star, 14 September 2026). Kenya's cloud demand has not shrunk; if anything, AI workloads have raised the stakes. And the single most important lesson has already been absorbed by the next project in the queue: Amaco's proposed $1.5 billion HERCULES AI campus in Mombasa is designed to generate its own electricity from an LNG-fired power barge precisely because the national grid cannot carry a hyperscale load at one site. Whether or not that project is ever built, its power design is a direct answer to the exact problem that stalled Microsoft and G42. ## What it means for Kenya's data centre market Three things are worth holding onto. **The demand case was always real.** The project did not fail because Kenya lacks customers for cloud and AI compute. It failed because the power and financing structures did not close. That distinction matters when you read the next announcement: ask where the megawatts come from and who guarantees what, before asking how many racks it has. **A local Azure region remains the prize.** Until the East Africa cloud region lands in some form, Kenyan enterprises keep writing traffic to Europe and paying the latency and foreign-exchange cost of doing so. If the initiative returns at smaller scale, the cloud region is the piece Kenya should want first, because it is the part that changes the economics of every other part. **Match the load to the generation, or generate your own.** That is now the defining rule of Kenyan hyperscale projects. The Olkaria corridor still holds the resource; the 7,000 to 10,000 megawatts of geothermal potential have not moved. What has changed is the discipline: proposals that arrive with their own power plan, like Amaco's barge or smaller phased campuses that grow with the grid, are the ones that can actually clear the bar the Microsoft project could not. We track the project in our [Kenya data centre directory](/directory) as early stage with no confirmed site, and we will update that entry the moment anything changes. The physics has not changed, the resource has not changed, and the demand has not changed. Only the timetable has. ## MTN and Al Ashram's Africa Data Hub, Explained URL: https://data-centers-254.vercel.app/articles/mtn-africa-data-hub-ai-data-centres Cluster: Market Analysis Published: 2026-08-29 Updated: 2026-08-31 Summary: MTN Group and Dubai investor Tarek Al Ashram launched Africa Data Hub to build AI data centres, starting with 150MW in Nigeria and South Africa. Africa's largest mobile network operator is putting serious capital behind the continent's AI infrastructure gap. MTN Group, which operates across 16 African markets, has agreed with Dubai-based investor Tarek Al Ashram to develop AI-focused data centres through a new venture called Africa Data Hub Holding Ltd, according to Bloomberg reporting published on 27 August 2026 and formally confirmed by MTN in an announcement on 31 August 2026. The announcement is one of the most significant commitments yet to building AI-ready compute capacity on the continent, and it arrives at a moment when Africa accounts for less than 1% of the world's AI data centre capacity despite having its fastest-growing and youngest population. ![AI data centre hall with GPU racks and blue status lights](/images/africa-ai-data-hub-mtn.webp) For Kenya's data centre community, the story matters even though the first facilities will not be built here. It signals that pan-African operators and Gulf capital now view AI infrastructure as an investable, revenue-generating asset class on the continent, and it adds competitive pressure to the race that Kenya's own flagship projects, including the planned one-gigawatt Microsoft and G42 facility, have been running. This article was first published on 29 August 2026 and updated on 31 August 2026 after MTN formally confirmed the partnership structure. ## The Venture: Africa Data Hub Holding Ltd The new company sits at the intersection of two complementary strengths. On one side is MTN Group, the continent's largest mobile operator by scale, with network infrastructure, enterprise relationships, and land and power positions across its footprint. On the other is Tarek Al Ashram, a Dubai-based investor who co-founded Gulf Data Hub, a large-scale data centre platform backed by American private equity giant KKR & Co. Al Ashram is backing the African venture through his own investment firm. "We see significant potential to bring our experience in developing and operating large-scale data centre platforms to African markets," Al Ashram said in a response to queries, adding that Africa Data Hub will "help shape that growth" of the continent's digital economy. In his framing, Africa now presents a "similarly compelling opportunity, driven by strong underlying demand, rapidly expanding digital economies and an increasing need for resilient and scalable infrastructure" to what the Middle East experienced during its own data centre boom. That comparison is instructive. The Gulf states transformed themselves from data centre bystanders into a regional hub by pairing sovereign-backed capital with experienced operators and guaranteed anchor tenants. Africa Data Hub is an attempt to replicate that playbook with MTN as the anchor, and with an operator who has already done it once in the UAE leading delivery. ## Confirmed on 31 August: MTN Digital Infrastructure, Bayobab, and the Platform Structure The formal announcement, published by IT News Africa on 31 August 2026, added useful structural detail to the original reporting. The agreement was signed by MTN Digital Infrastructure, the group subsidiary that houses MTN's towers, fibre, and data centre assets, and the venture will operate through Africa Data Hub Holding Limited as the platform through which future digital infrastructure opportunities are developed and scaled, with an initial focus on South Africa and Nigeria. Two details stand out for infrastructure watchers. First, Bayobab, MTN's digital connectivity business, is not just a supplier but a shareholder in the partnership, and will provide open-access connectivity and go-to-market support across its pan-African network. Bundling wholesale fibre with data centre capacity into one platform mirrors the integrated land-and-power-and-fibre model that successful data centre platforms use to shorten time-to-revenue. Second, the announcement framed the venture explicitly as part of MTN's Ambition 2030 strategy, with group digital infrastructure CEO Mazen Mroué describing a platform "capable of supporting hyperscalers, cloud providers, enterprises and technology innovators across Africa", language that positions Africa Data Hub as wholesale infrastructure rather than an MTN-internal cloud play. Mroué's framing also clarified the demand thesis: "Africa's digital economy is entering a new era, driven by rapid advances in artificial intelligence, cloud adoption and digital innovation. Realising this opportunity requires more than connectivity. It requires the infrastructure platforms that will power the next generation of digital services, businesses and innovation." For Al Ashram, the pitch is experience transfer: having watched data centre platforms accelerate cloud and AI growth in the Middle East, he described Africa as presenting "a similarly compelling opportunity" driven by demand, expanding digital economies, and the need for resilient, scalable infrastructure. ## Inside the Buildout: 150MW, Nigeria and South Africa First The venture is starting deliberately narrow. MTN chief executive Ralph Mupita said the company is taking a phased approach and will begin by building out 150 megawatts of AI data centre capacity, with facilities planned in Nigeria and South Africa. Those two markets are logical first moves: Nigeria offers West Africa's largest economy and a fast-growing cloud and content market, while South Africa hosts the continent's deepest data centre ecosystem and grid-adjacent industrial land. The sites will serve a dual purpose. First, they will process MTN's own AI workflows, from network optimisation to customer analytics, giving the venture a guaranteed internal workload from day one. Second, they will rent capacity to hyperscalers, other enterprise businesses, and governments, a customer segment Mupita noted is "increasingly asking for more sovereignty over their data sets." That sovereignty demand is a recurring theme across African markets, where regulators and public institutions want data resident in-country but have historically had limited local options. MTN has also been doing unglamorous groundwork that experienced developers will recognise: purchasing land and negotiating power agreements to support a multi-phased buildout across the continent. Securing grid capacity and land banks ahead of demand is exactly how mature data centre platforms position themselves to move quickly when customers commit, and it suggests the 150MW figure is an opening position rather than a ceiling. ## Why MTN Is Getting Into Data Centres MTN's motivation is straightforward: telcos globally are watching AI shift value toward compute infrastructure, and they want a share of the income stream that hyperscalers are chasing. Connectivity alone captures a thin slice of AI-driven revenue. The physical facilities that house AI training and inference clusters, however, are generating some of the strongest returns in digital infrastructure, with occupancy and pricing holding up across major markets. Building also fits MTN's balance sheet strategy. The operator has spent years monetising non-core assets and repositioning around infrastructure, and data centres complement its fibre, towers, and mobile network assets. Partnering with an experienced external developer rather than building alone, as Mupita explained ("the entity that we are partnering with has built data centres in the UAE"), reduces execution risk and brings in capital and operational know-how that African telecom groups have often lacked in-house. ![Grid technician working on overhead power distribution lines](/images/grid-technician-pole-work.webp) AI workloads also change the engineering requirements dramatically. Training clusters demand power densities per rack that far exceed traditional enterprise colocation, along with advanced liquid cooling, high-capacity interconnects, and hardened power chains. Operators who can deliver AI-ready halls today are effectively selling scarce capacity into unmet demand, which is precisely the opportunity Africa Data Hub is chasing. ![Cutaway infographic of an AI data centre's systems](/images/whats-inside-ai-data-center.webp) ## Africa's AI Capacity Problem in Numbers The context behind this deal is a stark imbalance. According to World Economic Forum data cited in the Bloomberg report, Africa houses less than 1% of the world's AI data centre capacity even as it holds the world's fastest-growing and youngest population. Hyperscalers including Microsoft, Alibaba, and Amazon offer cloud services on the continent, but the large-scale AI training facilities that have reshaped other regions have largely not materialised here yet. That gap has consequences. African enterprises building AI products often train or host models overseas, exporting both spend and latency penalties. Governments crafting data localisation rules find limited in-country compute to localise into. And the continent's much-discussed AI talent pipeline lands in an environment with few large-scale facilities to work in, reinforcing the shortfall. Kenya illustrates both the ambition and the gap. The planned one-gigawatt Microsoft and G42 facility, announced with fanfare as part of a wider Kenya-UAE technology partnership, "has so far not materialised", as the Bloomberg report notes. Projects of that scale depend on patient capital, firm power supply, and anchor tenants aligning simultaneously, and the continent's track record of converting headline announcements into energised data halls remains thin. ## What This Means for Kenya MTN does not operate a mobile network in Kenya, and the first Africa Data Hub sites are earmarked for Nigeria and South Africa. So why should Kenyan readers care? Three reasons stand out. First, the deal validates the demand thesis that Kenya's own projects rest on. When Africa's largest operator commits capital to AI data centres, it signals that the underlying demand signals, sovereignty requirements, cloud growth, and enterprise AI adoption, are strong enough to underwrite multi-phase industrial buildouts. That is the same demand case that iXAfrica, Africa Data Centres, and other Kenyan operators make when expanding Nairobi capacity. Second, it sharpens competition for the hyperscaler investments Kenya has been courting. If Africa Data Hub delivers AI-ready capacity in Nigeria and South Africa ahead of Kenya's flagship projects, regional hyperscaler and enterprise workloads could consolidate in those markets first. Nairobi's counterweights remain real, including dense subsea connectivity, a strong talent base, and a policy environment actively courting digital investment, but timelines now matter more. Third, it creates a template Kenya's market could follow: a large anchor tenant pairing with a proven international operator and specialist capital. A similar structure around a Kenyan anchor, whether a bank, a telco-adjacent enterprise group, or a government workload, is exactly what would convert the country's pipeline of announced projects into steel, concrete, and energised capacity. Readers exploring the local landscape can compare operators in our [data centre directory](/directory) and our analysis of [AI data centres in East Africa](/articles/ai-data-centres-east-africa). ## From Announcement to Energised Halls: The Execution Question African infrastructure history counsels measured optimism. The continent has seen repeated cycles of ambitious digital infrastructure announcements, some transformative and some still on paper years later. The gap between a signed agreement and an operating AI data centre runs through land title, grid connections that can take years to secure, cooling and power engineering for high-density racks, anchor tenant contracts, and construction capacity, each a point where timelines slip. What differentiates credible projects is visible progress: land actually acquired, power agreements actually signed, and phases sized to match realistic demand. On those measures, MTN's groundwork, purchasing land and negotiating power agreements before the venture was announced, and the choice to start with 150MW rather than promising gigawatts, are encouraging signs of discipline. The involvement of an operator with Gulf Data Hub's delivery record adds further weight. The financing environment will matter too. AI infrastructure is capital-hungry, and African projects compete globally for it. A venture backed by an operator of MTN's scale and an investor with KKR-linked credentials is better positioned than most to attract institutional capital, but execution milestones, not press releases, will decide whether that capital follows. ## What to Watch Next Three markers will tell us whether Africa Data Hub becomes real. Watch for site-specific announcements in Nigeria and South Africa, including locations, grid arrangements, and cooling designs that reveal the density ambitions. Watch for anchor tenant disclosures beyond MTN's own workflows, particularly hyperscaler or government commitments that de-risk the first phase. And watch whether the venture's roadmap extends beyond its first two markets, with East Africa and Kenya a natural consideration in later phases given the region's connectivity and demand growth. For Kenya's data centre sector, the announcement is best read as both validation and a starting gun. The AI infrastructure race on the continent now has another well-funded runner, and the markets that move first on power, land, and anchor tenants will collect the workloads that follow. ## Where Should Kenya's Next Data Centre Be Built? URL: https://data-centers-254.vercel.app/articles/nairobi-vs-mombasa-data-centre-locations Cluster: Infrastructure Published: 2026-08-28 Updated: 2026-08-28 Summary: Nairobi hosts 80% of Kenya's data centre capacity. Mombasa, Kisumu, and Konza each offer advantages. The factors behind site selection. Kenya's data centre industry is geographically concentrated to an extraordinary degree. Approximately 80% of the country's total IT power capacity is located within a 15-kilometre radius in Nairobi, primarily along the Mombasa Road corridor. This concentration is not accidental, it reflects the economics of connectivity, demand, and infrastructure that make Nairobi the natural location for data-intensive facilities. But as the industry grows and new use cases emerge, the question of where the next data centre should be built is becoming increasingly relevant. ![Wide aerial view of the Port of Mombasa](/images/mombasa-port-wide.webp) Site selection for a data centre is a multi-million-dollar decision that cannot be easily reversed. Once a facility is built, its location is fixed for 15–25 years. The building, the power connections, the fibre routes, the cooling systems, all of it is purpose-built and immovable. Getting the location wrong means either accepting suboptimal performance and higher costs for the facility's entire life, or writing off the investment and building elsewhere. This is why site selection is arguably the most important decision in the data centre development process, and why it deserves rigorous analysis. ## Nairobi: The Established Hub Nairobi's dominance in Kenya's data centre industry is built on four pillars that are difficult to replicate elsewhere. ### Customer Proximity The most important factor in data centre site selection is proximity to customers. Data centres exist to serve the organisations that need computing, storage, and connectivity, and in Kenya, those organisations are overwhelmingly concentrated in Nairobi. The city is home to the headquarters of all major Kenyan banks (KCB, Equity, Co-operative Bank, Standard Chartered, Absa, NCBA), the largest telecom operators (Safaricom, Airtel, Telkom Kenya), international organisations (the United Nations Environment Programme, UN-Habitat, numerous diplomatic missions), and the central government ministries that are driving digitalisation. For colocation providers, this customer concentration means that a facility in Nairobi can attract customers from multiple sectors, reducing the risk of dependence on any single industry. For enterprise data centres built by banks or telecom companies, being in Nairobi means low latency to their own headquarters and branch networks. The proximity advantage is structural and self-reinforcing, as more data centres are built in Nairobi, more customers and partners locate nearby, which justifies further investment. ### Connectivity Nairobi's connectivity advantage is significant. Multiple fibre operators (Liquid Intelligent Technologies, Safaricom, Telkom Kenya, and Jamii Telecommunications) have dense metro fibre networks covering the city's commercial and industrial areas. The Mombasa Road corridor, where most data centres are located, is served by multiple diverse fibre routes from multiple providers, enabling the redundancy that enterprise customers require. Nairobi is also the primary hub for the [Kenya Internet Exchange Point (KIXP)](/articles/kixp-internet-exchange-point-kenya), which allows networks to exchange traffic locally rather than routing through South Africa or Europe. For data centres, KIXP presence means that traffic between Kenyan networks can be exchanged with sub-5ms latency, significantly improving performance for local users. ### Power Infrastructure Nairobi has the most reliable power supply in Kenya. While Kenya Power's grid reliability has faced challenges in some areas, Nairobi's central business districts, industrial areas along Mombasa Road, and Westlands generally experience fewer and shorter outages than other parts of the country. Kenya Power has dedicated feeders and higher reliability targets for industrial zones, and data centres supplement grid power with on-site generators and UPS systems. The cost of power in Nairobi is also favourable compared to many other African cities. Kenya's generation mix (dominated by geothermal energy from the Rift Valley) keeps electricity costs lower than in countries dependent on diesel or coal generation. Industrial tariffs in Nairobi range from KES 12 to KES 18 per kilowatt-hour, compared to KES 20–30 in some Nigerian cities or KES 15–22 in South African cities affected by load shedding. ### Workforce Data centres require a skilled workforce (network engineers, systems administrators, facility managers, cabling technicians, security personnel) and Nairobi has the largest pool of these skills in East Africa. The city's universities (University of Nairobi, Strathmore, Jomo Kenyatta University of Agriculture and Technology, Kenyatta University) produce thousands of IT and engineering graduates annually. While the specialised skills required for data centre operations are in short supply everywhere in Kenya, the probability of finding and recruiting qualified staff is highest in Nairobi. ### The Mombasa Road Corridor Within Nairobi, the Mombasa Road corridor (specifically the area around Enterprise Road, Sameer Business Park, and the adjacent industrial zones) has emerged as the primary data centre cluster. iXAfrica, Africa Data Centres, Safaricom, and several smaller operators all have facilities along this corridor. The reasons are practical: the area has good power infrastructure, is close to the Jomo Kenyatta International Airport, has ample land for expansion (compared to the CBD), and sits on the main fibre route between Nairobi and Mombasa. ## Mombasa: The Connectivity Play Mombasa's data centre potential rests on a single, powerful advantage: it is where the submarine cables land. SEACOM, TEAMS, EASSy, and DARE1 all have their Kenyan landing stations in or near Mombasa. A data centre in Mombasa could connect directly to these cable systems, achieving the lowest possible latency to international traffic, a genuine and measurable advantage for content delivery, financial trading, and any application where every millisecond of latency matters. ![Industrial port infrastructure at Mombasa](/images/mombasa-port-industrial.webp) However, Mombasa faces significant challenges as a general-purpose data centre location. The demand side is the most fundamental problem: Mombasa's economy is dominated by tourism, port logistics, and manufacturing, industries that generate less data centre demand than the financial services, technology, and government sectors concentrated in Nairobi. There are simply fewer potential customers in Mombasa who need colocation services. ![Nairobi skyline at dusk](/images/nairobi-skyline.webp) Power reliability in Mombasa, while adequate for most industrial purposes, does not match Nairobi's industrial zones. The coastal climate, while beneficial for natural cooling (Mombasa's ambient temperatures are warm but consistent, and sea breezes can potentially assist cooling), also brings humidity and salt air that increase corrosion risk for equipment. Security concerns, particularly following the 2014 Lamu attacks and periodic travel advisories for coastal Kenya, have made some international investors cautious about committing long-term infrastructure investments in the region. Mombasa's most realistic data centre role is as a content delivery point, a smaller facility that caches and serves popular content (streaming video, web content, software updates) locally for coastal and northern Tanzanian users, reducing the need for this traffic to travel to Nairobi and back. Several content delivery networks (CDNs) and cloud providers have established caching nodes in Mombasa for precisely this reason, though these are typically small deployments within existing telecom facilities rather than standalone data centres. ## Konza Technopolis: The Long-Term Vision Konza Technopolis, located approximately 60 kilometres south of Nairobi in Machakos County, is Kenya's most ambitious infrastructure project, a planned technology city intended to become "Africa's Silicon Savannah." The project, officially launched in 2008 and subsequently delayed multiple times, envisions a 5,000-acre city with residential areas, a university, research institutions, technology companies, and (critically) data centre infrastructure. Konza offers several theoretical advantages for data centre development. Land is available and relatively inexpensive compared to Nairobi's congested industrial areas. The city is planned with dedicated power infrastructure, including potentially direct connections to the geothermal generation plants in the Rift Valley. It sits on the Nairobi-Mombasa fibre corridor, ensuring connectivity. And if the city develops as planned, it would generate its own demand from technology companies, research institutions, and government digital services. ![Data centre rooftop chillers and generators](/images/dc-chillers-roof.webp) The challenge with Konza is timing. Data centres need customers, connectivity, and a skilled workforce from day one, not at some point in the future. As of 2025, Konza's development is well behind the original schedule. While some basic infrastructure (roads, water, power) is in place, the critical mass of technology companies, research institutions, and skilled workers that would justify a data centre investment does not yet exist. Konza may become a viable data centre location in the 2030s, but for current investment decisions, it is a long-term option rather than a practical choice. ## Secondary Cities: Kisumu, Nakuru, Eldoret Kenya's secondary cities have been discussed as potential data centre locations, driven by the concept of edge computing, placing smaller data processing facilities closer to end users to reduce latency. Kisumu, as the largest city in western Kenya and a gateway to the Lake Victoria basin (serving parts of Kenya, Uganda, and Tanzania), is the most frequently mentioned candidate. The case for Kisumu includes a growing business community, the expansion of fibre connectivity to western Kenya (by Liquid Telecom and the government's NOFBI programme), and the strategic importance of serving western Kenya and cross-border markets. However, the demand for commercial data centre services in Kisumu is currently very small. Most organisations in western Kenya that need data centre services are better served by connecting remotely to Nairobi facilities, where the ecosystem of providers, services, and skilled workers is far richer. Nakuru and Eldoret face similar demand constraints. Both are growing cities with improving connectivity, but neither has the enterprise base or the critical mass of technology companies that would support a commercial data centre today. Like Konza, these cities may become viable as edge computing grows and as Kenya's digital economy extends beyond Nairobi, but that is a future state, not the current reality. ## The Verdict For any data centre investment decision being made in Kenya today, Nairobi remains the clear choice for the vast majority of use cases. Its combination of customer proximity, connectivity, power reliability, and workforce availability is unmatched by any other Kenyan city. Within Nairobi, the Mombasa Road corridor offers the best combination of existing infrastructure, land availability, and cluster effects. Mombasa has a niche role for connectivity-optimised facilities, and Konza and secondary cities represent long-term options as the market develops. But for an investor deciding where to put money to work in 2025 or 2026, Nairobi is the answer, and the specific site within Nairobi should be chosen based on fibre route diversity, power feeder reliability, and proximity to the existing data centre cluster. ## Kenya Has the Steam. So Where Are the AI Data Centres? URL: https://data-centers-254.vercel.app/articles/olkaria-geothermal-ai-data-centre-race Cluster: Energy Published: 2026-10-16 Updated: 2026-10-17 Summary: Microsoft chose Olkaria's geothermal steam for a 100 MW AI campus in 2024. Two years on, the deal is stalled. The contract, the wires and the grid headroom explain why. On 22 May 2024, Microsoft and Dubai-based G42 announced a $1 billion digital ecosystem package for Kenya. Its centrepiece was a data centre campus at Olkaria, in the Rift Valley, run entirely on geothermal power. The 100 MW first phase was not in the announcement text; that figure entered the public record through project reporting carried by Data Center Dynamics, and it is the number DC254's directory carries as the announced capacity. The site choice was the whole pitch: Olkaria sits on one of the world's most productive geothermal fields, and geothermal generates through the night, which is precisely what an AI compute floor needs. Two years later, the campus is still an announcement. Semafor reported in May 2026 that the plan was being reconsidered, and quoted President William Ruto putting the facility's power appetite against the national system: "To switch on that one data center, we would need to shut off power for half the country. That's when I knew there was a problem." Semafor's account of that requirement puts it at a third of the country's total installed capacity of around 3,000 MW, roughly 1,000 MW. That is a different figure from the 100 MW first phase announced in 2024, and the distinction matters for everything that follows. In October 2026, Kenyan outlets reported that Microsoft plans to close parts of its Africa Development Centre in Nairobi, with roughly 400 jobs affected. The steam, meanwhile, keeps flowing. The gap between those two facts is the most instructive case study in African digital infrastructure right now. It shows where Kenya's AI data-centre race actually stands, and the lesson has less to do with geology than most commentary assumes. ## The terms, decoded This story leans on a handful of power-market words. Short versions: | Term | What it means | | --- | --- | | Conventional geothermal | Hot rock heats natural groundwater into steam; wells bring it up. This is Olkaria's system, in service for four decades. | | Enhanced geothermal (EGS) | No natural steam field. Wells drill sideways into hot dry rock, and water pumped through engineered fractures makes the reservoir. This is Fervo's system. | | Fervo Energy | The US enhanced-geothermal company behind Cape Station in Utah. Its first commercial block, 33 MW, reached commercial operation in September 2026. | | PPA (power purchase agreement) | A long-term contract to buy power at an agreed price. Lenders lend against signed PPAs. | | Capacity payment | Payment for power capacity to be available whether or not it is used. The fight over who guarantees these payments is what stalled the Olkaria deal. | | Single buyer | Kenya Power buys most of the country's generated electricity and resells it, so a large customer must work through or around that structure. | | Behind the meter | A facility generates power on its own side of the meter, and grid contracting never enters the deal. | | Live IT load vs designed capacity | What servers draw today versus what the halls were built to hold. DC254 tracks both, separately. | ## What the verified tracker actually shows DC254's directory tracks 31 facilities and projects across Kenya and the region. The verified numbers are small. Twenty Kenya facilities are operational, with a published in-service IT load of 10.5 MW. Designed capacity at those live sites totals 42.9 MW. A further 230 MW sits in the announced pipeline, of which 77 MW is genuinely under construction at three projects: iXAfrica's NBOX1.2 (18 MW), Africa Data Centres' Nairobi 2 (15 MW), and Nxtra by Airtel at Tatu City (44 MW). Now hold the figures against each other, each with its own basis. ![Bar chart on one linear scale against Kenya's installed generation: 10.5 MW verified live IT load, 42.9 MW designed capacity, 77 MW under construction, 100 MW announced first phase, 230 MW tracked pipeline, about 1,000 MW reported power requirement, 3,000 MW national installed capacity](/images/diagram-olkaria-scale-gap.webp) The announced first phase alone was 100 MW, close to ten times the country's entire verified live load. The requirement behind the scenes, as President Ruto described it, was nearer 1,000 MW. For reference, Kenya's record peak demand, reached in December 2025, was 2,439 MW (TechCabal, 20 August 2026). Either version of the campus would have been one of the largest individual loads the grid has ever been asked to carry, announced into a market whose whole operating fleet is a rounding error next to it. That scale gap frames everything else in this story. Kenya's data centre market is real, growing, and verified down to the megawatt. It is a fleet of tens of megawatts, and the AI race is being contested in hundreds. ## The steam was ready on day one Kenya's geothermal credentials are among the strongest outside Iceland and the United States. The installed fleet stood at 943.7 MW at the end of 2023, according to the regulator EPRA, and geothermal supplied 39.51 percent of the country's electricity in the year to June 2025, the largest single source in the mix. Olkaria V alone contributes 172 MW. KenGen, the state generator, is completing Olkaria VII, an addition in the 80 MW class, and the Ministry of Energy's long-standing estimate puts the Rift Valley's total potential at about 10,000 MW. None of that was ever in doubt. The resource could credibly support the campus. What Kenya could not put on the table in 2024 was a completed path from steam to a powered, contracted data hall, and that path runs through parts of the system the headlines rarely mention. ## Where the deal actually stalled Three kinds of evidence tell the stall story, and they say different things. The company announcement came first: on 22 May 2024, Microsoft and G42 put the initial investment at $1 billion and described a campus run entirely on renewable geothermal energy. Government statements came two years later: President Ruto, quoted by Semafor in May 2026, said the project was reconsidered once its power requirement became clear, at about a third of the country's installed capacity. Wire reporting filled in the mechanism. By May 2026, Reuters reporting carried by Bloomberg described talks stalled over the company's request for guaranteed annual capacity payments: Microsoft and G42 had asked the Kenyan government to commit to paying for a certain amount of capacity annually, and the talks broke down when the guarantees could not be provided at the level requested. [The directory record for the project](/directory/microsoft-g42-ai-dc) carries that attribution, with a caveat that the outlets share an upstream source, and holds Kenya's counter-position from the same wire: principal secretary John Tanui's line that the project "is not failed or withdrawn" and that "the scale of the data centre they wanted to do still requires some structuring." The numbers behind that caution came into focus later. Kenyan business press tracked the requested power growing from 60 MW in early discussions to 1,000 MW by mid-2026 (Kenyans.co.ke, 22 August 2026; Business Daily, 10 September 2026), roughly a third of installed generation and the same magnitude President Ruto described. The announced 100 MW first phase and the reported 1,000 MW requirement describe the same project at different moments, and much of the public confusion around this deal comes from reading one for the other. Put the evidence together and the shape of the problem is specific. Kenya can permit generation and drill steam. What it lacked was a bankable contract structure for a hyperscale customer, and grid headroom in the right place. A campus at Olkaria would also depend on transmission into Nairobi's metro network, where [the country's connectivity concentrates](/articles/kenya-power-infrastructure-data-centres): the KIXP peering point alone carries 144 member networks and 2,985 Gbps of connected port capacity, and the submarine systems that land in Mombasa ride terrestrial fibre to reach it. ![High voltage transmission towers crossing the Rift Valley toward Nairobi](/images/kenya-transmission-towers-riftvalley.webp) Power is one input in that stack. The Olkaria deal had to clear every input at once, and it had to do so as a single bespoke negotiation with a state utility in the middle. ![Diagram tracing the power chain from grid generation to GPU racks](/images/diagram-power-chain-grid-to-gpu.webp) ## What Utah did differently The useful comparison sits outside Africa. Fervo Energy, a US company, drills horizontal wells into hot rock and engineers the reservoir it needs, an approach called enhanced geothermal. There is no natural steam field to tap: water goes down one well, heats as it moves through engineered fractures in the rock, and comes back up another as steam to spin turbines. ![Two cross-sections side by side: a conventional hydrothermal field with wells into a natural steam reservoir, and an enhanced geothermal system with horizontal wells and engineered fractures in hot dry rock](/images/diagram-egs-vs-hydrothermal.webp) Fervo's first commercial project fed Google's data centres in Nevada. The climb since then reads like a financing ladder, and each rung has its own status. In June 2024, Fervo signed 15-year power purchase agreements totalling 320 MW with utility Southern California Edison for its Cape Station project in Utah. In 2025, Nevada regulators approved a 115 MW Clean Transition Tariff for the technology. In March 2026, the company closed $421 million in non-recourse project financing to fund Cape Station's first-phase construction. In September 2026, Google signed a 396 MW PPA with Fervo, the largest corporate geothermal deal on record. On 1 October 2026, Fervo announced that the first 33 MW GeoBlock at Cape Station had reached commercial operation, declared on 30 September, one day ahead of its contractual date, and already earning revenue under the PPA. ![Timeline of Fervo Energy milestones with status chips: 3.5 MW pilot delivered 2023, 320 MW Southern California Edison PPAs signed June 2024, 421 million dollar financing March 2026, 396 MW Google PPA September 2026, first 33 MW block delivered September 2026, Phase I due January 2027, 400 MW next phase building for 2028](/images/diagram-fervo-milestones.webp) Two rungs are delivered, two are signed contracts, one is closed financing, one is a regulatory approval. The rest of Phase I, about 100 MW across three standard 33 MW blocks, is due on contract by 1 January 2027, and a further 400 MW phase is under construction for 2028. Across the portfolio, Fervo counts more than 1 GW of binding PPAs. Look at what made that progression financeable, because it is all commercial. Google and SCE contracted directly with the generator, at fixed prices, over long terms, in block sizes a single PPA could carry. Standard blocks meant each financing closed on its own merits, and the drilling could be scheduled like a factory. Three years after a 3.5 MW pilot, the approach is delivering a 33 MW block and holding contracts for hundreds more. So make the comparison honestly. Kenya has the more proven resource, and Fervo has the more repeatable commercial model. The Rift is a natural hydrothermal field that has carried Kenya's grid for four decades; Cape Station's reservoir had to be engineered out of hot dry rock, a harder geology problem, and Fervo solved it anyway to reach a commercial delivery. What Fervo proved sits in the commercial layer: standard block sizes, direct long-term PPAs, and project finance closed against signed contracts. That model is the transferable part. The steam was never the hard part in Kenya. One more echo, and this one is DC254's read rather than an industry claim: Cape Station's entire first phase, about 100 MW being delivered as three standard blocks, is the size of the Olkaria campus's announced first phase. One company carried a 33 MW block on one contract, three times over, while the Kenya negotiation tried to land the same megawatts in a single bespoke state deal. The megawatts are similar. What differs is how many signatures each megawatt needs. ## What would turn announcements into live load Kenya's pipeline records share one honest gap: the tracker can show a megawatt figure and a status, and almost no project publishes where it stands on power procurement. Announced, contracted, under construction and commissioned are different stages with different risk, and the announced figure of 230 MW says nothing about how much of it has a signed path to power. That distinction is exactly what the [tracker](/tracker) now stages explicitly, project by project. Two moves would shift the market up that ladder. The first is modular procurement: 33 MW blocks that one contract can carry, rather than a 100 MW campus that needs the Treasury in the room. That first move is DC254's reading of the Fervo example, not a proven path for Kenya; the supporting sign is that the 77 MW under construction at three Nairobi projects shows Kenyan operators already build in that size range when the financing is standard. The second is a contract design Kenya is willing to defend. The capacity-payment dispute taught both sides where the other's limits sit, and the next attempt has to answer it, whether through direct corporate PPAs, behind-the-meter generation at the plant, or a reformed structure through the single buyer. The steam can carry any of those models. It has been carrying Kenya's grid for four decades. ## What to watch Three markers will tell you whether Kenya converts steam into compute. Watch [the Olkaria campus record](/articles/microsoft-g42-kenya-data-centre): if it moves from Early Stage to a signed structure, the contract question has an answer. Watch Olkaria VII's commissioning and the transmission work that accompanies it, because grid headroom is bought down wire by wire, not by press release. And watch the 77 MW under construction: if those three projects commission on schedule, Kenya's next negotiation starts from roughly a doubled live base, and the arithmetic that scared everyone in May 2026 looks different. The Fervo clock is running in parallel. Cape Station's remaining Phase I blocks are due on contract by 1 January 2027, a further 400 MW is under construction for 2028, and the portfolio now holds more than 1 GW of binding PPAs. Kenya's next hyperscale negotiation will be judged by the same standard Fervo just met: a price, a term, and a block size a bank can underwrite. ## Inside Olkaria's Geothermal Data Centre Corridor URL: https://data-centers-254.vercel.app/articles/olkaria-geothermal-data-centre-corridor Cluster: Renewable Energy Published: 2026-09-16 Updated: 2026-09-26 Summary: Why hyperscale projects are looking past Nairobi's grid to Olkaria's geothermal wells, what stalled the Microsoft and G42 build, and the latency math. Kenya has a power problem and a power surplus at the same time, and both point to the same place on the map. The country's grid carries roughly 3,000 megawatts of installed capacity, and the largest single slice of its electricity comes from steam fields at Olkaria, about 120 kilometres northwest of Nairobi. Yet the biggest data centre project ever announced in East Africa stalled in 2026 largely because of power, not money, not permits, and not demand. The idea now pulling investors and government officials alike is the geothermal corridor: putting compute close to the wells instead of stretching the grid thin to reach it. This article explains what the corridor actually is, what happened to the Microsoft and G42 project that put Olkaria on the data centre map, and the honest engineering trade-offs (fibre, latency, workforce) that will decide whether the Rift Valley ever hosts server halls at scale. ## The $1 Billion Project That Ran Into the Grid The project that defined the conversation was announced in May 2024, during President William Ruto's state visit to Washington. Microsoft and the UAE's G42 committed around $1 billion to a data centre intended to serve government and business customers on Azure, planned for a site roughly 100 kilometres northwest of Nairobi, in the direction of Naivasha and the Olkaria steam fields. Geothermal power was the headline: the facility was meant to run on the same renewable baseload that powers roughly 45 percent of Kenya's electricity generation. It never broke ground. Semafor reported on 6 May 2026 that President Ruto, speaking in Nairobi, said the project was reconsidered after it became clear it would require about a third of the country's roughly 3,000 megawatts of installed capacity. "To switch on that one data center, we would need to shut off power for half the country," he said. The project as announced was a data centre campus in the region of 100 megawatts of IT load, per the entry in our [Kenya data centre directory](/directory), but the president's arithmetic described the generation and grid build-out a campus of that ambition was expected to pull with it. Either way, the numbers describe a load Kenya's grid could not absorb at a single site in 2026. The paperwork trail matters as much as the quotes. Government officials told Semafor the technology ministry developed a project concept note and took it to the National Treasury, which never approved the funding. A meeting between Kenyan officials and Microsoft executives in August 2025 made clear the facility would miss its original May 2026 online target. Semafor also reported the context around the announcement: the deal had been framed as a symbol of strengthening Kenya-US relations, and MoUs signed at bilateral summits often move faster than feasibility studies. G42 declined to comment; Microsoft did not respond to a request for comment. ## What Kenya's Grid Can and Cannot Carry Kenya's power system is genuinely strong by regional standards, and genuinely finite. Installed capacity sits at roughly 3,000 megawatts, geothermal contributes approximately 950 megawatts of it, and demand from homes, industry and the Mombasa Road data centre corridor is growing at the same time. The [national grid picture](/articles/kenya-power-infrastructure-data-centres) is the binding constraint on every hyperscale conversation: Kenya Power cannot today deliver the kind of single-site load a flagship AI campus wants, which is precisely why the Microsoft and G42 project stalled and why our directory marks it early stage rather than pipeline. Two government numbers frame how the corridor idea fits in. The first is the target: President Ruto has argued Kenya must raise energy capacity to 10,000 megawatts by 2030 for projects of this class to work, part of a broader push to fund roughly $38 billion in energy infrastructure through asset sales and private capital. The second is the resource: Kenya's geothermal potential is estimated at 7,000 to 10,000 megawatts, of which only about 950 megawatts has been developed, almost all of it at Olkaria. The constraint is not the heat under the Rift Valley. It is the pace at which wells, plants and transmission lines can be financed and built. ## Why Move Compute Next to the Steam? The economics of siting compute near the wellhead are straightforward. Power bought close to generation avoids transmission losses and some distribution charges, and analysis in our [geothermal explainer](/articles/geothermal-energy-kenya-data-centres) puts the saving for a Rift Valley site at 30 to 40 percent below Nairobi levels. Olkaria also sits at roughly 2,000 metres altitude, where cooler air helps free-cooling economics. Geothermal is baseload: it runs through the night and does not depend on rainfall, which matters for the high utilisation rates that make data centre capital efficient. There is evidence the corridor is forming rather than just being discussed. Coverage of the ITW and Datacloud Africa 2026 summit in Nairobi (The Star, 14 September 2026) reported that Kenya's live data centre capacity is growing "with more under construction in Olkaria and Konza", a notable line because it puts an active Rift Valley site alongside the government-backed smart city. Konza Technopolis runs its own campus power model, [as our Konza deep dive explains](/articles/konza-technopolis-data-centres-kenya), and the same logic (build the power plan first, the halls second) is what a direct geothermal integration at Olkaria would formalise. ## The Fibre and Latency Reality Every honest corridor conversation has to clear the latency question, and the physics are friendlier than most people assume. Over the roughly 120 kilometres between Olkaria and Nairobi, a signal travelling on fibre needs about 1 to 2 milliseconds for a round trip even before equipment overheads (a rough estimate from the speed of light in glass, not a measured route figure). That is a rounding error compared with the delay of routing traffic to Europe or South Africa and back. Latency will not kill a Rift Valley data centre. The real connectivity questions are route diversity and interconnection density. Nairobi has the metro fibre, the internet exchange points and the carrier community; a greenfield site at Olkaria needs high-capacity fibre built or extended to it, ideally over more than one physical path, before any carrier treats it as a credible colocation market. AI training workloads tolerate this trade-off well, because they move huge datasets between a handful of locations on schedule. Transactional workloads that serve M-Pesa and banking customers want the densest possible interconnection, which for now keeps them in Nairobi. The sensible reading of the corridor is a split market: training and batch compute near the steam, customer-facing compute in the metro. ## What Would Make the Corridor Real Three things have to line up. First, generation: more wells and plants at Olkaria, Menengai and Suswa, developed by KenGen and the Geothermal Development Company, on the path to the government's ~1,600 megawatt geothermal target for 2030. Second, transmission: high-voltage lines (KETRACO's portfolio) sized for dedicated industrial loads, or a formal direct-integration framework that lets a large campus contract power at the generator gate. Third, commercial structure: the kind of purpose-built licensing clarity the [Communications Authority is currently consulting on](/articles/kenya-ca-standalone-data-centre-licence), and the investment incentives covered in [our tax piece](/articles/tax-incentives-data-centre-investment-kenya). The prize explains the persistence. Kenya is projected to account for more than three-quarters of fresh data centre capacity across East African markets by 2030, according to an African Union report cited by Semafor, and the Kenya market is projected to roughly triple to $805 million by 2031 (a 2026 market report cited by Semafor; market estimates vary, [as we explain here](/articles/why-kenya-data-centre-market-estimates-disagree)). A country that solves the power-at-scale question first will collect a disproportionate share of that build-out. The [compute hub plan](/articles/kenya-data-centre-compute-hub-plan-explained) the ICT ministry published in September 2026 makes exactly this argument: cables are done, compute is next, and power is the gate. ## Three Models Now Compete for the Same Load The corridor does not develop in a vacuum. As of late September 2026, three competing power models are on the table for Kenya's compute build-out, and each needs to be read on its own terms. The first is this article's model: compute-to-the-wells, siting campuses at or near geothermal generation so the load lives where the baseload is. The second is regional hydro imports: Ethiopian Electric Power plans to raise exports to Kenya from 200 to 400 megawatts from December 2026, subject to grid testing, which we track in [our Ethiopia power exports analysis](/articles/ethiopia-power-exports-kenya-data-centres). The third arrived on 26 September 2026, when Aliko Dangote announced that his proposed Lamu refinery would include an LNG-fired power plant of about 1,000 megawatts, with roughly 500 megawatts potentially offered to the Kenyan government, covered in [our Dangote Lamu analysis](/articles/dangote-lamu-refinery-power-plant-data-centres). That announcement is attached to a refinery that has not broken ground, with no power purchase agreement public. The discipline in reading all three is keeping their numbers apart. Olkaria's roughly 950 megawatts of installed geothermal is operating capacity on today's grid. Ethiopia's 400 megawatts is a planned trade volume entering the bulk system, not a private feed. Dangote's 500 megawatts is an announcement, the same status our directory gives the Microsoft and G42 campus's 100 megawatts. None of the three is reserved for data centres, and adding them together produces a number that describes nothing, because the figures sit at different stages of certainty and different points on the grid. What decides Kenya's compute future is which model turns its promise into contracted, transmitted, dispatchable supply first, and for the corridor that means wells, plants and a direct-integration framework, not a headline. ## What to Watch - Treasury action on energy infrastructure: the $38 billion funding push is the single biggest variable between the 3,000 MW grid of 2026 and the 10,000 MW ambition of 2030. - Named projects at Olkaria: the ITW 2026 coverage points to construction under way, but no operator, size or timeline is public yet. We will add it to the directory the moment it is confirmed. - Whether the Microsoft and G42 concept returns in any form: the directory entry stays early stage, and the lesson from the stall (match the load to the generation plan first) is now shaping every new proposal. - Direct-integration policy: how the government lets large campuses contract power at or near generation, which decides whether the corridor is a metaphor or an engineering programme. Sources: Semafor (6 May 2026, Ruto statements, Treasury concept note, market projections); The Star (14 September 2026, ITW and Datacloud Africa 2026, Olkaria and Konza construction); Business Daily (15 September 2026, regional connectivity context). Grid and geothermal figures (950 MW installed geothermal, Olkaria complex roughly 800 MW, 45 percent generation share, 120 km distance, 30 to 40 percent cost advantage, 7,000 to 10,000 MW potential) verified against our geothermal explainer, first published with dated sources. Facts verified 16 September 2026; updated 26 September 2026 to add the three-model competitive framing (Olkaria compute-to-the-wells, Ethiopian hydro imports, announced Lamu coastal LNG) following the Dangote announcement, with figures for each model kept separate. ## G2M Fibre Route Is Live: An East-West Data Artery URL: https://data-centers-254.vercel.app/articles/paratus-g2m-fibre-route-live Cluster: Connectivity Published: 2026-09-08 Updated: 2026-09-16 Summary: Paratus Group's 2,000 km Goma-to-Mombasa fibre route is live, linking DRC, Rwanda, Uganda and Kenya to subsea capacity. Why it matters. East Africa's data infrastructure runs north-south. Fibre comes inland from the Mombasa cable landing stations to Nairobi, then forks toward Kampala, Kigali, Juba and beyond. That topology has served the region well, but it means almost every landlocked network's international traffic rides the same corridor, and every network in the east of the DRC rides it longest. Paratus Group's announcement at ITW Africa 2026, confirmed by ITWeb Africa on 8 September, changes the geometry: the **Goma-to-Mombasa (G2M) route (roughly 2,000 km of terrestrial fibre linking Goma via Kigali, Kampala and Nairobi to the Kenyan coast) is now live and carrying traffic**. ![Fibre optic infrastructure on the new G2M route](/images/dc-fibre-optics.webp) ## What the G2M route actually is The G2M is not a new submarine cable, it is the terrestrial artery that connects eastern DRC's traffic to the submarine cables that *do* land on the coast. Its path stitches together the region's established network geography: Goma in the east of the DRC, through Rwanda's Kigali, Uganda's Kampala and Kenya's Nairobi, terminating where the subsea systems come ashore. Paratus describes it as an important east-west connection providing access to international subsea cable capacity landing at Mombasa, and, critically, the route is not just lit fibre on a map. Paratus is complementing it with **points of presence and interconnections at data centres and internet exchanges** along the way, which is what turns a physical route into capacity a customer can actually buy. The company behind it has been assembling this quietly. Paratus (a pan-African network operator with southern African roots) has been building East African presence through infrastructure, services and cross-border connections, recently added **Starlink services in Uganda** to its mix, and is represented in the region by group CTO Gert Duvenhage and Kenya country manager Joseph Kibwott. At ITW Africa 2026 in Nairobi this month, it is hosting the customers and partners the route was built for. ## Why an east-west artery matters more than it sounds The unglamorous truth about African bandwidth is that it fails in corollaries. Kenya's submarine capacity is extraordinary on paper (seven live cable systems land on the Kenyan coast, from SEACOM and TEAMS to the 16 Tbps PEACE route) but the region's *terrestrial* diversity has lagged. Cut or degrade the inland corridor and you learn exactly how much of the network was really diversified. Carnegie's April 2025 analysis put Mombasa's single-point-of-failure risk bluntly, and the concern compounds as traffic grows: [our own infrastructure map](/infrastructure/map) shows nine cable systems (seven live, with Africa-1 landed and Daraja in development) converging on one coastal bottleneck. The G2M adds three specific kinds of resilience: 1. **A second path to the sea for eastern DRC.** Goma's networks have historically backhauled through long, fragile inland corridors. A purpose-built route to Mombasa shortens the path and, more importantly, diversifies it. 2. **Ring potential for the northern corridor.** With G2M live alongside the established Kampala–Nairobi and Kigali–Kampala routes, operators can start building genuine rings, traffic that fails over between paths instead of waiting for a single fibre cut to be spliced. 3. **A deeper funnel for Mombasa's landing stations.** Every new terrestrial feeder makes the coastal cluster (iColo's MBA1 with its 94 registered networks, MBA2 two kilometres from the beach manhole) more valuable as an interconnection destination rather than a transit waypoint. ![Fibre technician working on the terrestrial network](/images/aerial-fibre-technician-utility-pole.webp) ## The Kenyan angle: what it changes for data centres and buyers For Kenya's data centre market, the G2M is quietly significant. Interconnection density, the metric we [track on every directory entry](/directory), grows when more networks can reach more buildings over more paths. A live east-west feeder means DRC, Rwandan and Ugandan networks have one more reason to place points of presence in Nairobi's carrier-neutral facilities and at Mombasa's coastal sites, rather than simply transiting. More networks present means more peering, lower latency for regional traffic, and better economics for the Kenyan businesses buying transit behind them. For buyers of connectivity in Kenya, the practical advice is unchanged but easier to execute: ask any provider how many *diverse physical paths* your traffic will take to the submarine cables, not just how many cables the provider "has access to". Routes like G2M are what make a good answer possible. ## What to watch Three signals will tell us whether the G2M becomes a load-bearing artery or another lit route on a map. First, whether Paratus lands wholesale and enterprise capacity deals along the corridor, the ITWeb report says the route is carrying traffic, but carrying *whose* traffic and at what volumes is the real test. Second, whether the region's internet exchanges (KIXP, RINEX, and Uganda's UIXP) see membership growth from DRC-facing networks, which would show the route being used for peering rather than pure transit. Third, whether it pressures the market toward more east-west builds, the region's next frontier is redundancy *between* corridors, not just along them. [Our map](/infrastructure/map) now carries the G2M alongside the terrestrial backbone, and we will keep the [connectivity explainers](/articles/submarine-cables-landing-mombasa) updated as the traffic story develops. ## PwC's $31.6 Trillion Forecast: Where Kenya Fits In URL: https://data-centers-254.vercel.app/articles/pwc-global-data-centre-capex-2050 Cluster: Market Analysis Published: 2026-09-04 Updated: 2026-09-04 Summary: PwC projects $31.6 trillion of data centre capex through 2050, and Africa's $255bn share is the only non-AI bet. Kenya's grid is the quiet edge. Railways. Electrification. The internet. Each required enormous capital and defined an era, and, according to PwC's data centre outlook published with modelling by Oxford Economics, the AI infrastructure cycle now underway dwarfs all three. In the central scenario, data centres attract US$31.6 trillion of capital expenditure globally through 2050, within a range of about $22 trillion to nearly $50 trillion depending on how fast AI adoption runs. Annual spending accelerates as it goes: roughly $800 billion in 2026, $1.1 trillion by 2030, $1.8 trillion by 2050. Unlike every previous infrastructure wave, this one does not tail off when the network matures, it resets every four to six years, because the capital goes into what fills the buildings. For an African readership, the forecast's most important page is not the topline. It is the region table, where Africa's $255 billion cumulative capex sits slightly below its share of global GDP but carries a growth profile PwC describes, paradoxically, as one of the lowest-risk capex stories on the map, and where Kenya's power grid, at approximately 95 percent renewable, gets named one of the most sustainably powered data centre markets in the world. This article unpacks the numbers, the five forces PwC says decide where capital lands, and what the two geopolitical scenarios PwC stress-tested would mean for East Africa's place in the flow. ## Why This Capex Cycle Never Ends The structural insight underneath the $31.6 trillion is a ratio most markets still underprice: every $1 of construction capex effectively commits the industry to roughly $12 of future ICT equipment spend. Buildings and utility connections have long useful lives; the servers, storage, networking, CPUs, and crucially GPUs inside them age out in four to six years, meaning a single data centre may need three to five rounds of equipment investment over a 20-year asset life. PwC's memorable formulation (the data centre as essentially a chip-replacement subscription with a building wrapped around it) inverts the pattern of every prior infrastructure wave, which front-loaded construction and let spending decay as networks matured. The composition shifts over time in a way that matters for equipment-heavy markets. ICT equipment rises from 70 percent of total capex in 2026 to 93 percent by 2050, driven by AI workloads that rely on expensive, energy-intensive accelerators following a fast innovation curve. Higher rack densities force power and cooling upgrades alongside, but the largest recurring share is the compute itself. For African markets, this is a quiet warning about import dependence: a region with limited semiconductor depth and weaker access to advanced ICT equipment captures the construction share of the cycle but leases the technology share, a distinction the downside scenario below makes painfully concrete. ![Rows of GPU compute clusters inside a data centre](/images/dc-gpu-cluster-2-wide.webp) The demand base widens too. The cloud era had essentially one customer profile, hyperscalers. The AI era adds at least five more: neoclouds, model developers, inference platforms, enterprises, and governments, each with different demand and risk profiles. About 30 percent of workloads today carry localised requirements (latency, data access, privacy, and sovereignty constraints that pull capacity toward the users) and PwC expects that share to grow, which is precisely the slice of demand African markets can realistically compete for. ## The Five Forces Deciding Where Capital Lands PwC's framework for who captures the spending comes down to five forces, and reading them through a Kenyan lens is instructive. Power sits at the top: affordable, reliable, increasingly low-carbon electricity at scale is the hardest requirement for most markets, and transmission capacity, substation availability, and multiyear transformer lead times have become the bottlenecks that determine which projects break ground and when. In Kenya, that is not an abstraction, [grid interconnection timing](/articles/kenya-power-infrastructure-data-centres) already shapes facility timelines, and the operators who secure megawatts fastest will capture a disproportionate share of regional demand. The next three forces sort markets by workload type. Latency and connectivity determine which regions can credibly serve consumer and enterprise inference, a contest Nairobi's [submarine cable gateway position](/articles/why-data-centres-cluster-nairobi-mombasa) and [internet exchange](/articles/kixp-internet-exchange-point-kenya) depth actually play into. Security and trusted-region hosting decide where regulated and sovereign data goes, which rewards markets with credible data protection regimes. GPU access and ecosystem depth govern the frontier: which markets can host training clusters at all, given chip availability and technical talent. Policy certainty and community consent aggregate the rest, converting underlying advantages into capacity that actually gets built. PwC is explicit that sustainability is not modelled as a separate variable but runs through all five forces as either advantage or constraint. Markets with renewable-heavy grids and credible decarbonisation pathways carry a structural comparative advantage the model picks up through their power positions. That single sentence is the most under-appreciated paragraph in the entire outlook for anyone sitting on the East African rift. ## Africa's $255 Billion: The Only Non-AI Bet on the Map The regional numbers deserve close reading. The Americas account for $16.5 trillion in cumulative capex through 2050, with the US alone at $15.1 trillion, around 48 percent of the global total. Asia-Pacific takes $8.2 trillion, Europe $5.6 trillion, the Middle East $1.1 trillion on the fastest CAGR in the forecast, and Africa $255 billion, the smallest number, but attached to the lowest-variance story. PwC's reasoning is worth quoting in spirit: Africa is the only region whose central scenario is not a bet on AI. The continent is largely buying foundational digital infrastructure (cloud storage, enterprise compute, connectivity) that pays off regardless of which AI scenario plays out. Slower AI adoption removes roughly $10 trillion from the global cycle and leaves stadium-sized facilities half empty in other regions; in Africa it simply means the foundational build continues on its own logic. South Africa anchors the region with the most established data centre base, while Kenya, Nigeria, and Ghana are named among the most promising emerging markets. And Kenya's distinctive position is its grid: approximately 95 percent renewable, a mix our [renewables analysis](/articles/kenya-renewables-industrial-power-data-centres) and [geothermal deep-dive](/articles/geothermal-energy-kenya-data-centres) examine from the engineering and economics sides, the kind of low-carbon baseload that PwC's framework treats as a structural power advantage rather than a press release. ![Solar panels at a Kenyan solar farm](/images/kenya-solar-farm-2.webp) ## Two Scenarios That Redraw the Map The forecast's stress tests matter as much as the base case, and they pull in opposite directions for Africa. The first scenario escalates today's export controls to peak trade-war intensity: advanced GPUs become harder to procure across a wide set of markets, and retaliatory raw-material restrictions propagate along the semiconductor supply chain. Global cumulative capex falls from $31.6 trillion to $25.5 trillion by 2050 (a roughly $6 trillion shortfall) with the Middle East the most proportionally exposed region and AI-heavy pipelines hit first. Africa is not a direct target of export controls but absorbs meaningful second-order disruption through limited semiconductor depth and weaker access to advanced ICT equipment, with cumulative capex falling from $255 billion to $193 billion. The lesson for African operators is about equipment strategy: hardware-flexible designs, phased procurement, and vendor diversification are hedges, not luxuries. The second scenario tests sovereignty as an organising principle: governments and regulated industries stop relying on foreign infrastructure for critical workloads while non-critical demand keeps moving. Total capex barely moves (down 6.7 percent to $29.5 trillion) but the redistribution is dramatic. Africa records the largest proportional uplift of any region, with cumulative capex rising roughly 12 percent from about $255 billion to $284 billion, supported by near-term onshoring of cloud storage and longer-term AI market development. It is one of the few scenarios in which Africa is a net winner, and it runs directly through the localisation frameworks already being legislated across the continent, including [Kenya's data sovereignty rules](/articles/kenya-data-sovereignty-localisation). ## What It Means for Kenya Translated into Kenyan strategy, PwC's recommendations compress into four moves. Compete on speed to power: pre-procure long-lead grid components and press for predictable interconnection queues, because transmission timing now decides which projects break ground. Design for optionality: phased grid connections, modular capacity, and hardware-flexible cooling preserve value if AI accelerates and limit stranded-asset exposure if it does not. Sign long-dated leases with resilient counterparties (hyperscalers, sovereign vehicles, regulated industries) whose demand survives the slow scenarios; the off-take, not the building, is the hedge. And plan the low-carbon build from the start, because markets where renewable capacity scales alongside data centre siting convert a compliance story into a siting advantage, the position Kenya already occupies, as our [AI infrastructure in East Africa](/articles/ai-data-centres-east-africa) analysis argues from the demand side. ![High-voltage transmission pylons in Kenya](/images/kenya-transmission-pylons.webp) The $31.6 trillion question, PwC concludes, is not whether the capital exists or whether the demand is real, it is which regions, operators, and institutions are positioned to capture it. Africa's answer is unlikely to be gigawatt training campuses. It is far more likely to be the sovereign, carbon-conscious, latency-bound workloads that the sovereignty scenario rewards and that Kenya's renewable grid is uniquely configured to host. The forecast's map of 2050 will look little like 2026's. Kenya's job, between now and then, is to make sure the redrawn lines run through Nairobi. ## Raxio vs Africa Data Centres in Kenya: A Fair Comparison URL: https://data-centers-254.vercel.app/articles/raxio-vs-africa-data-centres-kenya-comparison Cluster: Operators Published: 2026-09-16 Updated: 2026-09-18 Summary: Raxio and Africa Data Centres are both expanding in Kenya, but they stand at different stages today. A fact-checked, plain-English comparison for Kenyan buyers. Two names keep coming up in the same breath for anyone shopping for data centre space in Kenya: Raxio and Africa Data Centres. Both are pan-African operators, both describe themselves in nearly identical language (carrier-neutral, Tier III, enterprise-grade), and both have Kenya in their growth plans. The main difference between them today is timing: one is operational in Nairobi, while the other is still building toward its Kenya opening. This article walks through what each operator has delivered and what is still ahead of it, in plain English, so you can plan on verified facts. A fair comparison starts with verification. Before putting these two names side by side, we checked every claim in this article against primary sources: the operators' own websites, independent industry directories and trade press with named publication dates. Where something could not be confirmed, we say so. That is the house standard at DC254, and it applies to every operator we cover equally. ![Ribbon cutting at the NBO2 data centre launch](/images/nbo2-launch-ribbon-cutting.webp) ## First, the location claim, checked Some comparison notes describe this matchup as "Mombasa Road versus Upper Hill or Karen". We checked, and the second half does not hold up. Africa Data Centres' own website places its Nairobi flagship NBO1 at Sameer Business Park, Block A, on Mombasa Road (fetched 16 September 2026). PeeringDB, the industry's independent directory that operators themselves update, lists the same address (entry dated September 2025), and third-party facility trackers agree. No Africa Data Centres facility sits in Upper Hill or Karen. Why does an address deserve a whole section? Because practical decisions start there: a site visit, a pricing call, a carrier ordering last-mile fibre, all begin with a location. The confirmed picture is simple: NBO1 sits on Mombasa Road, inside Sameer Business Park, and that is where the most densely connected building in Kenya stands, with four internet exchanges operating on site. ![The Nairobi Expressway stretching over Mombasa Road](/images/nairobi-expressway.webp) ## Africa Data Centres: what is verifiable Africa Data Centres is the data centre arm of Cassava Technologies, a group headquartered in South Africa, and its Kenyan story is currently the more concrete one. The flagship, [NBO1](/directory/africa-dc-nairobi-1), is operational at Sameer Business Park as described above, with about 2,000 square metres of technical space listed in its network directory entry, and a reputation as the building where Kenyan networks meet: four internet exchanges under one roof, which is precisely what [carrier neutrality](/articles/what-is-colocation-kenya) is supposed to deliver. Then there is the new one. In September 2026, Digital Realty and Africa Data Centres launched NBO2 in Nairobi, with about 6.4 megawatts of commissioned capacity reported by trade press (ET Datacenters, 10 September 2026; first reported 7 September 2026). We examined the launch and the "gateway" claims around it in a separate piece on [NBO2](/articles/digital-realty-nbo2-launch-nairobi). A megawatt, for scale, is roughly the draw of a few hundred Kenyan homes, so 6.4MW of commissioned data centre capacity is a serious expansion of what Nairobi can host. The verifiable summary for ADC: one operational flagship with unmatched connectivity, one brand-new facility with a joint-venture partner, group ownership under a listed-technology family, and addresses you can physically visit tomorrow. ![Map showing data centre locations across Africa](/images/africa-dc-map.webp) ## Raxio: what is verifiable, and what is not Raxio Group, established in 2018, describes itself as Africa's leading carrier-neutral Tier III operator, and its expansion record so far supports that description. Its Kampala facility opened in 2021 and its first Angola facility launched in late 2025 (Raxio pages and launch coverage). The money is real too: in July 2026 the group secured up to US$380 million to fund African expansion (Data Centre Magazine, 13 July 2026), and its earlier build-outs carried a US$12.24 million guarantee from MIGA, the World Bank Group's investment insurance arm, approved in September 2022. Institutional backing of that kind is a meaningful signal: professional insurers and lenders ran their own diligence before underwriting the build. Kenya is squarely in the plan. Sector briefings, including the D4D Hub's East African market briefs (Xalam Analytics, January 2026), describe Kenya as one of Raxio's target locations, with a first facility suggested to go live in 2026. As of 16 September 2026, though, the first Nairobi facility has not opened yet: no launch announcement, customer-facing address or commissioning report has been published so far. Our directory, which holds every Kenyan facility to the same verification bar of primary confirmation, lists Raxio Nairobi as early stage. That is a status, not a criticism; every operational facility on our books was an announcement first. For buyers, it simply means Raxio belongs on the watch list for future tenders rather than the shortlist for capacity needed this quarter. The distinction matters because these two operators are at different stages of the same journey. Africa Data Centres' Kenyan facilities are open and taking customers; Raxio's are still on the way. Neither stage is better or worse in itself; what matters is matching the operator's stage to your timeline. If you need capacity this quarter, the open facilities decide themselves. If you are planning 2027 tenders, both names belong in the conversation. ![Business people reviewing investment documents](/images/dc-business-investment.webp) ## So who wins for a Kenyan buyer? If the question is where to place hardware this quarter, the answer does not actually involve a versus: it involves the facilities that are open. Nairobi's operational, verifiable options include [NBO1](/directory/africa-dc-nairobi-1) on Mombasa Road, the new NBO2, iColo Nairobi One under Digital Realty, iXAfrica's NBOX1 campus, and PAIX, all covered with a checklist in our [Nairobi colocation buyer's guide](/articles/colocation-data-centres-nairobi-buyers-guide). Choosing among them is about power headroom, connectivity needs, price and support, not brand slogans. Raxio's Kenya entry, when it lands, is good news even for people who will never buy from it. Nairobi's colocation market is thin on genuine carrier-neutral competition, and every new Tier III entrant tends to sharpen pricing and service across the market. The right posture is to watch, not wait: keep Raxio on the evaluation list for future tenders, and check our directory, which will update Raxio's status the day the facility opens. One last habit worth borrowing, and it works with any operator on your shortlist: ask for three things that are easy to confirm. An opening date you can verify, a Uptime Institute certificate (our [Tier ratings explainer](/articles/data-centre-tier-ratings-explained) shows what the labels mean), and the list of carriers physically on site. Established operators produce all three as a matter of routine, and asking the question itself signals that you know what to look for. That single habit will serve you well in this fast-moving market. ## Frequently asked questions ### Is Raxio open in Nairobi yet? Not as of 16 September 2026. Raxio Group has publicly targeted Kenya as a key market, and sector briefings have suggested a first Kenyan facility going live in 2026, but the first Nairobi facility has not opened yet: no launch announcement, address or commissioning report has been published so far. DC254's directory therefore lists Raxio Nairobi as early stage, and we will update it the moment that changes. Announcements turning into operating facilities is the normal rhythm of this industry; the label simply tracks where each operator stands today. ### Where is Africa Data Centres' NBO1 located? NBO1 sits at Sameer Business Park, Block A, on Mombasa Road in Nairobi. That address is confirmed by Africa Data Centres' own website and by PeeringDB's independent listing (September 2025). It is the most densely connected data centre building in Kenya, with four internet exchanges on site, which means networks meet and exchange traffic inside the building. ### Who owns Africa Data Centres? Africa Data Centres is part of Cassava Technologies, a technology group founded by Zimbabwean-born businessman Strive Masiyiwa and headquartered in South Africa. In Nairobi, the group also partnered with Digital Realty on the NBO2 facility, which launched in September 2026 with about 6.4MW of commissioned capacity reported by trade press. ### Who is behind Raxio Group? Raxio Group, established in 2018, describes itself as Africa's leading carrier-neutral Tier III data centre operator, with facilities including its Kampala site opened in 2021 and a first Angola facility launched in late 2025. In July 2026 it secured up to US$380 million to fund African expansion, and its earlier build-outs drew a USD 12.24 million guarantee from MIGA, the World Bank Group's investment insurance arm (approved September 2022). The Kenya facility is announced but not yet open. ### Which should a Kenyan business choose today? If you need rack space or compute capacity in Nairobi now, the operational options are Africa Data Centres NBO1, the new NBO2, iColo Nairobi One, iXAfrica NBOX1 and PAIX, all of which we cover in our colocation buyer's guide. Raxio is a name to watch: when it opens, carrier-neutral competition in Nairobi increases, which historically lowers prices for everyone. Until then, if Raxio comes up in a future tender, three questions will tell you most of what you need: when does it open, is it Tier III certified, and which carriers are on site? ## How to Reduce Cloud Latency for Users in Kenya URL: https://data-centers-254.vercel.app/articles/reduce-cloud-latency-kenya Cluster: Connectivity Published: 2026-09-16 Updated: 2026-09-16 Summary: Why Kenyan users feel slow apps, and how to fix it: edge locations in Nairobi, CDNs, AWS Local Zones and local hosting, explained in simple English. A Kenyan shop launches its online store. The design is sharp, the photos are light, the internet bundle is generous. Yet every tap takes a beat too long, and customers quietly drift away. The usual suspect is not the design or the bandwidth. It is distance: the server answering those taps sits in Europe or America, and every single click has to cross the equator, the Indian Ocean and back before anything happens on screen. This guide explains what latency is, why it punishes Kenyan apps hosted abroad, and the practical fixes available in 2026, from CDNs with Nairobi presence to an AWS Local Zone in the city and good old local hosting. Every technical term is unpacked in plain language as we go, because you should not need a networking degree to make an app feel fast. ![Fibre optic cables glowing inside network equipment](/images/dc-fibre-optics.webp) ## Latency, explained like a road trip Latency is the time a small piece of data takes to travel from a user's device to a server and back. It is measured in milliseconds (a millisecond is one thousandth of a second, so these are genuinely tiny numbers that add up fast). The key idea is the round trip. When a user taps a button, the phone sends a request to the server, the server thinks, and an answer comes back. That full there-and-back journey is what latency measures, the way you would time a motorbike sent to the shop and back. The motorbike is fast; the road is long. What decides the length of the road? Two things. First, physical distance: data in fibre optic cables moves at roughly two-thirds the speed of light, so thousands of kilometres simply cost time no engineer can delete. Second, hops: data passes through routers and switches (think of them as junctions) on the way, and every junction adds a small pause. Nairobi to London crosses tens of junctions and a submarine cable. Nairobi to Nairobi crosses a handful of junctions in one city. This is why fast internet and fast apps are not the same thing. Bandwidth, the number people see in internet adverts, is how much data can move at once, the width of the road. Latency is how long one round trip takes, the length of the road. Widen the road all you like; if the destination is far away, every trip still takes its time. Our guide to [Kenya internet speeds and what they mean](/articles/kenya-internet-speeds-data-centres) covers that distinction in more depth. ![Map of submarine cables connecting Africa to the world](/images/submarine-cables-map.webp) ## Why Kenyan apps hosted abroad feel slow When your server sits in Europe, every user action pays the distance tax again and again. Modern apps do not make one trip per page; they make dozens. Load the page, fetch the images, check the login, query the database, send the analytics. If each round trip costs, say, 150 milliseconds, thirty of them stack up into seconds of pure waiting, even on a full-bar fibre connection. That is the "international latency" feeling Kenyan users know too well. Three structural facts shape this in Kenya. One, our international traffic rides on [submarine cables that land in Mombasa](/articles/submarine-cables-landing-mombasa), which is excellent, but the destination still matters more than the pipe. Two, the local half of the journey is in good shape: fibre has pushed deep into Nairobi, so trips inside the city are short and quick. Three, the fixes have finally arrived locally: content delivery networks keep copies of your files in Nairobi, and cloud infrastructure now exists in the city itself. ## Fix one: put copies of your content in Nairobi (a CDN) A CDN, short for content delivery network, is a company with buildings full of servers in hundreds of cities. You point your website at the CDN, and it keeps copies of your static files (images, videos, style sheets, scripts) in cities near your users. Those stored copies are called a cache, and the buildings are called edge locations, edge because they sit at the edge of the network, close to users rather than close to your main server. The savings are direct. If your images load from Nairobi instead of Frankfurt, the user's phone collects them in a few milliseconds instead of a long round trip. Only the first user in a city pays the full journey; everyone after that gets the local copy. For most Kenyan websites, a CDN is the cheapest, fastest latency win available, and it does not require moving your application at all. The Kenyan edge map is real now. CDNPlanet's Kenya listings (fetched 16 September 2026) show Cloudflare with points of presence in Nairobi and Mombasa, Amazon CloudFront with an edge location in Nairobi, EdgeNext in Nairobi, and CDNetworks around Nairobi and Kiambu. Cloudflare's own blog records adding Nairobi to its network back in June 2021. What that means in practice: the big global CDNs can serve your Kenyan users from inside Kenya today, and switching one on is usually a settings change, not a rebuild. ![Fibre internet advertising billboard on a Nairobi street](/images/fibre-billboard-nairobi-street.webp) ## Fix two: run the app itself closer (Local Zones and local hosting) A CDN accelerates files, but the application still thinks wherever its server lives. For apps where every millisecond counts (chat, gaming, trading tools, anything with a database in the loop), the fix is to run the application close to the user. Two roads lead there from Kenya. The first is an edge cloud. AWS defines Local Zones as small outposts that extend a cloud region into a metro city, running core services like compute, storage and databases near users, with a design goal of single-digit millisecond latency (AWS, 2020). AWS announced Nairobi among 32 new Local Zone metros in February 2022 (Business Wire, 16 February 2022; Data Center Dynamics, 17 February 2022), and its official locations page lists the Nairobi, Kenya Zone today, attached to the Africa (Cape Town) region as its parent (fetched 16 September 2026). In plain terms: part of AWS now sits in Nairobi, close enough for the snappy round trips that used to require hosting in Europe. If your workloads already run on AWS, this is the road of least change. There is also a private pipe option: in September 2025 AWS opened a Direct Connect location inside East African Data Centres NBO1 near Nairobi (AWS announcement, 3 September 2025). Direct Connect is AWS's dedicated private link between your own equipment and AWS, bypassing the public internet entirely, which buys predictable latency and bandwidth for hybrid setups where part of the workload sits in Kenya and the rest in a full AWS region. The second road is Kenyan hosting outright: rent a VPS in a Nairobi data centre, or place your own servers in one. A VPS (virtual private server) is a slice of a machine you rent monthly, and our guide to [VPS hosting in Kenya paid with M-Pesa](/articles/vps-hosting-kenya-mpesa-payment) lists verified local providers. Hosting fully inside Nairobi keeps the entire round trip local, and it pairs naturally with [peering and interconnection in Kenya](/articles/data-centre-interconnection-peering-kenya), where networks exchange traffic directly inside the same buildings instead of detouring abroad. This is the deeper story of [edge computing in East Africa](/articles/edge-computing-east-africa): compute moving toward users, one facility at a time. ![Diagram explaining how network protocols carry data](/images/diagram-network-protocols.webp) ## Measure first, then fix Before spending money, spend ten minutes finding out where the delay actually lives. Three free tools do this, and all of them run from a laptop or even a phone app. **Ping** sends one tiny packet to a server and times the round trip. It answers the basic question: how long is the road? Run it against your current server from a Kenyan connection and write the number down. **Traceroute** lists every hop between you and the server, one line per router, with the time each hop took. It answers where the delay appears: inside Kenya, at the submarine cable landing, or at the destination's door. Sudden jumps tell you which leg of the journey is the expensive one. **MTR** (a blend of the two, short for my traceroute) runs continuously and sums the pattern, which exposes the hop that is consistently slow rather than occasionally busy. The comparison that matters: ping your European server, then ping a Nairobi edge location or a local VPS. The gap between those two numbers is exactly what edge hosting buys you. If your users complain about slowness and the numbers show a long international round trip, you have your answer in writing. ## A practical order of attack For most Kenyan teams, the cheapest wins come first. Switch on a CDN with Nairobi presence and let it cache your images and files; this alone can transform perceived speed, and it usually costs little more than the bandwidth you already pay for. Second, move anything genuinely latency-sensitive closer, either onto the Nairobi Local Zone if you are already on AWS, or onto a local VPS or colocation slot if you are not. Third, trim the round trips themselves: fewer redirects, fewer third-party scripts, database queries that need one trip instead of ten. Distance is the tax you cannot repeal; the number of trips is the tax you control. Latency used to be an unsolvable tax for Kenyan builders, a price of geography. In 2026 it is a design choice: cache in Nairobi, host in Nairobi where it counts, and measure the difference. The ocean is still far away, but the edge is no longer. ## Frequently asked questions ### What is latency in simple terms? Latency is the time it takes for a small piece of data to travel from a user's device to a server and back again. It is measured in milliseconds (thousandths of a second), and it is mostly decided by distance and the number of network devices the data passes through. A round trip from Nairobi to a server in Europe crosses thousands of kilometres plus submarine cables, so it collects far more delay than a trip to a server in Nairobi. ### Why does my app feel slow in Kenya even with fast internet? Because bandwidth and latency are different things. Bandwidth is how much data can move at once, like the width of a pipe. Latency is how long each round trip takes, like the length of the pipe. A fast fibre connection can still feel sluggish if every tap has to travel to a server in Europe and back before anything happens on screen. The fix is to shorten the trip, not just buy more bandwidth. ### Does AWS have servers in Kenya? AWS lists a Local Zone in Nairobi on its official Local Zones locations page (fetched 16 September 2026), with Cape Town as the parent region. A Local Zone is a small AWS outpost in a metro city that runs core services like compute and storage close to users, with a design goal of single-digit millisecond latency. AWS announced Nairobi among 32 new Local Zone metros in February 2022. Since September 2025 there is also an AWS Direct Connect location inside East African Data Centres NBO1 near Nairobi, a private network pipe straight into AWS. For bigger workloads, the nearest full AWS regions remain outside Kenya, so many teams mix Local Zones, Direct Connect or local hosts with a main region. ### What is a CDN and does it help in Kenya? A CDN (content delivery network) is a company that keeps copies of your static files, images, videos and pages, in many cities at once. Those copies are called a cache, and the buildings they sit in are edge locations. When a Kenyan user opens your site, the files come from Nairobi instead of Europe, which saves the whole long round trip. CDNPlanet lists Cloudflare with points of presence in Nairobi and Mombasa, Amazon CloudFront in Nairobi, EdgeNext in Nairobi and CDNetworks around Nairobi and Kiambu (fetched 16 September 2026), so yes, CDNs now have real Kenyan presence. ### How can I measure my own latency? Three free tools cover it. Ping sends one small packet to a server and reports the round-trip time in milliseconds. Traceroute lists every hop, every router, between you and the server, so you can see where delay appears. MTR combines both and keeps running, which makes patterns obvious. Run them from Kenya against your server, then against a Nairobi edge or a local VPS, and compare the numbers. The difference is what edge hosting would save you. ## Safaricom Data Centres: Red Hill, Waiyaki Way and M-Pesa URL: https://data-centers-254.vercel.app/articles/safaricom-data-centre-operations-kenya Cluster: Operators Published: 2026-08-29 Updated: 2026-09-28 Summary: Where does M-Pesa actually run? Inside Safaricom's Waiyaki Way hub and the Red Hill campus in Limuru, and how they differ from carrier-neutral Kenya. Safaricom is not a data centre company. It is Kenya's largest telecommunications operator, serving over 30 million mobile subscribers and generating revenue exceeding KES 300 billion annually. Yet within its portfolio sits what is arguably the single most critical data centre in East Africa: a Tier III facility on Waiyaki Way in Nairobi that processes over KES 20 trillion in M-Pesa transactions each year, hosts the core of Kenya's 4G and 5G mobile networks, and underpins an expanding portfolio of enterprise cloud services. No other data centre in the region carries this level of systemic importance to a national economy. ![Engineer working at server racks in a Kenyan facility](/images/technician-rack-cabling-kenya.webp) This profile examines Safaricom's data centre operations, the workloads they support, the company's role in Kenya's emerging AI infrastructure ecosystem, and how its facility strategy differs fundamentally from the carrier-neutral operators entering the Kenyan market. ## Safaricom's Position in Kenya's Digital Ecosystem To understand Safaricom's data centre strategy, you must first understand what Safaricom is. The company is a vertically integrated telecommunications and digital services provider with a market share of approximately 65% among Kenya's mobile subscribers. Its M-Pesa mobile money platform alone processes transactions equivalent to roughly half of Kenya's GDP, making it not just a commercial service but a piece of national critical infrastructure comparable to the payment rails of a central bank. Safaricom's business has evolved significantly beyond voice and SMS. The company now operates across mobile voice and data, M-Pesa financial services, enterprise and cloud services (Safaricom Cloud), fixed-line broadband, and a growing fibre-to-the-home segment. Each of these business lines generates substantial compute, storage, and network infrastructure demands that must be served from within Kenya, both for latency reasons and to comply with data localisation requirements under Kenya's Data Protection Act. This is the fundamental difference between Safaricom and a carrier-neutral colocation provider like iXAfrica or Africa Data Centres. Those companies build facilities and sell capacity to multiple tenants. Safaricom builds facilities to serve its own workload first and foremost. Its data centre exists because its business requires it, not because operating data centres is the business itself. ## The Waiyaki Way Data Centre: Specifications and Role Safaricom's primary data centre facility is located along Waiyaki Way in Nairobi's Westlands area. The facility holds a Tier III certification from the Uptime Institute, meaning it is designed to be concurrently maintainable with no single points of failure in the power and cooling distribution paths. This is the same reliability standard that major enterprise customers expect from commercial colocation providers, and Safaricom has invested in meeting it because the cost of downtime to M-Pesa alone would be measured in billions of shillings per hour. Key specifications of the Waiyaki Way facility include: - **Tier classification:** Tier III (Uptime Institute certified) - **IT power capacity:** approximately 1.5 MW - **Rack capacity:** approximately 120 racks - **Cooling:** precision air conditioning with N+1 redundancy - **Power backup:** diesel generators with on-site fuel storage, UPS systems for seamless switchover - **Connectivity:** direct access to Safaricom's extensive terrestrial fibre backbone - **Security:** multi-layered physical security including biometric access controls, CCTV surveillance, and on-site security personnel At 1.5 MW and 120 racks, the Waiyaki Way facility is modest by comparison to the hyperscale facilities being developed by iXAfrica (targeting 16 MW across its campus) or Africa Data Centres (planning 30 MW at its Mzansi campus). But raw capacity is not the metric that matters here. What matters is the criticality of the workloads running inside those 120 racks. ![Transmission tower beside a commercial high-rise](/images/dc-challenges.webp) ## Workloads: M-Pesa, Mobile Network Core, and Enterprise Cloud The Waiyaki Way data centre supports three primary workload categories, each of which would justify a dedicated facility on its own. ![Labelled anatomy of a telecommunications mast](/images/diagram-telecom-mast-anatomy.webp) ### M-Pesa Transaction Processing M-Pesa is the dominant mobile money platform in Kenya, with over 51 million active customers across seven countries. In Kenya alone, the platform processes transactions valued at more than KES 20 trillion annually. Every single one of those transactions requires real-time authentication, ledger updates, and settlement processing. The M-Pesa platform runs on a distributed architecture, but the core transaction processing and database layers are housed within Safaricom's Waiyaki Way facility. The physical closeness of those systems is what makes the service feel instant. When a subscriber checks a balance or sends money, the request travels from the handset to the network core and into the Waiyaki Way facility, and a response comes back before the person holding the phone has registered a delay. The figure sometimes quoted (that a full M-Pesa round trip completes in under two seconds) is an estimate, not a published specification: Safaricom does not release a formal USSD latency service-level target, so DC254 derives this range from reported USSD session behaviour and typical transaction throughput during peak periods. Treat it as an order-of-magnitude answer to the question "how fast is fast enough to move half of Kenya's GDP" rather than an audited number. Our methodology page explains how estimates like this are handled and labelled, and how they differ from the verified figures in the DC Directory. The reliability requirements are extraordinary. M-Pesa operates effectively as Kenya's primary payments infrastructure. A prolonged outage would not just inconvenience Safaricom's customers; it would halt commerce across the country, from informal sector traders in Nairobi's markets to salary disbursements for millions of employed Kenyans. This is why Safaricom has invested in Tier III redundancy and maintains robust disaster recovery capabilities, including secondary sites that can assume the primary processing role in the event of a facility-level failure. ### Mobile Network Core: 4G LTE and 5G Safaricom's mobile network core, the intelligence layer that manages subscriber authentication, session control, policy enforcement, and handover between cell sites, runs from the Waiyaki Way data centre. With over 4,000 base stations across Kenya and an expanding 5G footprint that now covers parts of Nairobi, Mombasa, and Kisumu, the network core must process millions of signalling messages and bearer sessions every second. The 5G rollout has added new demands on the facility. 5G network functions, particularly in the core network, are increasingly containerised and deployed using cloud-native architectures. This has driven Safaricom to modernise its data centre infrastructure to support Kubernetes orchestration, software-defined networking, and distributed storage systems that were not part of the original facility design. The transition from traditional 3G/4G network functions running on dedicated hardware to virtualised and containerised network functions is a multi-year programme that touches every layer of the data centre stack. ### Safaricom Cloud and Enterprise Services Safaricom has been building an enterprise services business that directly competes with global cloud providers and local IT companies. Safaricom Cloud offers infrastructure-as-a-service, platform-as-a-service, and software-as-a-service products to Kenyan enterprises, with the value proposition of local data residency, low latency, and integration with Safaricom's network infrastructure. Enterprise customers using Safaricom Cloud benefit from data localisation, a requirement under the Kenya Data Protection Act for certain categories of personal data. Hosting workloads within Safaricom's facility, rather than routing them to cloud regions in South Africa or Europe, reduces latency for local users and ensures regulatory compliance. Key enterprise customers include banks, county governments, and large corporations that need local infrastructure with the reliability guarantees that only a Tier III facility can provide. ## Not Carrier-Neutral: A Deliberate Strategic Choice One of the most important distinctions about Safaricom's data centre is what it is not: it is not carrier-neutral. Unlike iXAfrica, Africa Data Centres, or Wingu Africa, Safaricom's facility does not operate Meet-Me Rooms where multiple telecom operators, internet service providers, and cloud platforms interconnect. There is no open access policy for third-party networks. You cannot, as an independent enterprise, purchase rack space and bring your own connectivity. This is a deliberate strategic choice, not an oversight. Safaricom's data centre exists to serve Safaricom's workloads. Opening the facility to competing network operators or cloud providers would create conflicts of interest and potential security concerns, particularly given the sensitivity of the M-Pesa platform and the mobile network core. The company has no commercial incentive to help competitors or third parties interconnect within its facility. For enterprises that require carrier-neutral colocation with access to multiple network providers, the commercial data centre operators in Nairobi remain the appropriate choice. Safaricom's facility serves a different function: it is the private backbone of Kenya's most important digital services company. ## The Microsoft-G42 AI Data Centre Partnership In May 2024, Microsoft and United Arab Emirates-based technology group G42 announced a landmark $1 billion investment to build a state-of-the-art AI data centre in Kenya. The announcement, made during a visit by UAE President Sheikh Mohamed bin Zayed Al Nahyan to Nairobi, named Safaricom as the local partner for the project. It would be the largest single foreign direct investment in Kenya's digital infrastructure. The proposed facility was designed to provide AI compute capacity to serve not just Kenya but the broader East African region. G42, which has emerged as a significant player in AI infrastructure globally, would bring technical expertise and investment capital. Microsoft would provide Azure AI services and cloud integration. Safaricom would contribute its local market position, regulatory relationships, fibre infrastructure, and operational capabilities within Kenya. ### Why the Project Is Stalled As of mid-2026, the project has not broken ground. The primary bottleneck is power. The proposed AI data centre would require in excess of 100 MW of dedicated power capacity, a demand that Kenya Power and Lighting Company (KPLC) is currently unable to deliver to a single site. While Kenya has an installed generation capacity of over 3,000 MW, the transmission and distribution infrastructure is not designed to deliver concentrated loads of this magnitude to a single customer at a single location. This power constraint is not unique to the Microsoft-G42 project. It is the single most significant barrier to hyperscale data centre development in Kenya. Our analysis of [Kenya's power infrastructure as it relates to data centres](/articles/kenya-power-infrastructure-data-centres) covers this challenge in detail. The constraint affects every operator and prospective investor in the Kenyan market, but it is particularly acute for AI-scale facilities that require power densities an order of magnitude higher than traditional enterprise data centres. Possible pathways forward include dedicated transmission lines from geothermal plants in the Rift Valley, on-site renewable generation with battery storage, or siting the facility where the grid can accommodate high-density loads. Each option involves significant capital, permitting, and timelines measured in years. ### Safaricom's Role as Local Partner Safaricom's selection as the local partner for the Microsoft-G42 project reflects its unique position in the Kenyan market. No other company in East Africa combines Safaricom's regulatory relationships, existing fibre infrastructure spanning thousands of kilometres, enterprise customer base, and experience operating mission-critical data centre infrastructure. Even with the project stalled, the partnership signals a strategic direction for Safaricom. The company is positioning itself not merely as a telecom operator with a data centre, but as a infrastructure platform company capable of hosting and enabling the next generation of AI workloads in East Africa. If the power challenges are resolved, Safaricom would transition from operating a single 1.5 MW facility to having access to and operational involvement in what could be the largest AI computing platform on the African continent. ## Safaricom's Data Centre Strategy Within the Broader Enterprise Business Safaricom's enterprise business has become an increasingly important revenue stream, and data centre infrastructure is central to this strategy. The company's Enterprise business unit serves large corporations, government agencies, and SMEs with a portfolio that includes managed IT services, cloud computing, cybersecurity, and connectivity solutions. The data centre underpins this entire enterprise offering. Safaricom Cloud competes with global providers by offering local hosting, data residency, and integration with Safaricom's network. Enterprise customers that purchase MPLS circuits, SD-WAN solutions, or dedicated internet access from Safaricom can have their compute workloads hosted within the same ecosystem, simplifying procurement, support, and fault management. The enterprise strategy also extends to managed hosting and application management services. Safaricom hosts applications for banks, insurance companies, and government agencies within its data centre, managing the infrastructure layer so that customers can focus on their applications. This is a higher-margin business than raw connectivity, and it directly leverages the data centre investment that Safaricom has already made. ## Future Plans and Expansion Safaricom has not publicly announced plans for additional data centre facilities beyond Waiyaki Way. However, several factors suggest expansion is likely, even without the Microsoft-G42 partnership in its original form. The growth of M-Pesa across seven countries will continue driving demand for compute and storage. New product features such as savings accounts, lending products, and merchant payments all require additional infrastructure. The 5G rollout will further strain the network core as coverage expands beyond initial urban deployments and session volumes increase. While edge computing can offload some processing, core network functions will remain centralised. The enterprise cloud business is also growing as Kenyan enterprises move workloads to local platforms, driven by data protection regulations and latency requirements. The AI compute opportunity, whether through the Microsoft-G42 partnership or other arrangements, represents a potential step change in scale. Safaricom's existing data centre expertise and market position make it a natural partner for any hyperscale infrastructure investment in Kenya. ## How Safaricom Compares to Other Kenyan Data Centre Operators Understanding Safaricom's data centre operations requires situating them within the broader Kenyan market. The following comparison highlights the key differences: - **Safaricom (Waiyaki Way):** Approximately 1.5 MW, 120 racks, Tier III, non-neutral, serves internal workloads (M-Pesa, network core, enterprise cloud). The most critical facility in the country by workload importance, but not the largest by capacity. - **iXAfrica (Mombasa Road):** 4.5 MW operational, targeting 16 MW campus, carrier-neutral, AI-ready densities up to 50 kW per rack. Serves cloud providers, enterprises, and satellite operators. The largest and most advanced commercial facility in the region. - **Africa Data Centres (Sameer Business Park and Mzansi):** Multiple facilities in Nairobi, carrier-neutral, part of the Pan-African REMCO platform. Focuses on enterprise colocation and hyperscale customers. - **Wingu Africa (Various Nairobi locations):** Smaller-scale carrier-neutral facilities focused on enterprise and SME colocation. Safaricom's facility is unique in this landscape. It is not competing for third-party colocation customers. It is not marketing itself to hyperscale cloud providers. It exists because Safaricom's business demands it, and its value to the Kenyan economy is measured not in rack prices or megawatts sold, but in the continuity of M-Pesa, the reliability of the mobile network, and the availability of local cloud services for Kenyan enterprises. ## Conclusion Safaricom's data centre on Waiyaki Way is the hidden engine of Kenya's digital economy. It does not appear in carrier-neutral interconnection rankings or commercial colocation market reports, yet it processes more financial value and serves more end users than any other facility in East Africa. The workloads it supports, from M-Pesa transactions to 5G network core functions, are so fundamental to Kenya's economic activity that its reliability is effectively a matter of national importance. Looking ahead, Safaricom's data centre strategy is at an inflection point. The Microsoft-G42 partnership, if it overcomes power constraints, would transform the company's role from infrastructure operator to infrastructure platform partner at a scale that would reshape the entire East African data centre market. Even without that partnership, the growth of M-Pesa, the expansion of 5G, and the increasing demand for local cloud services will require Safaricom to expand its data centre footprint in the coming years. For organisations exploring the [data centre directory](/directory) to understand Kenya's infrastructure landscape, Safaricom's facility represents a critical but distinctive asset: a purpose-built, non-commercial facility that is simultaneously the most important data centre in the country and the one least accessible to the open market. --- *Data Centre 254 provides independent analysis of Kenya's data centre industry. For a broader view of the market, read our analysis of [Kenya's power infrastructure challenges](/articles/kenya-power-infrastructure-data-centres) and our outlook on [AI data centres in East Africa](/articles/ai-data-centres-east-africa).* ## Solar Power for Data Centres in Kenya: The Economics URL: https://data-centers-254.vercel.app/articles/solar-power-data-centres-kenya Cluster: Renewable Energy Published: 2026-08-28 Updated: 2026-09-28 Summary: Kenya's strong solar irradiance makes PV compelling for data centres. The economics, hybrid configurations, battery storage, and feasibility. Kenya sits on the equator, receives more direct sunlight than most countries on Earth, and generates over 90% of its electricity from renewable sources. For an industry that consumes as much power as a small town, this should be a dream combination. Yet the question of whether Kenya's data centres can meaningfully harness solar power is more nuanced than simply pointing photovoltaic panels at the sky. The answer involves grid economics, battery storage costs, land availability, and the practical reality that data centres cannot afford even a second of downtime. The short answer: solar alone cannot power a Kenyan data centre, but grid-tied solar is already cheaper than grid power for operators who can install it, and it is becoming a meaningful layer of Kenya's data centre energy stack rather than its foundation. ![Workers at a large solar farm in Kenya](/images/kenya-solar-farm.webp) Solar power for data centres is not a new idea. Google, Amazon, and Microsoft have all built massive solar farms to power their data centres in the United States and Europe. What makes Kenya different is that the country's grid is already predominantly renewable, powered by geothermal energy from the Rift Valley, hydroelectric dams, and growing wind and solar installations. A Kenyan data centre running on grid power already has a lower carbon footprint than almost any competitor in Africa or the Middle East. The question, then, is not whether solar can make data centres greener, but whether it can make them cheaper, and whether the economics work at the scale that Kenyan operators are building. ## Kenya's Solar Resource Kenya's position astride the equator gives it a solar resource that ranks among the best in the world for utility-scale photovoltaic generation. The country receives an average of 4 to 6 kilowatt-hours of solar energy per square metre per day (kWh/m²/day), with the highest irradiance found in northern and eastern Kenya, the arid and semi-arid lands (ASALs) that cover roughly 80% of the country's land area. Even Nairobi, which sits at 1,795 metres above sea level and experiences occasional cloud cover, receives approximately 5 kWh/m²/day on average, which is comparable to solar leaders like Spain, California, and Northern India. The equatorial location provides another advantage that is often overlooked: seasonal consistency. Unlike higher latitudes, where solar output can vary by 50% or more between summer and winter, Kenya's solar resource is relatively stable year-round. Day length varies by only a few minutes between the longest and shortest days of the year. This consistency makes it easier to size solar systems accurately and to predict energy output, reducing the risk of over-investment or under-performance that plagues solar projects in more variable climates. ## Why Solar for Data Centres Makes Economic Sense The economic case for solar in Kenyan data centres rests on the gap between the cost of grid power and the levelised cost of solar energy. Kenya Power charges industrial and commercial customers between KES 12 and KES 18 per kilowatt-hour (approximately $0.08–$0.12), depending on the tariff category, time of use, and demand charges. The levelised cost of solar PV in Kenya, based on current panel prices, inverter costs, and installation labour, is approximately KES 6–8 per kWh ($0.04–$0.06) over a 25-year system life. This means solar can produce electricity at roughly half the cost of grid power, a compelling saving for an industry where power is the single largest operating expense. To put this in concrete terms: a 1 megawatt data centre operating at a power usage effectiveness (PUE) of 1.5 consumes approximately 1.5 megawatts of total facility power, or about 13,140 megawatt-hours per year. At KES 15/kWh from the grid, this costs approximately KES 197 million ($1.3 million) annually. A solar system sized to offset 30% of this consumption (approximately 450kW of solar PV capacity) would generate about 810 MWh per year, saving roughly KES 12 million ($80,000) annually. With an installation cost of approximately KES 50–70 million for a 450kW system, the payback period would be 4–6 years, well within the 25-year design life of the solar panels. ![Rows of solar panels stretching across a Kenyan site](/images/kenya-solar-farm-4.webp) ## Grid-Tied Hybrid: The Practical Configuration The most practical and economically viable configuration for a Kenyan data centre is a grid-tied hybrid system. In this setup, solar panels generate DC electricity during daylight hours, which is converted to AC by inverters and fed directly into the data centre's power distribution system. When solar output exceeds the data centre's demand, the excess can be exported to the grid (where net metering is available) or curtailed. When solar output is insufficient (at night, during heavy cloud cover, or during peak demand periods) the grid seamlessly supplies the balance. This hybrid approach eliminates the need for large battery banks, which are currently the most expensive component of off-grid solar systems. Battery storage costs have fallen dramatically (lithium-ion battery packs now cost approximately $120–$150 per kilowatt-hour, down from over $1,000 a decade ago) but they still represent a significant capital investment and add complexity in terms of maintenance, thermal management, and replacement cycles. The key design decision in a grid-tied hybrid system is how much solar capacity to install relative to the data centre's load. This is not a simple ratio calculation, because data centres run 24 hours a day, but solar only generates during daylight hours (approximately 6:00 AM to 6:30 PM at the equator). Oversizing the solar array means more energy is exported to the grid (at potentially unfavourable rates) rather than used directly by the data centre. Undersizing means leaving money on the table. The optimal sizing typically targets a solar offset of 20–40% of total energy consumption. This range maximises the use of solar generation during peak daylight hours without excessive curtailment or grid export. For a facility with a 1.5MW total load, this translates to approximately 300–600kW of solar PV capacity. ## Battery Storage: When It Makes Sense While grid-tied systems without batteries are the most cost-effective configuration for most Kenyan data centres today, there are scenarios where adding battery storage makes economic sense. The primary case is for facilities that want to participate in demand response programmes or reduce their peak demand charges. Kenya Power's industrial tariffs include a demand charge component based on the maximum power draw recorded during the billing period. A data centre that occasionally draws more power during peak hours (for example, when workloads spike or when cooling systems work harder during hot afternoons) pays higher demand charges for the entire billing period. A battery system can smooth out these peaks by drawing power from the grid (or solar) during low-demand periods and discharging during high-demand periods, reducing the recorded maximum demand and therefore the demand charge. A secondary case for batteries is as a short-term backup during grid outages. While data centres already have UPS systems and diesel generators for this purpose, batteries can bridge the gap between a grid failure and generator start-up more cleanly and with lower emissions than flywheel UPS systems. Lithium-ion batteries can also reduce diesel fuel consumption during extended outages by handling the first 15–30 minutes before generators take over, and by managing load transitions during generator testing. ## The Geothermal vs. Solar Question Kenya's unique position in the global renewable energy landscape raises a question that does not arise in most data centre markets: if the grid is already 90% renewable (primarily geothermal), does on-site solar provide meaningful additional environmental benefit? The baseload that anchors this question comes from [Kenya's geothermal energy](/articles/geothermal-energy-kenya-data-centres), with the [Olkaria corridor](/articles/olkaria-geothermal-data-centre-corridor) its clearest expression. The answer depends on the data centre operator's goals. From a pure carbon perspective, a Kenyan data centre running on grid power already has an exceptionally low carbon intensity. Kenya Power's generation mix produces approximately 50–100 grams of CO2 per kilowatt-hour, compared to 400–500 g/kWh for the European grid average and 800+ g/kWh for coal-heavy grids in South Africa and India. Adding solar panels on-site might reduce this to 20–50 g/kWh for the portion of consumption offset by solar, which is a meaningful improvement but not a transformative one. ![Steam rising from a Rift Valley geothermal field](/images/kenya-geothermal-plant.webp) However, the economic argument for solar remains strong regardless of the grid's greenness. Solar is cheaper than grid power in Kenya. Period. Any data centre operator who can install solar at $0.04–$0.06/kWh instead of buying grid power at $0.08–$0.12/kWh is saving money. And for international companies with corporate renewable energy targets (companies that have committed to 100% renewable energy or net-zero carbon) on-site solar provides verifiable, attributable renewable energy that is easier to report and certify than purchased grid renewables. ## Case Studies from Other African Markets While Kenya has not yet seen a large-scale solar-powered data centre, other African markets provide instructive examples. In South Africa, several data centres have installed rooftop and ground-mounted solar arrays, driven primarily by the country's load shedding crisis and high grid costs. Teraco, Africa's largest colocation provider, has installed solar at multiple facilities and has signed power purchase agreements (PPAs) with independent solar producers. In Ghana, a pilot project at the Ghana-India Kofi Annan Centre of Excellence in ICT demonstrated a 100kW solar installation supporting a small data centre, achieving 25–30% solar offset with a grid-tied configuration. These examples confirm that the technology works in African conditions and that the economics are viable. Kenya's advantage over both South Africa and Ghana is its more reliable grid, which makes grid-tied systems more practical (there is less need for battery backup to compensate for load shedding) and lower grid costs (which means the savings from solar, while real, are a smaller percentage of total costs than in markets with very expensive grid power). ## Land Requirements and Site Selection One of the practical constraints on solar for data centres is land. A 1MW solar PV array in Kenya, using modern monocrystalline panels with approximately 20–22% efficiency, requires approximately 2–3 hectares (5–7 acres) of land when accounting for panel spacing, access paths for maintenance, inverter pads, and perimeter security. A 5MW array (enough to offset a significant portion of a large data centre's consumption) needs 10–15 hectares. This land requirement has implications for data centre site selection. Facilities in Nairobi's congested industrial areas (along Mombasa Road, in Enterprise Road, or in South B) typically sit on 1–2 acre plots and have no room for meaningful solar installations. Newer facilities being built on larger sites (like iXAfrica's campus, which has room for expansion, or potential sites in Konza Technopolis) are better positioned to incorporate solar from the design stage. ## The Path Forward Solar power for Kenyan data centres is not a question of if, but of when and how fast. The economics are already favourable, the solar resource is excellent, and the regulatory framework (net metering, renewable energy targets) is supportive. The primary barriers are land availability at existing sites, the capital investment required, and the fact that grid power is already relatively clean and reasonably priced, which reduces the urgency compared to markets with dirty or unreliable grids. The most likely trajectory is that new data centre projects (those not yet built or in early planning stages) will incorporate solar from the design phase, allocating land and electrical infrastructure for 20–30% solar offset. Existing facilities will add solar incrementally, starting with carport installations (solar panels above parking areas, which require no additional land) and rooftop arrays where structural capacity allows. Within five years, it is reasonable to expect that most new Kenyan data centres will have some form of on-site solar generation, and that solar offset will be a standard feature in colocation marketing materials. ![Technician installing solar panels](/images/kenya-solar-farm-5.webp) For Kenya, solar-powered data centres represent an opportunity to reinforce its positioning as Africa's greenest digital infrastructure hub. In a world where environmental, social, and governance (ESG) considerations are increasingly influencing investment decisions, the combination of geothermal grid power and on-site solar could make Kenyan data centres some of the most environmentally attractive in the world, a genuine competitive advantage in the global market for digital infrastructure investment. ## Starlink in Kenya: What It Means for Data Centres URL: https://data-centers-254.vercel.app/articles/starlink-kenya-data-centres-backup-link Cluster: Connectivity Published: 2026-09-16 Updated: 2026-09-18 Summary: Starlink reached 25,000 Kenyan subscribers by March 2026. How satellite internet fits enterprise networks, backup links and disaster recovery, and where fibre still wins. When Starlink switched on in Kenya in July 2023, the conversation was about homes: rural families, remote workers, the places fibre had never reached. Three years later the more interesting story is enterprise. Kenya has 25,000 Starlink dishes as of March 2026 (Communications Authority data via Tech-ish, 19 June 2026), a licensing regime that just made satellites considerably more expensive to operate (Techlabari, 8 April 2026), a network that hit capacity limits in July 2026 (TechCabal, 7 July 2026), and a growing number of businesses asking the question this article answers: what is satellite actually for, in a country whose data centres run on fibre and undersea cables? ![Starlink dish and Wi-Fi router installed indoors, city skyline outside the window](/images/starlink-dish-router-indoor.webp) ## The numbers: fast growth, small share, real limits Start with the facts. Starlink launched in Kenya in July 2023 (The Kenya Times, 23 September 2025) and recorded nearly 2,000 percent subscriber growth in its first year, with aggressive pricing that put a mini kit at KSh 27,000 and promotional monthly service around KSh 1,300 by October 2024 (Space in Africa, 14 October 2024). Growth then moderated: 7,933 net additions in the year to March 2026, against 12,562 the year before (Space in Africa, 21 June 2026). The base stood at 24,999 subscribers in March 2026, about 0.9 percent of fixed internet (Tech-ish, 19 June 2026), and in July 2026 the company limited new signups as the Kenyan network reached capacity (TechCabal, 7 July 2026). Read those numbers like a data centre operator would. A 0.9 percent share is not a fibre competitor; Kenya's fixed market is fibre-first and the big providers kept their customers by cutting prices and improving packages. But 25,000 diverse, independent, non-terrestrial links is a meaningful resilience layer for the enterprises that bought them, and the July 2026 capacity clamp is the reminder that satellite capacity is a shared, finite resource in a way a fibre pair is not. ![Illustration of the Starlink kit box with dish, router and cabling](/images/starlink-kit-box.webp) ![Diagram of satellite internet connection types from dish to terminal](/images/diagram-satellite-connector-types.webp) ## What Starlink is actually good for in enterprise Kenya The honest use cases fall into three groups. **Backup and out-of-band management.** A business whose primary link rides one fibre route has a single point of failure the provider's marketing will not fix. A satellite terminal on the roof is a physically separate path: different medium, different provider, different failure modes. Network engineers value it even more as an out-of-band management connection, the path you use to reach equipment when the main network is exactly what is broken. After Kenya's 2024 undersea cable cuts forced businesses to activate business continuity plans for weeks at a time (Citizen Digital, 14 May 2024; Internet Society, 23 July 2024), the case for diverse links stopped being theoretical. **Sites fibre will never reach economically.** Kenya's wind farms, mining operations, conservancies, border posts and rural bank branches sit far from any trench. The economics that make a Makueni fibre pole line viable for a trading centre make it absurd for a single remote site, which is why satellite terminals are becoming standard kit for remote industrial and public-sector locations (Mawingu, 27 February 2026). The February 2026 announcement that Mawingu would lead a Microsoft-Starlink rural connectivity initiative in Kenya signals exactly this: satellite plus local fixed wireless, reaching where trenching cannot. **Rapid deployment.** A branch that opens Monday, an election centre, a disaster response camp, a construction site office: satellite delivers a working link in hours, not the months a wayleave and trench require. For Kenyan organisations that operate where infrastructure is seasonal or unstable, that speed is the product. ![Technicians erecting a fibre distribution pole in rural Makueni](/images/last-mile-pole-erection-makueni.webp) ## Where fibre and data centres keep winning The comparison matters most at the top of the market. A data centre's product is dense interconnection: dozens of carriers, cross-connects between tenants, peering at the Kenya Internet Exchange Point, and capacity measured in gigabits per second delivered over diverse physical routes. Satellite serves none of that. Its capacity per link is modest, it is shared with every other terminal in the beam, weather degrades it, and its latency, while far better than old satellite, still sits above metro fibre's single-digit milliseconds. The right mental model is complementary layers, not substitutes. Fibre carries the workload; satellite carries the lifeline. A Kenyan enterprise running its systems in a Nairobi colocation facility, with diverse fibre entries and a satellite terminal for management and emergency access, has a network posture most banks would call healthy. The same enterprise with only a satellite link has a consumer product with a business invoice. The regulatory picture sharpens that divide. The Communications Authority's April 2026 licensing overhaul lifted satellite licence fees to as much as KES 45 million, roughly 28 times the previous level (Techlabari, 8 April 2026; ITWeb Africa, 9 April 2026), a signal that Kenya intends to treat satellite as one input among many in a competitive market, not a bypass around it. ![Wind turbines at a power installation in Kenya](/images/kenya-wind-turbines-4.webp) ## What to watch next Three threads are worth following. First, capacity: the July 2026 signup limits will ease or they will not, and either outcome tells enterprise buyers how dependable satellite is as a lifeline at scale. Second, price: every Starlink price cut in Kenya was answered by fibre providers, and the discipline runs both ways. Third, the hybrid model: the Microsoft-backed rural deployments with local partners like Mawingu suggest the future Kenyan network is a braid, fibre for density, fixed wireless for the middle mile, satellite for the margins. For the data centre industry, the takeaway is short. Starlink does not move a single rack into or out of a Kenyan facility. It does make every facility and every enterprise network more survivable, and in a country that has watched cable cuts and national DDoS attacks disrupt the internet within living memory, survivable is a feature worth paying for. ## Starting a Data Centre Business in Kenya URL: https://data-centers-254.vercel.app/articles/starting-data-centre-business-kenya Cluster: Business Guide Published: 2026-08-28 Updated: 2026-08-28 Summary: Starting a data centre company in Kenya takes KES 500M-2B+, an NFP-T2 licence, land, and a market strategy. The investment and regulatory steps. Building a data centre is one of the most capital-intensive infrastructure investments possible in Kenya's technology sector. A small colocation facility requires hundreds of millions of shillings in upfront capital, takes 18-36 months from concept to operational, and generates returns over a 15-25 year asset life. It is not a startup in the traditional sense, it is an infrastructure play that requires patience, deep pockets, and a clear understanding of the market. ![Nairobi City Hall, seat of the Nairobi City County government](/images/nairobi-city-hall-county.webp) This guide is for investors, entrepreneurs, and organisations considering entering the Kenyan data centre market. It covers the investment requirements, the regulatory steps, the market dynamics, and the strategic decisions that determine whether a new data centre succeeds or fails. ## Understanding the Market Opportunity Kenya's data centre market, as detailed in our [market outlook analysis](/articles/kenya-data-centre-market-outlook-2025-2030), is growing at 15-20% annually and is projected to reach $400-500 million by 2030. The current supply of approximately 15 megawatts of IT power capacity is expected to need 25-30 megawatts by 2030, creating a gap that new entrants can fill. But market growth alone does not guarantee success for a new entrant. The existing operators (iXAfrica, Africa Data Centres, and Safaricom) have established customer relationships, brand recognition, and the operational expertise that comes from running facilities in Kenya. A new entrant must offer something they do not: a different customer segment, a different location, a different technology approach, or a different pricing model. ## Capital Requirements ### Construction Costs The cost of building a data centre in Kenya varies depending on the size, tier rating, and specifications. As a rough guide, construction costs (excluding land) range from KES 30,000-60,000 per square metre for the building shell, and KES 100,000-200,000 per square metre when fully fitted out with power, cooling, fire suppression, and security systems. For a typical small colocation facility of 500-800 square metres of white space (enough for 100-200 racks), the total construction cost is approximately KES 300-600 million. For a larger facility of 1,500-2,500 square metres (enough for 300-600 racks), the cost rises to KES 800 million to 2 billion. ### Power Infrastructure ![Aerial view of a data centre campus with rooftop solar](/images/dc-environment-sustainability.webp) Power infrastructure typically accounts for 30-40% of total construction costs. This includes the high-voltage connection from Kenya Power (which the operator must pay for), transformers, UPS systems, diesel generators, fuel storage tanks, and the power distribution system (switchgear, PDUs, and cabling). A 2MW facility with N+1 generator redundancy might spend KES 150-300 million on power infrastructure alone. ### Cooling Infrastructure Cooling infrastructure accounts for 15-20% of construction costs. This includes precision air conditioning units, ductwork, containment systems, and the building management system that controls temperature and humidity. Cooling costs are lower in Nairobi than in hotter climates (thanks to the moderate ambient temperature and altitude), which is a cost advantage for Kenyan facilities. ### Land Land costs vary enormously by location. Along Mombasa Road, industrial land suitable for a data centre costs approximately KES 100-200 million per acre. In Konza Technopolis or other areas further from Nairobi, land may cost KES 20-50 million per acre. A small data centre needs 0.5-1 acre; a larger campus needs 2-5 acres. ### Working Capital In addition to construction costs, a new data centre operator needs working capital to fund operations during the ramp-up period, typically 12-24 months before the facility reaches breakeven occupancy. This includes staff salaries, utility bills, marketing, and insurance. Working capital of KES 100-300 million should be budgeted. ### Total Investment Summary | Facility Size | IT Load | Construction | Land | Working Capital | Total | |--------------|---------|-------------|------|-----------------|-------| | Small | 1-2 MW | KES 300-600M | KES 50-200M | KES 100-150M | KES 450-950M | | Medium | 3-5 MW | KES 800M-1.5B | KES 100-400M | KES 150-300M | KES 1-2.2B | | Large | 10+ MW | KES 3-5B+ | KES 200-1B | KES 300-500M | KES 3.5-6.5B+ | ## Regulatory Requirements ### Communications Authority Licence The [NFP-T2 licence](/articles/kenya-data-centre-licensing-framework) from the Communications Authority of Kenya is the primary regulatory requirement. The licence costs KES 15 million for a 15-year term and requires the operator to meet specific technical and operational standards. The application process involves submitting technical plans, financial statements, and evidence of compliance with CA requirements. ### Construction Approvals Building a data centre requires standard construction approvals from the relevant county government (Nairobi City County for facilities in Nairobi), including building plans approval, environmental impact assessment (EIA) from the National Environment Management Authority (NEMA), and fire safety certification from the Kenya Fire Service. ![National government building at dusk](/images/dc-policy-regulation.webp) ### Kenya Power Connection Connecting to the Kenya Power grid requires a formal application and agreement process. For a data centre, this typically involves a high-voltage connection (11kV or 33kV), which requires the operator to fund the connection infrastructure (transformers, cabling, and substation equipment). The connection process can take 6-18 months, depending on the location and the required capacity. It is critical to engage Kenya Power early in the project timeline. ### Data Protection Registration As discussed in [our Data Protection Act guide](/articles/kenya-data-protection-act-data-centres), data centre operators must register as data processors with the Office of the Data Protection Commissioner (ODPC). This is a straightforward process but should be completed before commercial operations begin. ## Market Entry Strategies ### Strategy 1: Carrier-Neutral Colocation This is the model used by iXAfrica and Africa Data Centres. Build a Tier III carrier-neutral facility and offer colocation, interconnection, and managed services to multiple customers. This is the most proven model in Kenya but also the most competitive, as the incumbents have established market positions. Differentiation options for a new colocation entrant include: targeting a specific customer segment (SMEs, government, or a specific industry like healthcare), offering a different price point (lower cost through innovative design), or specialising in a technology (GPU-ready zones for AI, or high-security zones for government and defence). ### Strategy 2: Hybrid Cloud Hub Build a facility designed specifically as a hub for hybrid cloud deployments, where customers run some workloads in the data centre and some in the public cloud, with dedicated, high-speed connections between the two. This model targets enterprises that want data residency and low latency for sensitive workloads while leveraging cloud elasticity for other workloads. The facility would include Direct Connect/ExpressRoute capabilities for all major cloud providers. ### Strategy 3: Edge and Regional ![Nairobi Expressway stretching across the city](/images/nairobi-expressway-wide.webp) Build smaller facilities (100-500kW) in locations outside Nairobi (Mombasa, Kisumu, Nakuru, or Konza Technopolis) serving regional demand and edge computing use cases. This strategy avoids direct competition with the Nairobi incumbents and serves a market that is currently underserved. The challenge is that demand in secondary cities is smaller and less proven. ### Strategy 4: Specialised Facility Build a facility optimised for a specific use case: GPU computing for AI, high-security for government, or disaster recovery for financial services. Specialised facilities can command premium pricing and face less competition because they serve a defined market segment that general-purpose facilities cannot serve well. ## Revenue Model and Financial Projections Colocation revenue comes from recurring monthly charges for rack space, power, and connectivity. A typical pricing model for a new Kenyan facility might be: - Full rack (42U, 5kW): KES 80,000-120,000/month - Half rack (21U, 2.5kW): KES 45,000-70,000/month - Power over base allocation: KES 20-25/kWh - Cross-connects: KES 5,000-15,000 one-time - Remote hands: KES 2,000-5,000/incident For a 200-rack facility at an average of KES 70,000 per rack per month and 70% occupancy (140 racks occupied), monthly colocation revenue would be approximately KES 9.8 million (KES 118 million annually). Adding interconnection revenue, remote hands, and other services, total annual revenue might reach KES 150-200 million. Operating expenses (power, staff, maintenance, insurance, connectivity costs, property costs) typically total 50-60% of revenue for a well-run facility. Net operating income of KES 60-100 million on a KES 500 million to 1 billion investment implies a 6-20% return on investment, attractive for infrastructure, though the payback period is 5-10 years. ## Key Success Factors The data centre business rewards operational excellence. The facilities that succeed are those that deliver reliable power and cooling, provide responsive customer service, maintain high security standards, and continuously invest in infrastructure upgrades. In Kenya's market, where the customer base is relatively small and relationships matter enormously, reputation and trust are critical differentiators. For any investor seriously considering the Kenyan data centre market, the recommendation is to start with thorough market research, engage experienced data centre design and operations consultants, secure land and power commitments early, and plan for a 10-15 year investment horizon. The opportunity is real and growing, but so is the competition, and only well-executed projects will capture their share of this expanding market. ## Which Submarine Cables Land at Mombasa? 7 Live Systems URL: https://data-centers-254.vercel.app/articles/submarine-cables-landing-mombasa Cluster: Connectivity Published: 2026-08-26 Updated: 2026-09-28 Summary: TEAMS, SEACOM, EASSy, LION2, DARE1, 2Africa and PEACE land at Mombasa. Landing stations, capacities, and why Africa-1 is not counted as live yet. Mombasa sits at the eastern edge of Kenya's coastline, and beneath its waters run the fibre optic cables that connect over 300 million people in East and Central Africa to the global internet. Seven live submarine cable systems land at stations along the Mombasa coast, with two more landed or in development, making this city the single most important internet gateway in the region. Every WhatsApp message sent from Nairobi, every M-Pesa transaction processed, and every Netflix stream watched in Kenya begins or ends its oceanic journey here. Submarine cables are the physical backbone of the internet. Despite the proliferation of satellite services like Starlink, more than 99% of intercontinental data traffic travels through these undersea fibre lines. For Kenya and its landlocked neighbours (Uganda, Rwanda, Burundi, South Sudan, and eastern DRC) Mombasa is the doorway to the digital world. ![Wide aerial view of the Port of Mombasa](/images/mombasa-port-wide.webp) ## The Cables That Land in Mombasa Kenya's first submarine cable, TEAMS (The East African Marine System), landed in 2009. Before TEAMS, Kenya's international internet traffic was routed through satellite links or through neighbouring countries' cables, resulting in high latency and prohibitive costs. TEAMS was a game-changer: it reduced international bandwidth costs by over 90% and catalysed Kenya's tech boom. The cables currently landing in Mombasa include: - **TEAMS** (2009), Owned by Telkom Kenya on behalf of the Kenyan government. Initial capacity 1.28 Tbps, subsequently upgraded. Connects Kenya to the United Arab Emirates and onwards to Europe and Asia. - **SEACOM** (2009), Privately owned, connects South Africa, Mozambique, Tanzania, and Kenya to Europe and India. Played a crucial role in breaking the bandwidth monopoly in East Africa. - **EASSy** (2010), The Eastern Africa Submarine System, a 21-member consortium cable running from South Africa to Sudan, with a branch landing in Mombasa. Connects to Europe via multiple landing points. - **LION2** (2012), The Lower Indian Ocean Network 2, an Orange-led cable linking Mombasa to Madagascar and Réunion. Ready for service since April 2012. - **DARE1** (2021), The Djibouti Africa Regional Express, in service since February 2021 and connecting Djibouti, Mogadishu and Mombasa. Designed to provide diversity and redundancy for landlocked East African nations. - **2Africa** (2024), One of the largest subsea cable projects globally, built by a consortium led by Meta. Its Kenya segment went live in June 2024, it lands at Mombasa and Mtwapa, and it delivers 180 Tbps of design capacity, dwarfing all previous cables. - **Africa-1** (landed 2024), A Telecom Egypt-led consortium cable connecting Africa to Asia via the Indian Ocean, with landings in Mombasa, Karachi and Egypt. The cable has come ashore at Mombasa but its ready-for-service date has not been announced, so DC254 does not count it among the live systems yet. - **PEACE** (2022), Pakistan and East Africa Connecting Europe, with a branch to Mombasa. Provides a new route to Europe via Pakistan and the Middle East. The cumulative design capacity of the live systems approaches 300 Tbps, an almost incomprehensible amount of bandwidth that positions Mombasa as one of the most connected coastal cities on the African continent. One more system is on the way. Daraja, a Meta-backed cable hosted by Safaricom, was announced in October 2025 to run from Salalah in Oman to Mombasa with 24 fibre pairs. It is in development and does not change the live count yet. Meta's subsea ambitions extend beyond African landings: its [Petal system](/articles/meta-petal-petabit-subsea-cable-kenya), announced in September 2026, will be the first petabit-class transoceanic cable, a signal of where the trunk capacity above Mombasa's routes is heading, even though Petal itself does not land in Africa. One development changes the geography without changing that count. At ITW Africa 2026 (September 2026), a consortium led by INDOI Ltd. (IOX) and Blue Trade Investments announced **LuLu**, a planned ~500km coastal cable system running from Mombasa to Lamu with landing points at Vipingo Special Economic Zone, Kilifi and Malindi in between. LuLu is not a tenth international system, it is Kenya's first dedicated coastal diversity route, pairing a subsea span (144 fibre pairs, up to 60 Tbps per pair) with a protected terrestrial path along the whole corridor. Its readiness-for-service target is Q2 2028, subject to contract execution; as of the announcement, no supplier was named and construction had not started. Full status check: [the LuLu coastal cable system](/articles/kenya-lulu-coastal-cable-system). ![Diagram of subsea cable architecture from landing station to POP](/images/submarine-cable-diagram.webp) ## From Ocean Floor to Nairobi: How the Connection Works A submarine cable does not plug directly into your phone. The journey from ocean floor to data centre involves several stages, each with its own infrastructure. When a submarine cable reaches shore, it terminates at a **cable landing station**, a purpose-built facility typically located in a secure, nondescript building near the coast. Inside, the cable's fibres are connected to terminal equipment that amplifies and splits the optical signals. From the landing station, the connection travels overland via **terrestrial fibre** (typically along railway lines, road reserves, or power line easements) to data centres in Nairobi, a distance of approximately 500 kilometres. In Nairobi, the cables connect to facilities like the [Kenya Internet Exchange Point (KIXP)](/articles/kixp-internet-exchange-point-kenya), where local traffic is handed off between ISPs, and to international carrier-neutral data centres like iXAfrica and Africa Data Centres, where global content providers and cloud platforms maintain their presence. The architecture means that traffic from a user in Nairobi to a Google server in Europe travels: user device → local ISP → terrestrial fibre → Mombasa landing station → submarine cable → European landing station → terrestrial fibre → Google data centre. The round trip takes roughly 100-150 milliseconds, which is fast enough for real-time applications like video calls and online gaming. ## Why Mombasa and Not Elsewhere? Mombasa's dominance as a cable landing point is not accidental. Several factors make it the natural choice: **Geography.** Mombasa is Kenya's principal deep-water port, with a natural harbour and proximity to the open Indian Ocean shipping lanes where cables are laid. The ocean floor topology off the Mombasa coast is also suitable for cable deployment. **Proximity to Nairobi.** The 500 km terrestrial fibre route from Mombasa to Nairobi is one of the shortest in the region. Nairobi, as East Africa's largest economy and the location of most data centres, is the natural destination for international bandwidth. **Existing infrastructure.** Once the first cable (TEAMS) established landing station infrastructure in Mombasa, subsequent cables could share or co-locate at the same facilities, reducing the cost and complexity of new landings. **Market size.** Kenya's large and growing economy, combined with its role as a gateway for landlocked neighbours, creates sufficient demand to justify multiple cable landings. No other Kenyan coastal city offers the same combination of market access and infrastructure. ![Fibre optic termination equipment in a data centre](/images/dc-fibre-optics.webp) ## What Happens When a Cable Is Cut? Submarine cables are vulnerable to damage from ship anchors, fishing trawlers, underwater landslides, and occasionally deliberate sabotage. Cable cuts are more common than most people realise, the Submarine Cable Almanac records dozens of outages globally each year. Kenya's multi-cable architecture provides natural redundancy. If TEAMS is cut, traffic can reroute through SEACOM, EASSy, or 2Africa. The [data centre licensing framework](/articles/kenya-data-centre-licensing-framework) also requires operators to maintain diverse connectivity paths, which further protects against single points of failure. The real risk is not a single cable cut but a simultaneous outage affecting multiple cables. This can happen if cables share a common geographic hazard, for example, if several cables pass through the same narrow strait and a single event damages all of them. The Indian Ocean has experienced such events, but Kenya's cable diversity (landing at geographically separated stations where possible) mitigates this risk. Repair times vary. Shallow-water repairs near shore can be completed in days. Deep-water repairs in the open ocean typically take two to six weeks, depending on the availability of repair ships and weather conditions. ![Submarine cable coming ashore at a Kenyan beach landing](/images/mombasa-cable-landing.webp) ## The Satellite Question: Is Mombasa Still Relevant? Low Earth orbit (LEO) satellite constellations like Starlink, OneWeb, and Amazon's Project Kuiper are often presented as alternatives to submarine cables. For some use cases (particularly in remote areas far from fibre infrastructure) they are transformative. But for Kenya's international connectivity, satellites complement rather than replace submarine cables. Submarine cables still offer orders of magnitude more bandwidth at lower cost per bit than any satellite system. A single modern cable like 2Africa (180 Tbps) carries more capacity than all LEO satellites combined. Cables also have lower and more consistent latency, which matters for financial trading, real-time collaboration, and cloud computing workloads. Where satellites excel is in providing connectivity to areas where terrestrial fibre is not available or economically viable. In Kenya, this means remote northern and north-eastern regions, and potentially as backup for critical government services. But the backbone of Kenya's digital economy remains the submarine cables landing in Mombasa. ![Crews hauling a submarine cable ashore at Mombasa](/images/mombasa-cable-landing-2.webp) ## What This Means for Data Centres The density of submarine cable connectivity in Mombasa is a significant competitive advantage for Kenya's data centre industry. International content providers and cloud platforms (Google, Microsoft, AWS, Netflix) need to be close to their users to deliver low-latency, high-quality experiences. Kenya's cable connectivity, combined with the data centre capacity in Nairobi, makes the country a natural hub for serving the East African market of over 300 million people. This is why companies like iXAfrica and Africa Data Centres are investing heavily in Nairobi, not just for the Kenyan market, but to serve as a regional hub for East and Central Africa. The submarine cables that land in Mombasa are the fundamental reason this strategy works. ## Tax Incentives for Data Centres in Kenya URL: https://data-centers-254.vercel.app/articles/tax-incentives-data-centre-investment-kenya Cluster: Policy & Regulation Published: 2026-08-28 Updated: 2026-09-16 Summary: The tax incentives for data centre development in Kenya: SEZ benefits, VAT exemptions, capital allowances, and how they compare regionally. ![Business meeting over data centre investment plans](/images/dc-business-investment.webp) The decision to build or expand a data centre in Kenya ultimately comes down to numbers. While market demand, connectivity, and power infrastructure are critical factors, the **tax and investment incentive landscape** often determines whether a project's financial model works or fails. For investors evaluating Kenya's data centre market, understanding the full spectrum of fiscal benefits (and how they compare with competing African jurisdictions) is essential. This guide provides a comprehensive breakdown of every significant tax incentive, investment benefit, and fiscal consideration for data centre development in Kenya, with practical comparisons that inform real investment decisions. ## Corporate Tax Rates: Standard and Incentivised ![Transmission tower beside a commercial high-rise](/images/dc-challenges.webp) Kenya's standard **corporate tax rate** for resident companies stands at **30%** under the Income Tax Act (Cap 470). This is the baseline rate that applies to data centre operators established outside any special incentive zone. However, the **Special Economic Zones Act 2015** provides a significantly reduced rate for qualifying operations. Data centres established and operating within a gazetted Special Economic Zone (most notably **Konza Technopolis**) benefit from: - **10% corporate tax rate** for the first 10 years of operation - **15% corporate tax rate** for the subsequent 10 years - After 20 years, the standard 30% rate applies This represents a substantial fiscal advantage. For a data centre generating KES 500 million in annual profit, the difference between a 30% and 10% tax rate translates to **KES 100 million per year in tax savings** during the first decade of operation. Over a 20-year investment horizon, the cumulative tax benefit under the SEZ framework is enormous. It is worth noting that **non-resident companies** operating data centres in Kenya through branches are subject to a **37.5% corporate tax rate**, making the SEZ incentive even more critical for international investors. ## The Special Economic Zones Act and Konza Technopolis Benefits The **Special Economic Zones Act 2015** is the centrepiece of Kenya's investment incentive framework for large-scale infrastructure projects. The Act establishes the legal basis for creating zones with preferential tax treatment, simplified regulations, and streamlined business processes. **Konza Technopolis**, gazetted as an SEZ under this Act, offers the most comprehensive incentive package available for data centre developers in Kenya. The benefits extend well beyond corporate tax reductions: - **VAT exemption** on goods and services supplied to and within the SEZ - **Stamp duty exemption** on execution of any instrument relating to the acquisition of land or property within the SEZ - **Import duty exemption** on raw materials, intermediate goods, and capital goods imported for use within the SEZ - **Streamlined licensing** through a single SEZ authority rather than multiple national and county agencies - **Work permit facilitation** for expatriate staff, a significant benefit for data centres that require specialised engineering talent The Konza Technopolis Development Authority (KoTDA) operates a **one-stop facilitation centre** that coordinates all regulatory approvals for SEZ licensees. For data centre developers, this means a single point of contact for construction permits, environmental licences, utility connections, and investment registration, dramatically reducing the time and cost of project establishment. ## VAT Exemptions on Imported Data Centre Equipment One of the most significant cost considerations for data centre development in Kenya is the **Value Added Tax (VAT)** on imported equipment. Data centres require substantial capital imports (servers, storage systems, networking equipment, UPS systems, precision cooling units, and electrical switchgear) much of which is not manufactured locally. Kenya's standard VAT rate is **16%**, which applies to most imported goods. However, several provisions in the **VAT Act (2013)** and the **EAC Customs Management Act** provide relief for qualifying ICT infrastructure imports: - **Zero-rating** applies to certain telecommunications infrastructure equipment imported by licensed operators - **Exemptions** exist for specific categories of ICT equipment under the **EAC Common External Tariff (CET)** schedules - Licensed telecommunications operators and **Network Facilities Providers (NFPs)** can apply for VAT relief on qualifying network infrastructure imports In practice, **servers and storage equipment** imported as part of an ICT infrastructure project may qualify for VAT exemption if the importing entity holds the appropriate Communications Authority of Kenya licence and the equipment is classified under the correct tariff headings. **UPS systems and cooling equipment** present a more nuanced situation, while they are essential to data centre operations, their classification as "ICT equipment" versus "industrial equipment" can affect their VAT treatment. The [Kenya Revenue Authority (KRA)](https://www.kra.go.ke/) has issued periodic guidelines on the interpretation of VAT exemptions for ICT infrastructure. However, the application of these exemptions remains an area where **professional tax advice is essential**, as misclassification can result in significant unexpected tax liabilities. ## Import Duty Considerations Under the EAC Common External Tariff Kenya is a member of the **East African Community (EAC)**, which operates a **Common External Tariff (CET)** applied to imports from outside the EAC bloc. The CET has four bands: - **0%** for raw materials and capital goods not available within the EAC - **10%** for intermediate goods - **25%** for finished goods - **35%** for sensitive goods (specific items listed by each partner state) Data centre equipment generally falls under the **capital goods** category, attracting a **0% import duty rate** under the CET. This is a critical advantage, it means that while VAT may still apply, the customs duty on servers, networking equipment, and other data centre hardware is typically zero when imported from outside the EAC. For equipment sourced from within the EAC (Tanzania, Uganda, Rwanda, Burundi, South Sudan, and the DRC), no import duty applies under EAC internal trade arrangements, though VAT may still be chargeable. Operators must ensure accurate **HS (Harmonised System) code classification** for each equipment category. The correct classification determines whether the 0% capital goods rate applies, and errors can be costly. ## Capital Allowances on Plant and Machinery Kenya's **Income Tax Act** provides capital allowances (effectively tax depreciation) that allow data centre operators to deduct the cost of capital assets from their taxable income over time. The most significant allowance for data centres is the **100% first-year investment deduction** on plant and machinery. While this allowance was traditionally associated with manufacturing, the expanded definition of qualifying assets under recent tax amendments and the interpretation by the [Kenya Investment Authority (KenInvest)](https://www.invest.go.ke/) has made it accessible to technology infrastructure operations. Key capital allowance rates relevant to data centres include: | Asset Category | Allowance Rate | Basis | |---|---|---| | **Plant and Machinery** (servers, UPS, cooling, electrical) | **100%** | First-year investment deduction | | **Computers and IT Equipment** | **100%** | First-year investment deduction | | **Industrial Buildings** | **2.5%** per annum | Straight-line basis over 40 years | | **Furniture and Fittings** | **12.5%** per annum | Reducing balance basis | | **Motor Vehicles** | **25%** per annum | Reducing balance basis | For a **KES 2 billion** data centre capital investment, of which approximately KES 1.2 billion is in plant and machinery (servers, power systems, cooling), the 100% first-year allowance provides a **tax shield of KES 360 million** in the first year alone at the standard 30% corporate rate. Within an SEZ at the 10% rate, the first-year tax shield is KES 120 million, though the lower rate itself provides far greater long-term savings. ## Withholding Tax on Dividends and Management Fees Data centre investors (particularly international operators and private equity-backed developers) must account for **withholding tax (WHT)** on cross-border payments: - **Dividends** paid to non-residents attract WHT at **20%** (which may be reduced under Double Taxation Agreements, Kenya has DTAs with the UK, Mauritius, and several other investment source countries) - **Management fees** paid to non-residents attract WHT at **20%** - **Technical service fees** paid to non-residents attract WHT at **20%** - **Royalties** paid to non-residents attract WHT at **20%** (reduced to 5% for certain DTA partner countries) For international data centre operators with regional headquarters in Mauritius or Dubai, **Double Taxation Agreements** can significantly reduce the effective WHT rate. The Kenya-Mauritius DTA, for example, reduces the dividend WHT rate for qualifying companies. Structuring the investment vehicle to benefit from available DTAs is a standard tax planning consideration for cross-border data centre investments. ## Stamp Duty on Land Acquisition Data centres are **land-intensive infrastructure**, requiring significant parcels for facility construction, compound security, transformer yards, and future expansion. In Kenya, the acquisition of land attracts **stamp duty**: - **4%** stamp duty on property transfers within municipalities (applicable to Nairobi, Mombasa, and other urban areas where data centres are typically located) - **2%** stamp duty on property transfers outside municipalities For a data centre requiring 5 acres of land in Nairobi at KES 300 million per acre, the stamp duty alone would amount to **KES 60 million**. This is a material upfront cost. However, **SEZ-located data centres** at Konza Technopolis are **exempt from stamp duty** on land acquisition within the zone. For large-scale data centre developments, this exemption alone can represent savings of tens of millions of shillings and is a significant incentive favouring SEZ-based development. ## The Kenya Investment Authority (KenInvest) and Investor Facilitation The **Kenya Investment Authority (KenInvest)** is the government's primary investment promotion and facilitation agency. For data centre investors, KenInvest provides several critical services: - **Investment certification**, which qualifies the investor for certain incentives and protections - **One-stop shop** facilitation for business registration, work permits, and regulatory approvals - **Investment guarantees** against nationalisation and expropriation under the Investment Promotion Act - **Repatriation guarantees** for capital and dividends earned by foreign investors - **Aftercare services** to support licensed investors post-establishment The investment certification process involves submitting a detailed business plan, proof of capital commitment, and compliance with environmental and regulatory requirements. Once certified, investors gain access to the full suite of facilitation services and are eligible for additional incentives that may be negotiated on a case-by-case basis for large-scale investments. KenInvest has been actively promoting **data centre investment** as a priority sector, recognising its strategic importance to Kenya's digital economy ambitions. The authority's engagement with international data centre operators has been a factor in several recent investment announcements. ## Proposed Digital Infrastructure Tax Incentives The Kenyan government has signalled its intention to introduce **additional tax incentives** specifically targeted at digital infrastructure, including data centres. While the specific provisions remain under discussion, proposals under consideration include: - **Enhanced capital allowances** for data centre equipment, potentially extending the 100% first-year deduction to cover a broader range of assets - **Accelerated depreciation** for data centre buildings and civil works, increasing the current 2.5% rate - **Tax credits** for investments in renewable energy used to power data centres, aligning with Kenya's climate commitments - **Reduced VAT** on data centre services to promote cloud adoption and digital transformation These proposals reflect a growing recognition within government that **targeted fiscal incentives** are necessary to attract the scale of data centre investment required to support Kenya's digital economy ambitions. For investors, staying engaged with the legislative process through industry associations and direct engagement with the National Treasury is essential to understanding and influencing these developments. External validation for this direction arrived in September 2026, when the Principal Secretary for ICT and the Digital Economy argued that Kenya's incentives must reward local computing services, AI capacity, additional renewable generation, skills and digital exports, not merely construction, and should include clawbacks where promised outcomes are missed. He pointed to Malaysia, which attracted over RM110 billion in approved data centre and cloud investment from 2021 by tying incentives to outcomes and sustainability rules. The full argument, with our fact-check against the market, is in [Kenya's data centre and compute hub plan, explained](/articles/kenya-data-centre-compute-hub-plan-explained). ## Comparative Analysis: Tax Incentives Across African Markets ![Power lines crossing Kenyan farmland](/images/kenya-transmission-pylons-4.webp) For investors evaluating data centre opportunities across Africa, the tax incentive landscape varies dramatically between jurisdictions. The following comparison highlights the key differences: ### Kenya - **Standard corporate tax**: 30% (37.5% for non-resident branches) - **SEZ incentive**: 10% for 10 years, 15% for 10 years (Konza Technopolis) - **VAT on equipment**: 16% standard, exemptions available for qualifying ICT infrastructure - **Import duty**: 0% on capital goods under EAC CET - **Capital allowance**: 100% first-year on plant and machinery - **Key strength**: Comprehensive SEZ framework with broad incentives; established DTA network ### Rwanda - **Corporate tax**: 30% standard, **up to 8-year tax holiday** for ICT companies - **VAT on equipment**: **Exemption** on imported ICT equipment - **Import duty**: 0% on ICT equipment - **Capital allowance**: 40-50% accelerated depreciation - **Key strength**: Most aggressive incentive package in East Africa; fastest registration and licensing (24-48 hours for business registration) ### Nigeria - **Corporate tax**: 30% standard - **SEZ/Free Zone incentive**: **Tax holiday of 3-5 years** in designated free zones (Lekki Free Trade Zone, etc.) - **VAT on equipment**: 7.5% standard, limited exemptions - **Import duty**: 5-20% depending on equipment classification - **Pioneer status incentive**: Up to 5-year tax holiday for qualifying industries - **Key strength**: Massive market size; multiple incentive pathways, though regulatory complexity is higher ### South Africa - **Corporate tax**: 27% standard - **Specific DC incentives**: None, relies on broad **Section 12I** manufacturing investment incentive and **Section 13 quat** renewable energy allowance - **VAT on equipment**: 15% standard, no specific DC exemption - **Import duty**: 0-20% depending on classification - **Capital allowance**: 40% in year 1, 20% in year 2, 20% in year 3 on qualifying assets - **Key strength**: Largest, most mature market; deep capital markets; but least generous specific DC incentives ### Ethiopia - **Corporate tax**: **No corporate tax for 5 years** in industrial parks (extended by 2-5 years for export-oriented investors) - **VAT on equipment**: Exemptions available in industrial parks - **Import duty**: 0% on capital goods in industrial parks - **Key strength**: Zero tax environment for initial years; rapidly growing market; but regulatory environment remains underdeveloped and foreign exchange challenges persist ## Practical Cost Comparison Table The following table illustrates the estimated tax burden on a **hypothetical KES 2 billion data centre investment** in each market over a 10-year period: | Cost/Tax Factor | Kenya (SEZ) | Kenya (Standard) | Rwanda | Nigeria (FTZ) | South Africa | |---|---|---|---|---|---| | **Corporate Tax Rate** | 10% | 30% | 0% (8 yr holiday) | 0% (5 yr holiday) | 27% | | **Estimated Annual Tax (KES M)** | 50 | 150 | 0 | 0 | 135 | | **10-Year Tax Cost (KES M)** | 500 | 1,500 | 0 | 0 | 1,350 | | **VAT on Equipment (KES M)** | 0 (exempt) | 96 | 0 (exempt) | 48 | 120 | | **Import Duty (KES M)** | 0 | 0 | 0 | 60 | 40 | | **Stamp Duty (KES M)** | 0 (exempt) | 60 | 10 | 40 | 30 | | **Total 10-Year Tax Burden (KES M)** | **500** | **1,656** | **10** | **148** | **1,540** | *Note: Figures are illustrative based on a KES 2 billion investment, 40% in plant/machinery (KES 800M), 60% in buildings/land (KES 1.2B), assuming KES 500M annual taxable profit. Actual figures depend on specific investment structure, equipment mix, and prevailing tax rates.* ## How Tax Policy Affects the Business Case for Data Centre Investment The cost comparison above demonstrates that **tax policy can be the difference between a viable and unviable investment**. For a KES 2 billion data centre, the 10-year tax burden in Kenya's SEZ framework (KES 500 million) is dramatically lower than the standard framework (KES 1,656 million), a saving of over **KES 1.1 billion**. Real investment decisions have been directly influenced by these incentives. **Africa Data Centres** selected Nairobi as a key expansion market partly because of the combination of SEZ benefits, capital allowances, and the zero-rated import duty regime that reduces upfront capital costs. Similarly, the investment case for [starting a data centre business in Kenya](/articles/starting-data-centre-business-kenya) is significantly strengthened when the full incentive framework is factored into financial models. However, investors should also consider the **non-tax factors** that affect the total cost of ownership: - **Power costs**: Kenya's electricity tariffs, while competitive regionally, remain a significant operating expense - **Regulatory compliance costs**: Licensing, environmental assessments, and ongoing regulatory compliance require resources - **Currency risk**: As with all African markets, KES-denominated revenues against USD-denominated capital expenditure creates foreign exchange exposure The [Kenya data centre market outlook 2025-2030](/articles/kenya-data-centre-market-outlook-2025-2030) projects that the combination of growing demand and available incentives will attract an additional 20-30 MW of data centre capacity over the period, with SEZ-based facilities capturing a significant share of new investment. ## Due Diligence: Verifying Incentive Availability Before committing to an investment based on expected tax benefits, investors must conduct thorough [data centre due diligence in Kenya](/articles/data-centre-due-diligence-kenya) to verify: - **Current incentive availability**: Tax incentives can be amended or withdrawn through legislative changes. The proposed digital infrastructure incentives demonstrate that the framework is evolving. - **Eligibility criteria**: SEZ benefits require gazettement and compliance with zone regulations. Capital allowance claims depend on correct asset classification. - **Compliance requirements**: Obtaining incentives often requires meeting specific investment thresholds, employment targets, or operational milestones. - **Transfer pricing considerations**: Related-party transactions (common in international data centre groups) must comply with Kenya's transfer pricing rules. Engaging qualified Kenyan tax advisors with specific experience in technology infrastructure investment is strongly recommended. The interaction between tax incentives, CA licensing requirements, SEZ regulations, and county-level permits creates a complex compliance landscape that requires specialist knowledge. ## Conclusion: Kenya's Competitive Fiscal Position for Data Centre Investment Kenya's tax incentive framework for data centres, while not the most aggressive in Africa, offers **substantial and meaningful benefits** when leveraged effectively. The **SEZ framework at Konza Technopolis** provides the deepest incentive pool, with 10% corporate tax, VAT exemptions, and stamp duty relief combining to create a compelling fiscal environment. For operators outside the SEZ, the 100% first-year capital allowance on plant and machinery and zero import duty on capital goods under the EAC CET still provide significant tax efficiency. Compared with Rwanda's more aggressive but smaller-market incentives, and South Africa's mature but less generous framework, Kenya occupies a **competitive middle position**, offering meaningful incentives with the added advantage of East Africa's largest economy, the region's most developed connectivity infrastructure, and a growing demand base driven by digital transformation across both the public and private sectors. For investors, the key is to **structure investments to maximise available incentives** (particularly SEZ location where feasible) while maintaining compliance with an evolving regulatory framework. With the proposed digital infrastructure incentives under discussion, the fiscal case for data centre investment in Kenya is likely to strengthen further in the coming years. ## Frequently Asked Questions **What is the corporate tax rate for data centres in Kenya?** The standard corporate tax rate in Kenya is 30% for resident companies. However, data centres established within a Special Economic Zone such as Konza Technopolis benefit from a reduced rate of 10% for the first 10 years of operation, and 15% for the subsequent 10 years, under the Special Economic Zones Act 2015. **Are data centre servers and equipment exempt from VAT in Kenya?** Yes, under the EAC Customs Management Act and Kenya's VAT Act, certain imported ICT equipment including servers, networking equipment, and some UPS systems qualify for VAT exemption or zero-rating when imported for use in licensed ICT infrastructure. The specific applicability depends on the equipment classification under the EAC Common External Tariff and the operator's licensing status. **What capital allowances are available for data centre plant and machinery?** The Kenyan Income Tax Act provides a 100% first-year capital allowance on plant and machinery used in manufacturing. While data centres are not strictly classified as manufacturing, many operators successfully claim investment deduction allowances of 100% on qualifying ICT equipment. Buildings and civil works attract different rates, typically 2.5% per year on industrial buildings and 4% on hotel buildings. **How does Konza Technopolis SEZ status benefit data centre investors?** Konza Technopolis is gazetted as a Special Economic Zone, offering reduced corporate tax rates (10% for 10 years, then 15% for 10 years), exemption from VAT on services and goods supplied within the zone, exemption from stamp duty on land transactions, and streamlined import procedures. These benefits can reduce the effective tax burden on a data centre operation by 40-60% compared to operating outside the SEZ. **How do Kenya's data centre tax incentives compare with Rwanda and Nigeria?** Rwanda offers up to 8 years of corporate tax holiday for ICT companies and VAT exemption on imported ICT equipment, currently the most aggressive incentive package in East Africa. Nigeria provides tax holidays of 3-5 years in special economic zones. Kenya's SEZ incentives at Konza are competitive but narrower in scope, while its broader fiscal framework (capital allowances, VAT exemptions on specific equipment) provides meaningful but less dramatic benefits. ## UPS Systems and Backup Power for Kenyan Data Centres URL: https://data-centers-254.vercel.app/articles/ups-backup-power-kenyan-data-centres Cluster: Energy & Power Published: 2026-08-28 Updated: 2026-10-11 Summary: A guide to UPS systems in Kenyan data centres: types, battery technologies, sizing, redundancy, generators, and Kenya-specific challenges. ## Why UPS Systems Are Critical in Kenyan Data Centres In a data centre, even a **millisecond** of power interruption can crash servers, corrupt databases, and disrupt services for thousands of users. This is why an Uninterruptible Power Supply (UPS) system is the most critical piece of electrical infrastructure in any facility, and in Kenya, where [Kenya Power infrastructure](/articles/kenya-power-infrastructure-data-centres) presents unique challenges, the UPS is not a luxury but an absolute necessity. ![Data centre power distribution and backup systems](/images/dc-power-systems.webp) A UPS system performs three essential functions. First, it **conditions the power**, smoothing out voltage fluctuations, filtering harmonics, and ensuring servers receive clean, stable electricity regardless of what comes from the grid. Second, it **bridges power gaps**, instantly supplying battery power during outages until backup generators start. Third, it **protects equipment**, shielding sensitive IT hardware from surges, spikes, and electrical noise that could cause premature failure. In Kenya specifically, UPS systems must contend with several challenges that make them even more critical than in markets with highly stable grids. ### Kenya-Specific Power Challenges **Kenya Power (KPLC) outages**: While KPLC has made meaningful improvements to grid reliability, unplanned outages remain common, particularly during the heavy rain seasons of March–May and October–December. Storms frequently bring down power lines and damage transformers, causing interruptions that can last from minutes to hours. **Voltage fluctuations**: Many parts of Nairobi and other towns experience voltage sags and swells, particularly during peak demand periods (typically 6–10 PM). Industrial areas where data centres are often located can see voltage drop by 15–20% during heavy load periods. **Frequency instability**: Kenya's grid frequency, which should ideally be stable at 50 Hz, can fluctuate during periods of generation-demand imbalance. While UPS systems with online double-conversion topology regenerate the power waveform, this constant conditioning adds to system wear and energy consumption. **Momentary interruptions**: Brief interruptions lasting less than one second (caused by automatic reclosers on power lines) are common and can be devastating to IT equipment if not properly bridged by a UPS. ## Types of UPS Systems Not all UPS systems are created equal. Understanding the three main topologies is essential for anyone evaluating data centre power infrastructure. ### Online Double-Conversion UPS The **online double-conversion UPS** is the gold standard for data centres. It operates by continuously converting incoming AC power to DC (rectification), then converting it back to clean AC power (inversion). The IT load is always powered by this double-converted, perfectly clean power, the UPS is always "online" in the power path. When a power failure occurs, the UPS seamlessly draws from its batteries without any transfer time, because the inverter is already running and simply changes its DC source from the rectifier to the batteries. This zero-transfer-time characteristic is why online double-conversion is the **only acceptable UPS topology for Tier III and Tier IV data centres**. Every major Kenyan data centre, including Africa Data Centres' facilities along Mombasa Road, PAIX Nairobi and Mombasa, and Liquid Intelligent Technologies' facilities, uses online double-conversion UPS systems from manufacturers such as **Eaton, Schneider Electric (APC), and Vertiv**. ### Line-Interactive UPS A **line-interactive UPS** monitors incoming power and can adjust voltage using an automatic voltage regulator (AVR) without switching to batteries. It only draws from batteries when the voltage goes outside the AVR's correction range. While line-interactive systems are common in small office environments and network closets, they are **not suitable for data centre environments** because they have a transfer time of 2–10 milliseconds when switching to battery mode, fast enough for computers but potentially problematic for sensitive server equipment and storage arrays. ### Standby / Offline UPS A **standby UPS** sits idle until a power failure is detected, at which point it switches to battery power. Transfer times range from 5–25 milliseconds, making them entirely unsuitable for data centre use. These are typically used for individual desktop computers and point-of-sale systems. ## Battery Technologies: VRLA vs Lithium-Ion The choice of battery technology significantly impacts a UPS system's performance, lifespan, maintenance requirements, and total cost of ownership. ### VRLA (Valve-Regulated Lead-Acid) Batteries VRLA batteries have been the traditional choice for data centre UPS systems for decades. They are sealed, maintenance-free (no water topping up required), and relatively affordable. **Advantages:** - Lower upfront cost, approximately KES 15,000 to 25,000 per 12V/100Ah unit - Well-understood technology with established supply chains in Kenya - Predictable performance and widespread manufacturer support **Disadvantages:** - Limited lifespan: 3 to 5 years in Nairobi's warm climate (vs. 7–10 years in cooler environments) - Large footprint: VRLA battery rooms can occupy 20–30% of a data centre's ground floor - Temperature sensitivity: every 8°C above 20°C halves battery life, a significant concern in Kenya where ambient temperatures regularly exceed 25°C - Environmental concerns: lead-acid batteries contain hazardous materials and require careful disposal - Weight: VRLA battery banks are extremely heavy, requiring reinforced flooring ### Lithium-Ion Batteries Lithium-ion technology is rapidly gaining ground in African data centres. Major manufacturers including **Eaton, Schneider Electric, and Vertiv** now offer lithium-ion battery cabinets purpose-built for UPS applications. **Advantages:** - Extended lifespan: 10 to 15 years, roughly double that of VRLA in tropical climates - Compact size: approximately 40% smaller and 60% lighter than equivalent VRLA banks - Higher operating temperature tolerance: performs well up to 40°C, reducing cooling costs - Faster charging: can recharge to 90% in under 2 hours - Better cycling capability: handles more charge-discharge cycles without degradation **Disadvantages:** - Higher upfront cost: approximately KES 35,000 to 50,000 per equivalent unit, 1.5 to 2 times more expensive - Thermal runaway risk: requires sophisticated battery management systems (BMS) - Limited local supply: most lithium-ion UPS batteries must be imported, adding lead time and shipping costs - Fewer local technicians with lithium-ion UPS maintenance expertise **Cost comparison in KES:** A typical 100 kVA UPS system with 10 minutes of VRLA battery runtime might cost KES 10–12 million for the battery bank alone. The equivalent lithium-ion system might cost KES 16–20 million, but over a 12-year operational life, the lithium-ion system may actually be cheaper when you factor in one battery replacement cycle for VRLA (KES 10–12 million at year 5) versus none for lithium-ion. ## UPS Sizing and Redundancy ### How UPS Systems Are Sized UPS sizing must account for the total IT load plus a safety margin. A properly sized UPS runs at **40–60% of its rated capacity** under normal conditions, leaving headroom for load growth and ensuring the system doesn't operate at maximum stress. For example, if a data centre has a current IT load of 200 kW, the UPS system might be sized at 400–500 kVA (accounting for power factor). This allows for future expansion without requiring a UPS upgrade. ### Redundancy Configurations ![Electrical switchgear and UPS cabinets](/images/dc-ups-batteries-wide.webp) **N+1 Redundancy**: Provides one additional UPS module beyond what is needed. If the facility needs 300 kW, an N+1 system might use four 100 kW modules, three carry the load, one is a hot standby. This is common in **Tier II and Tier III** facilities and provides protection against a single module failure. **2N Redundancy**: Provides two fully independent, parallel power paths, each capable of supporting 100% of the load. This is the standard for **Tier III and Tier IV** data centres. Africa Data Centres' Nairobi facilities use 2N UPS configurations, meaning every server rack receives power from two completely independent UPS systems. If the entire primary UPS system fails (controllers, batteries, and all) the secondary system takes over without any interruption to the IT load. **2(N+1) Redundancy**: An even higher level combining the principles of both approaches, two independent power paths, each with N+1 modules. This is reserved for the most critical facilities globally. ## Runtime Considerations: Bridging to Generator Start A common misconception is that UPS batteries should power a data centre for hours. In reality, **UPS runtime in a data centre is typically designed for 10 to 30 minutes**, just enough to bridge the gap between a utility power failure and the start of diesel generators. The sequence works as follows: 1. **Power failure detected** (0 milliseconds) 2. **UPS switches to battery power** (0–4 milliseconds for online double-conversion) 3. **Automatic Transfer Switch (ATS) signals generator start** (within 100 milliseconds) 4. **Diesel generator starts and stabilises** (10–30 seconds) 5. **ATS transfers load to generator** (after generator frequency and voltage are stable) 6. **UPS continues conditioning generator power** (ongoing) 7. **Generator runs until utility power is restored** (minutes to hours) Designing for extended battery-only runtime (beyond 15–30 minutes) would require impractically large and expensive battery banks. It is far more cost-effective to invest in reliable, fast-starting generators. ## Diesel Generators: The Backbone of Kenyan DC Power Continuity While the UPS handles the critical first seconds of a power failure, **diesel generators provide the long-term backup power** that keeps a data centre running through extended outages. ### Generator Sizing and Configuration Kenyan data centres typically size their generators to support the **entire facility load**, not just the IT load, but also cooling systems, lighting, and security. A 5 MW IT load might require 7–8 MW of generator capacity when cooling and other systems are included. Redundancy follows the same N+1 or 2N philosophy as UPS systems. A facility might have three 3 MW generators in an N+1 configuration, or four 3 MW generators in an N+2 configuration for additional resilience. ### Fuel Storage Requirements Most Kenyan data centres store **24 to 72 hours** of diesel fuel on site. Africa Data Centres' facilities typically maintain 48-hour fuel reserves, with contractual agreements for emergency fuel delivery within 12 hours from suppliers such as **KenolKobil, TotalEnergies, and Rubis Energy**. Fuel storage must comply with Kenya's environmental and fire safety regulations, including the **Environmental Management and Co-ordination Act (EMCA)** and the Kenya Bureau of Standards (KEBS) requirements for fuel tank installation. ### Alternative Generator Technologies While diesel remains dominant, alternative approaches are emerging: - **Biogas generators**: Companies like **Biogas International** and several pilot projects supported by the Kenya Renewable Energy Association (KEREA) are exploring biogas as a cleaner fuel source for backup power. - **Natural gas**: The discovery of natural gas reserves in Kenya and the development of gas infrastructure could eventually support gas-fired backup generators, which produce fewer emissions than diesel. - **AMEA Power and renewable hybrid systems**: AMEA Power's solar and wind investments in Kenya could eventually support hybrid power systems where renewables provide baseload power and batteries (rather than diesel generators) handle short-term backup. ## Automatic Transfer Switches (ATS) The **Automatic Transfer Switch (ATS)** is the electrical device that manages the transition between utility power and generator power. It continuously monitors both power sources and can transfer the load within seconds when it detects a failure. In a well-designed Kenyan data centre, the ATS operates in this sequence: 1. Monitors KPLC utility power voltage and frequency 2. Detects failure or unacceptable power quality 3. Sends start signal to diesel generator(s) 4. Waits for generator to stabilise (voltage and frequency within acceptable limits) 5. Transfers the load from utility to generator 6. Continues monitoring utility power 7. When utility power is restored and stable, transfers load back to utility 8. Allows generator to run for a cool-down period before shutting down A properly maintained ATS is critical because it is the **coordination point** between the UPS, the generator, and the utility grid. ATS failures (rare but possible) can result in the UPS batteries depleting before the generator takes over. ## Maintenance and Testing Schedules UPS and generator systems require rigorous, scheduled maintenance to ensure they perform when needed. The statistics show why this discipline matters. The [Uptime Institute](https://uptimeinstitute.com/resources/research-and-reports)'s Annual Outage Analysis 2026 found that power-related issues caused **more than one in five IT service outages** reported over the past three years. Within those power-related incidents, **UPS failures were the leading cause at 39%**, ahead of transfer switch failures (34%) and generator failures (28%), as reported by [Data Center Knowledge](https://www.datacenterknowledge.com/uptime/the-ups-is-uninterruptible-by-design-its-future-is-not-) citing the report. Uptime's research also flags a rising fire risk from lithium-ion UPS batteries. Most of these failures trace back to preventable maintenance and testing gaps, which is why the schedules below matter. ### UPS Maintenance - **Weekly**: Visual inspection of UPS status indicators, battery voltage readings, and alarm logs - **Monthly**: Battery impedance or conductance testing, thermal imaging of electrical connections, cleaning of air filters - **Quarterly**: Full load bank testing (where possible), calibration of monitoring sensors, review of event logs - **Annually**: Comprehensive preventive maintenance by the manufacturer or certified service provider, including firmware updates, capacitor inspection, and fan replacement - **Battery replacement**: VRLA batteries every 3–5 years, lithium-ion batteries every 10–15 years ### Generator Maintenance - **Weekly**: No-load exercise runs (10–15 minutes) - **Monthly**: Loaded exercise runs (at least 30% of rated capacity for 30 minutes) - **Quarterly**: Full load bank testing, fuel quality analysis, coolant system inspection - **Annually**: Comprehensive service including oil changes, filter replacement, alternator testing, and fuel system cleaning - **Fuel management**: Regular fuel polishing (removing water and contaminants from stored diesel), typically every 6–12 months ## Total Cost of Ownership Understanding the [data centre PUE (Power Usage Effectiveness)](/articles/data-centre-pue-power-usage-effectiveness) is essential because the UPS system directly impacts energy efficiency. Online double-conversion UPS systems typically operate at 93–96% efficiency, meaning 4–7% of the power is lost as heat during the double-conversion process. This inefficiency directly increases PUE and [energy costs](/articles/kenya-power-tariffs-data-centres). ### TCO Breakdown for a 500 kVA UPS System (Over 10 Years) | Cost Component | VRLA Battery System (KES) | Lithium-Ion System (KES) | |---------------|---------------------------|------------------------| | UPS units (500 kVA, N+1) | 45,000,000 | 45,000,000 | | Initial batteries | 25,000,000 | 40,000,000 | | Battery replacement (year 5) | 25,000,000 | 0 | | Installation and commissioning | 12,000,000 | 12,000,000 | | Maintenance (10 years) | 8,000,000 | 5,000,000 | | Cooling for battery room (10 years) | 6,000,000 | 2,500,000 | | **Total 10-year TCO** | **121,000,000** | **104,500,000** | As this simplified model shows, the **lithium-ion system is approximately 14% cheaper over 10 years** despite its higher upfront cost, primarily because it avoids a mid-life battery replacement and requires less cooling. ![Sustainable energy and battery storage systems at a modern data centre](/images/kenya-transmission-pylons-5.webp) ## Real Installations in Kenyan Data Centres **Africa Data Centres (Nairobi)**: Their facilities along Mombasa Road use Schneider Electric Galaxy VX online double-conversion UPS systems in 2N configurations, with both VRLA and increasingly lithium-ion battery options. Their UPS rooms are among the most advanced in East Africa, with real-time monitoring integrated into their building management system. ![Labelled diagram of a 42U server rack](/images/diagram-server-rack-42u.webp) **PAIX (Nairobi and Mombasa)**: PAIX facilities use Eaton 9395 UPS systems with N+1 redundancy. Their Mombasa facility, located near the submarine cable landing stations, has particularly robust power systems given its coastal location where power quality can be more variable. **Liquid Intelligent Technologies (Nairobi)**: Liquid's facilities deploy Vertiv Liebert EXL UPS systems with comprehensive battery monitoring. Their focus on energy efficiency has seen them pioneer advanced power management techniques in the Kenyan market. ## How the AI Era Is Reshaping UPS Design Everything above describes how Kenyan data centres build and run UPS systems today. Globally, however, the UPS is being reconsidered from both ends of the power chain, and buyers planning new capacity should understand where the technology is heading. The pressure comes from AI load profiles. Large GPU training clusters can swing their power demand by **tens of megawatts in seconds**, producing rapid load cycles that conventional UPS systems were never designed to smooth. Alex Cordovil, research director at Dell'Oro Group, told Data Center Knowledge that this has effectively flipped the UPS's job: "It's now less about protecting the IT load from the grid and more about protecting the grid from the IT load." His conclusion on the technology's survival is blunt but reassuring: "The UPS has been given a death sentence a few times, and every time it has come back in a new form." Three architecture shifts are driving that reinvention. First, hyperscale operators have adopted Open Compute Project direct-current designs with **distributed, rack-level battery backup units (BBUs)** alongside or in place of centralized UPS, and the move toward **800 V DC distribution** in AI factories extends this approach through sidecar power cabinets. Second, **battery energy storage systems (BESS)** are emerging as UPS alternatives at campus scale: in June 2025, BESS supplier FlexGen Power Systems and electrical contractor Rosendin announced a partnership to develop utility-scale BESS that can support data centres without traditional UPS systems. Third, **solid-state transformers (SSTs)** are on the horizon: Dell'Oro projects they will begin weighing meaningfully on UPS demand from 2029, initially in large AI factories that have largely moved away from UPS-based architectures. Supporting innovations include lithium iron phosphate (LFP), sodium-ion and nickel-zinc battery chemistries, plus supercapacitors that can absorb and release power in milliseconds. None of this means the centralized UPS disappears. Data Center Knowledge's October 2026 analysis concludes the opposite: with a large installed base serving non-AI workloads and inference workloads bringing availability requirements back to the fore, the UPS is evolving rather than vanishing, shifting from a pure resiliency device toward power-quality management and arbitrating between multiple energy sources. For Kenyan operators and buyers, two watch items follow. Kenya's grid already draws **over 90% of its electricity from renewable sources** (a generation mix figure, not a guarantee of site-specific power delivery), and as wind and solar capacity grows, grid power becomes comparatively clean but increasingly dynamic, the same pattern that is driving UPS redesign globally. And as Kenya positions itself for AI and GPU workloads through initiatives such as the country's [data centre compute hub plan](/articles/kenya-data-centre-compute-hub-plan-explained), buyers specifying new halls should ask vendors about LFP battery options, BESS-ready electrical rooms and how designs would accommodate higher-voltage DC distribution if AI-grade densities arrive. No Kenyan operator has announced an 800 V DC or BBU-based design; treat these as global trends to monitor, not local deployments. ## Frequently Asked Questions ### How long should a UPS system run during a power outage in a Kenyan data centre? A UPS system in a data centre is not designed for extended runtime. Its primary purpose is to **bridge the gap** between a power failure and the start of diesel generators, typically **10 to 30 seconds**. Some facilities design for 5 to 15 minutes of UPS runtime to provide additional buffer for generator start-up and synchronisation. Extended runtime on batteries alone would require impractically large battery banks and is not the standard approach. The generator, not the UPS, provides long-term backup power. ### What is the difference between N+1 and 2N UPS redundancy? N+1 redundancy means having one extra UPS module beyond what is required to support the full load. If you need 400 kW, an N+1 configuration might use four 100 kW modules (three to carry the load, one as backup). If any single module fails, the remaining three can still support the full load. 2N redundancy is far more robust, it provides a fully duplicate, independent power path. You have two complete UPS systems, each capable of supporting 100% of the load independently. If the entire primary UPS system fails, the secondary takes over seamlessly. **2N is the standard for Tier III and Tier IV data centres** and is what facilities like Africa Data Centres deploy in Kenya. ### Why are lithium-ion batteries gaining popularity in Kenyan data centres? Lithium-ion batteries offer several advantages over traditional VRLA lead-acid batteries: they last **10 to 15 years** compared to 5 to 7 years for lead-acid, they are roughly 40% smaller and lighter, they operate at higher temperatures (reducing cooling requirements), and they charge much faster. The main barrier has been cost, lithium-ion batteries are approximately 1.5 to 2 times more expensive upfront. However, the total cost of ownership over 10 to 15 years is increasingly competitive, especially when factoring in reduced replacement cycles, lower cooling costs, and smaller battery rooms. Newer facilities in Nairobi are increasingly specifying lithium-ion. ### How much does a UPS system cost for a Kenyan data centre? Costs vary enormously based on capacity and configuration. A 100 kVA online double-conversion UPS system from a tier-one manufacturer (Eaton, Schneider, Vertiv) costs approximately **KES 8 to 15 million** (USD 55,000 to 100,000) for the unit alone. Add battery cabinets, installation, switchgear, and commissioning, and a complete 100 kVA UPS room can cost KES 15 to 25 million. For a larger facility (say 1 MVA) with N+1 or 2N redundancy, the UPS infrastructure alone can range from KES 120 million to KES 250 million or more. These figures include import duties, shipping, and local installation costs. ### Will AI workloads change UPS design in Kenyan data centres? Globally, yes. Hyperscale operators are moving toward rack-level battery backup units, 800 V DC distribution and battery energy storage systems configured as UPS alternatives, and Dell'Oro projects solid-state transformers will begin weighing on UPS demand from 2029. Kenyan data centres continue to run conventional centralized UPS and generator designs, which remain the right baseline for today's colocation and enterprise loads. The practical step for Kenyan buyers is to ask vendors about lithium iron phosphate batteries, BESS-ready electrical rooms and higher-voltage DC readiness when specifying new halls. ### What role do diesel generators play alongside UPS systems? The UPS and generator work as a team. When utility power from Kenya Power fails, the UPS instantly takes over, it detects the interruption in under 4 milliseconds and begins supplying power from its batteries. Within 10 to 30 seconds, the automatic transfer switch (ATS) starts the diesel generator, lets it stabilise, and transfers the load from UPS batteries to generator power. The UPS continues to operate in 'online' mode, conditioning the generator's power to ensure clean, stable electricity reaches the servers. The generator provides power for as long as the outage lasts, limited only by fuel supply. Most Kenyan data centres store **24 to 72 hours** of diesel fuel on site. ## US Backs Africa's AI Data Centre Push: Where Kenya Fits URL: https://data-centers-254.vercel.app/articles/us-africa-ai-data-centre-frontier-kenya Cluster: AI & Infrastructure Published: 2026-09-14 Updated: 2026-09-14 Summary: The US is backing AI data centre deals across Africa as power limits bite at home. Lesotho, Gabon, Ivory Coast, and where Kenya fits. African data centres have moved from regional story to geopolitical one. In a Fox News Digital report published on 13 September 2026, analysts and US officials framed the continent as the next major frontier for AI infrastructure, precisely because the US and Europe are hitting limits of their own: power scarcity, grid congestion, permitting fights and environmental rules are slowing new data centre construction in mature markets. The US State Department put its position plainly, telling Fox News Digital that it is "committed to ensuring that American AI, including enabling infrastructure like data centers, continues to be the gold standard worldwide." The report matters for Kenya for a simple reason: the country is on the shortlist that accompanies the narrative. Arizton Advisory & Intelligence consultant Abhishek Goyal named South Africa, Kenya, Nigeria, Morocco and Egypt as the African markets attracting large-scale AI projects. But frontier stories attract pledges the way floodlights attract moths, so this article does what DC254 always does: separate what is signed from what is sited, and what is sited from what is switched on. ## What the Frontier Narrative Actually Says The argument from the US side has three parts. First, demand is real: governments, banks and large companies across Africa increasingly want local cloud and AI services, and they want sensitive data stored inside their own borders, which regulators and procurement rules are starting to require anyway. Second, supply is constrained elsewhere: as Goyal put it, "many mature markets face increasing environmental and power constraints," while Africa's gaps in infrastructure present "an opportunity rather than a disadvantage." Third, the US government is actively promoting American vendors: a State Department spokesperson described the goal of keeping American AI the "gold standard" and the US the "partner of choice" for governments upgrading telecoms networks and AI capabilities, and pointed to new undersea cable construction across the Atlantic as evidence that connectivity is improving faster than compute. ![Map of submarine cable systems touching African shores](/images/submarine-cables-map.webp) That third part is the genuinely new element. African data centre demand has been discussed for a decade as a commercial story. What the Fox report captures is its emergence as an arena of state-backed competition, with US embassies brokering deals that explicitly displace rival bids. The State Department framing ("partner of choice") and the competition references in the deal flow below are not our words; they are the words of the people doing the deals. ## The Deals Behind the Headlines Three US-linked packages carry the story in 2026, and each is at a different stage of verification. **Lesotho, $6.2 billion.** American company Convalt Energy won a package covering a data centre and a pumped-hydro power project (the Kobong scheme) with the government of Lesotho, in a deal the report says was facilitated with US government assistance. It is the largest single number in the narrative, and the structure is the interesting part: the data centre and the generation are bundled, because one cannot work without the other. **Gabon, $600 million in 2025 plus $100 million in 2026.** The US Embassy in Libreville helped American companies win $600 million of business in 2025, and this year assisted in securing a $100 million contract to supply two data centres and a gas-to-power plant, reportedly against strong competition from China. Two data centres plus dedicated generation in one package is the same pattern as Lesotho at a tenth of the headline size. **Ivory Coast, $170 million.** The US Embassy in Abidjan helped facilitate a $170 million contract for US company Cybastion with the government of Ivory Coast, which includes a data centre. The report carried images from a construction site visit, which puts this deal further along the verification path than most: renderings are promises, construction photos are evidence. The pattern across all three is worth naming. None of these is a hyperscaler buying land and waiting for a grid connection. Each bundle pairs compute with generation, because Africa's binding constraint is exactly what the analysts say it is: electricity. ## The Sober Counterpoint Arthur Goldstuck, CEO of World Wide Worx and author of "The Hitchhiker's Guide to AI: The African Edge," provided the reality check in the same report, and his quotes deserve as much attention as the deal values. "Africa is experiencing a data centre boom, but it begins from a very small base," he said, and "Africa will not replace either market as any kind of center of global computing." The attraction, in his framing, is serving a rapidly growing African cloud market and processing African data closer to the people and businesses that generate it. He also named the operational difficulty that headline numbers skip: weak electricity grids make construction and operation harder, and Africa remains far behind the US and Europe in installed capacity. His most useful observation for operators is the one about energy economics: "In some countries, a data center can also become the guaranteed customer that makes a new renewable energy project financially viable." That inverts the usual risk story. The data centre is not the customer that power projects fear; it is the anchor tenant that makes them bankable. ## Where Kenya Fits Kenya's name on Arizton's list (Goyal's five) is not a forecast; it is a recognition of what already exists, and our own directory makes it checkable. DC254 tracks 27 facilities in Kenya, of which 20 are operational, concentrated in Nairobi and Mombasa. The AI-specific positioning is already marketed: iXAfrica bills its Nairobi campus as Kenya's first hyperscale AI-ready facility, and the government's digital infrastructure push is active on the regulatory front, with the Communications Authority consulting on a standalone data centre licence. On the input side, Kenya's geothermal fleet gives operators something most frontier markets lack: cheap, firm, low-carbon baseload power, which is why energy-hungry compute keeps appearing in Kenya-focused investment conversations. The honest gaps are the ones Goldstuck names for the continent. Kenya's grid is more stable than most in the region but it is not unlimited, transmission into new industrial areas takes years, and the country's total installed data centre capacity is measured in dozens of megawatts, not the hundreds a single US campus can hold. That is why the Kenya angle in the frontier story is best read as a demand signal (US vendors and financiers are actively looking here) rather than a capacity claim. ## How to Read Frontier Numbers This site's standing advice applies with extra force to frontier narratives, because the numbers involved are pledges, and pledges compound enthusiasm. A $6.2 billion package is not $6.2 billion of installed capacity; it is a programme with a hydro project inside it, and pumped hydro takes the better part of a decade. An $80 million announcement is not an $80 million building until a groundbreak and a commissioning date exist. The discipline that works is the one this directory uses: track facilities, status and MW against dated sources, and re-verify when the story changes. For Kenya, that instrument already exists (our directory and its quarterly re-verification); for the rest of the frontier, the same habit is the difference between following the story and repeating it. Sources: Fox News Digital, "Could Africa be the next AI frontier as US and Europe hit limits?" by Paul Tilsley (13 September 2026), including quotes from Abhishek Goyal (Arizton Advisory & Intelligence), a US State Department spokesperson, and Arthur Goldstuck (World Wide Worx). Deal figures as reported therein. Kenya capacity and status figures from the DC254 directory (September 2026 verification). ## US DFC Bets Up to $155M on WIOCC: Kenya's Windfall URL: https://data-centers-254.vercel.app/articles/us-dfc-wiocc-155m-kenya-digital-infrastructure Cluster: Connectivity Published: 2026-09-17 Updated: 2026-09-18 Summary: The US DFC is putting up to $155m into WIOCC, its largest ever equity deal. What the money means for Kenya's bandwidth, open access and data centres. On 16 September 2026, Reuters reported that the US International Development Finance Corporation (DFC) will invest up to $155 million in WIOCC Group, the pan-African wholesale digital infrastructure firm. The agency called it the largest equity investment in its history, money it says will help build "the next generation of digital infrastructure for Africa's growth" (Reuters, 16 September 2026). WIOCC operates undersea cables, terrestrial fibre networks and, through its OADC division, carrier-neutral data centres. If you run a website, an office network or a server fleet in Kenya, investor headlines like this can feel distant. They are not. Wholesale capital decides how much international capacity lands on the Kenyan coast, how many carriers you can choose inside a Nairobi data centre, and how much a megabit of transit costs your provider next quarter. This article explains what the deal actually is, who WIOCC is, and what it changes for Kenyan buyers today. ## What the US DFC deal actually is The DFC is the United States government's development finance institution. It backs private companies operating in developing markets with equity, debt and political risk insurance, and it expects its capital back with a return. That commercial discipline matters for how you should read this news: this is not aid, and it is not a subsidy. An investment committee in Washington looked at African wholesale connectivity and concluded it is bankable. Three details in the announcement deserve attention. First, the structure: this is equity, up to $155 million, not a loan, which means the DFC shares both the risk and the upside of WIOCC's build-out. Second, the scale: Reuters describes it as the DFC's largest ever equity investment (16 September 2026), a benchmark that says as much about the asset class as about WIOCC itself. Third, the phrase "up to": commitments of this size are typically drawn in tranches as milestones are hit, so the money follows delivery, which is exactly how infrastructure finance should work. There is also precedent. The International Finance Corporation, the World Bank Group's private investment arm, made its own equity investment in WIOCC in September 2022 (IFC, 28 September 2022). Development finance institutions have therefore treated this company as investable infrastructure for four years. The DFC move extends that verdict and widens the pool of public capital behind African digital infrastructure. ## Who WIOCC is and why Kenyan networks already touch it WIOCC is a wholesalers' wholesaler. It does not sell home fibre or mobile bundles; it sells large blocks of capacity to the operators, internet service providers and big enterprises that then serve everyone else. Its asset base includes stakes in subsea cable systems, an open-access terrestrial fibre network, and a data centre division, with more than US$950 million invested across those three layers (Submarine Networks, 4 September 2026). ![A submarine cable landing station on the Kenyan coast](/images/mombasa-cable-landing-4.webp) The data centre arm matters most for this article. Open Access Data Centres (OADC) operates 13 core facilities and more than 30 edge sites, all carrier-neutral, across Nigeria, South Africa and the Democratic Republic of Congo, with direct connections to the Equiano and 2Africa subsea systems (OADC website, fetched 17 September 2026). Carrier-neutral is the key term: the facilities do not favour any single network, which is the same model that made Nairobi's best-connected buildings valuable. Kenya is already inside WIOCC's map, just not yet as a data centre market. WIOCC is one of the network providers present inside Kenyan carrier-neutral facilities such as Africa Data Centres' NBO1 on Mombasa Road, where tenants pick from multiple carriers rather than accepting whoever owns the building. When a Kenyan business buys connectivity in a facility like that, wholesale operators like WIOCC are often the ones carrying the traffic once it leaves the building. The relationship exists; the announcement scales it. ## Six weeks that reframed African digital infrastructure money The DFC announcement does not stand alone. Read together with the surrounding weeks, a pattern emerges about where institutional capital believes African digital infrastructure returns will come from. On 13 July 2026, Raxio Group secured up to US$380 million to fund data centre expansion across Africa (Data Centre Magazine, 13 July 2026). On 1 September 2026, WIOCC Group confirmed a $300 million investment from Africa Finance Corporation and Saudi Arabia's Vision Invest, announced at the LEAP conference in Riyadh, earmarked for three priorities: accelerating the deployment and consolidation of data centre capacity, extending open-access terrestrial fibre, and expanding subsea infrastructure (WIOCC Group, 3 September 2026). Two weeks later, the DFC committed its up-to-$155 million equity stake (Reuters, 16 September 2026). That is more than $800 million announced for African digital infrastructure in roughly nine weeks. ![SC connectors and yellow patch cords on a fibre patch panel](/images/fibre-patch-panel-sc-connectors.webp) The demand side explains the timing. The Africa Data Centres Association counts 360 megawatts of active capacity on the continent, with 238 megawatts under construction and 656 megawatts planned (ADCA economic report, 16 February 2026). Every megawatt under construction needs upstream capacity: subsea bandwidth to the coast, open-access fibre inland, and neutral halls where networks can meet. The investors announcing money since July are buying the picks and shovels of that build-out, not the consumer brands on top of it. Notice, too, what kind of infrastructure is attracting the capital. All three deals back multi-market, open-access, wholesale-first models rather than single-country consumer plays. Our earlier analysis of [why Africa's fibre boom needs a commercial operating model](/articles/africa-fibre-commercial-operating-model) argued the same point from the operator side: the constraint is no longer building the network, it is running it as a business many partners can buy from. Institutional investors appear to have reached the same conclusion. ## What the WIOCC investment means for Kenya Kenya sits downstream of all of this in a good position, and it is worth being precise about what changes and what does not. **More upstream capacity pressure, and that is a price story.** Seven live submarine cable systems make landfall near Mombasa (SEACOM, TEAMS, EASSy, LION2, DARE1, PEACE and 2Africa), with Africa-1 landed at Mombasa and Meta's Daraja system in development, the fullest cable portfolio in East Africa. Capital that expands subsea systems and open-access terrestrial fibre adds supply on exactly the routes Kenya depends on. Historically, every major capacity expansion on Kenyan routes has pushed wholesale transit prices down; more money upstream keeps that pressure on. **More carrier choice inside Kenyan buildings.** WIOCC already competes for wholesale traffic inside Nairobi's carrier-neutral facilities. A better capitalised WIOCC can bid more aggressively for that role, and its open-access fibre gives Kenyan ISPs and enterprises an alternative to incumbent backhaul routes. For a tenant, that shows up as more quotes to compare at cross-connect time, which is precisely what carrier-neutral buildings are supposed to deliver. **A credible candidate for Kenya's edge data centre map.** OADC's playbook is core-and-edge: big facilities in commercial hubs, small nodes deeper in the network. That playbook runs today in Nigeria, South Africa and the DRC (OADC website, fetched 17 September 2026). Kenya, with the region's densest facility cluster and its busiest cable landings, is an obvious future market. The honest caveat: no OADC facility in Kenya has been announced as of 17 September 2026, so treat this as a watch item, not a fact. **Validation that makes Kenya's own fundraise easier.** When the US government's development finance institution makes its largest ever equity bet on African digital infrastructure, every Kenyan project pitching for capital gains a reference point. The DFC's diligence becomes an argument Kenyan developers can cite: this asset class passes institutional scrutiny at the highest level. **Announced is still not open.** We made this point in our comparison of [Raxio and Africa Data Centres in Kenya](/articles/raxio-vs-africa-data-centres-kenya-comparison): a market is built out of commissioned facilities, not press releases. The same discipline applies to this deal. The number to watch is not $155 million; it is how much capacity actually goes live, where, and when. ![The Nairobi Expressway crossing above Mombasa Road](/images/nairobi-expressway.webp) ## What Kenyan buyers should watch next The practical moves are simple. First, track wholesale transit quotes over the next two to three quarters; capital announced in September shows up in price lists slowly, but the direction should be down. Second, when you evaluate facilities, ask which carriers and wholesale providers are physically present, and compare at least two quotes; our [Nairobi colocation buyer's guide](/articles/colocation-data-centres-nairobi-buyers-guide) lists the questions in order. Third, watch for any OADC announcement in Kenya specifically; if the edge playbook lands here, it will most likely appear near existing interconnection hubs rather than in greenfield suburbs. Fourth, treat every funding headline as a prompt to check what has actually opened, which is the habit this site is built on. None of this requires action today for most Kenyan businesses. What it changes is your negotiating position over the next year: more suppliers chasing your traffic, more capacity chasing your building, and more reasons for providers to compete on service rather than scarcity. Buyers who keep receipts and ask dated questions will capture most of that value. ## Frequently asked questions ### What is the US DFC? The US International Development Finance Corporation is the United States government's development finance institution. It invests in private companies working in developing markets, using equity, loans and political risk insurance. Unlike aid, DFC capital is expected to be repaid with a return, which is why an equity commitment of up to $155 million in WIOCC signals that the agency believes African digital infrastructure is a commercially sound investment (Reuters, 16 September 2026). ### Is the $155 million a grant or a loan? Neither. It is structured as an equity investment of up to $155 million, and Reuters describes it as the DFC's largest ever equity investment (16 September 2026). The phrase up to matters: commitments of this type are usually drawn in tranches as projects hit milestones, so the full amount lands only as the funded infrastructure is actually built. ### Does WIOCC own data centres in Kenya? Not as of 17 September 2026. WIOCC's data centre arm, Open Access Data Centres (OADC), lists 13 core and more than 30 edge facilities across Nigeria, South Africa and the Democratic Republic of Congo, with no announced Kenyan facility (OADC website, fetched 17 September 2026). WIOCC's carrier services do reach Kenyan facilities: it is among the network providers present inside carrier-neutral sites such as Africa Data Centres' NBO1 in Nairobi. ### What will the WIOCC investment fund? WIOCC's stated priorities for its current funding round are accelerating the deployment and consolidation of data centre capacity, extending open-access terrestrial fibre, and expanding subsea cable infrastructure (WIOCC Group, 3 September 2026). The DFC equity sits on top of the $300 million committed by Africa Finance Corporation and Vision Invest on 1 September 2026, so the group enters its next phase with up to $455 million of fresh institutional capital. ### Will this lower internet prices in Kenya? It pushes in that direction, but it is not a price cut. Wholesale capacity competition upstream of Kenya has historically compressed transit prices, and more capital for subsea and open-access fibre adds supply. Kenyan buyers see the effect indirectly: more carrier choices inside data centres, keener wholesale quotes and faster route diversity. The effect plays out over quarters and years, not on announcement day. ### How is the DFC deal different from the $300 million deal? The $300 million came from two commercial and institutional investors, Africa Finance Corporation and Saudi Arabia's Vision Invest, and was announced at LEAP 2026 in Riyadh on 1 September 2026 (WIOCC Group, 3 September 2026). The DFC money is a United States government agency making its largest ever equity investment, up to $155 million (Reuters, 16 September 2026). Same company, same build plan, different kinds of investors, which is itself the story: development finance is now treating African digital infrastructure as core allocation. ## VPS Hosting in Kenya: Providers That Take M-Pesa (2026) URL: https://data-centers-254.vercel.app/articles/vps-hosting-kenya-mpesa-payment Cluster: Infrastructure Published: 2026-09-16 Updated: 2026-09-16 Summary: A plain-English guide to VPS hosting in Kenya with M-Pesa payments: verified local providers, starting prices, and how to choose. Updated September 2026. Renting a server in Kenya used to have a strange barrier: not technical skill, not ideas, but the checkout page. Most global hosting companies want an international credit card, and plenty of talented Kenyan developers and small businesses simply do not have one. M-Pesa changed that. Today a handful of providers let you rent a full virtual server and pay for it the same way you buy airtime, straight from your phone. This guide explains what a VPS actually is, why paying with M-Pesa matters more than it sounds, and which providers we verified as accepting it, with their starting prices as of 16 September 2026. We wrote it for people who are new to servers, so every technical term is unpacked as we go. ![Network cables being patched into a server rack](/images/server-rack-patch-cabling.webp) ## What is a VPS, really? Let us build the idea from the ground up. A server is just a computer that stays on all the time and answers requests from the internet: when someone opens your website or your app, a server is the machine that sends them the pages. Servers live in data centres, which are secure buildings with backup power, cooling and fast internet connections. If you want the full picture, we have a plain-English walkthrough of [what happens inside a data centre](/articles/what-is-a-data-centre) and a separate guide to [what a server actually is](/articles/what-is-a-server). A VPS, short for virtual private server, is one slice of one of those computers. The provider takes a big physical server and splits it into several private pieces using software called a hypervisor (think of it as a very strict referee that gives each slice its fair share of memory and processing time). Each slice runs its own operating system, has its own disk space, and can be switched off and restarted by you alone. Three words in that name carry the meaning. Virtual, because it is software pretending to be a machine rather than a machine you can touch. Private, because your slice is sealed off from the slices next door. Server, because its whole job is to serve your website or application to visitors. In practice it feels like renting a small office in a shared building: the landlord handles the building, you handle what happens inside your room. ## Why M-Pesa changes the game On the surface, a payment method sounds like a small detail. In the Kenyan market it is often the deciding one, and there are three practical reasons why. First, the card problem. International clouds (Amazon Web Services, Google Cloud, Microsoft Azure and most big American hosts) bill with credit or debit cards, and many of them reject Kenyan-issued cards or place holds that a new developer cannot predict. If you cannot pass the billing check, the rest of the platform does not matter. M-Pesa removes that wall completely: the money moves from your mobile money balance, no card required. Second, currency clarity. Local providers bill in shillings, so there is no surprise when the exchange rate moves or your bank adds a foreign transaction fee on top of a dollar invoice. A plan that costs KSh 1,120 today is easy to budget for, and it is one less place where a small business can get quietly overcharged. Third, habit. Truehost, a Kenyan host, put a striking number in its February 2026 payment guide: over 90 percent of Kenyans use M-Pesa for daily payments. A product you can buy the way people already buy everything else is simply easier to adopt. For a student shipping a first project or a shop owner finally taking their business online, that friction matters more than any spec sheet. ## The providers, verified We checked provider websites directly on 16 September 2026 and listed only what we could verify on their own pages. Prices change often, so treat these as starting points and confirm on the checkout page before you pay. We have no commercial relationship with any provider on this list; this is reporting, not a recommendation to buy. **Hostnali.** A Kenyan provider selling VPS hosting from a Nairobi data centre, with Kenyan IP addresses included. Its site states that it is fully integrated with M-Pesa, along with other local payment methods such as Airtel Money (fetched 16 September 2026). That combination, local payments and a local IP, is exactly what the M-Pesa-first buyer is looking for. Kenyan IP matters more than people expect: some local services and search rankings behave differently when your server appears to be in Kenya. ![Close-up of a server processor chip](/images/dc-server-chip.webp) **Truehost.** One of the most visible budget hosts in Kenya. Its managed cloud server plans start at about KSh 1,120 per month (pricing page, fetched 16 September 2026), and its own payment guide from 17 February 2026 confirms M-Pesa acceptance. Managed means the provider handles routine server chores like updates for you, which suits first-time buyers who do not want to learn server administration from scratch. Older reviews put entry plans even lower, around KSh 860, so expect the exact figure to move with promotions. **MozOut.** Advertises Linux VPS plans from KSh 913 per month and lists M-Pesa, card and PayPal as accepted payment methods on its Kenya VPS page (fetched 16 September 2026). Linux VPS means the server runs the Linux operating system, which is the standard choice for websites and web apps because it is free, stable and well documented. **Others we checked, without the M-Pesa stamp.** Nescom advertises VPS plans from KSh 3,500 per month (fetched 16 September 2026), and Servercore sells cloud servers in Kenya with ready-to-use system images and public IP addresses. Neither page we reviewed confirmed M-Pesa in the sections we could verify, so if you consider them, ask support directly before signing up. Internationally, comparison sites like HostAdvice (February 2026) list global providers serving Kenyan users, but those almost always mean card payments and servers sitting outside the country. ## What your money gets you So what does roughly a thousand shillings a month buy? As a rule of thumb, entry VPS plans in this bracket offer something like one or two processor cores, a few gigabytes of memory, and tens of gigabytes of fast storage. A Kenyan cloud cost guide from July 2026 put entry cloud plans at USD 5 to 20 per month, roughly KES 700 to 3,000, which lines up with the local price points above. ![A network engineer connecting cables at a patch panel](/images/woman-network-engineer-patch-panel.webp) That is enough for a business website, a blog, a WhatsApp-adjacent booking system, a small API, or a portfolio that needs to survive a surge of visitors. It is not enough for heavy databases or AI work, and that is fine: the point of a first VPS is to learn cheaply and grow deliberately. Upgrading a VPS is usually a click and a reboot, so start small and scale when real traffic, not imagined traffic, demands it. One habit worth building from day one: backups. Ask the provider what backup options exist and what they cost, then actually turn them on. Storage is cheap, and the first time you delete the wrong file or a disk dies, the backup is the difference between a five-minute restore and losing the project. ## Local or global: how to choose The honest answer is that both have a place, and the choice gets easier once you know what you are optimizing for. Choose a local Kenyan VPS when you want M-Pesa billing, when your visitors are mostly in Kenya, or when you need a Kenyan IP. Server distance shows up directly in loading speed, and a visitor in Mombasa reaching a server in Nairobi travels a few hundred kilometres instead of several thousand. We explain this distance effect in detail in our guide to [Kenya internet speeds and what they mean](/articles/kenya-internet-speeds-data-centres). Choose a global cloud when you need services a small host does not offer, such as managed databases, advanced AI tools, or many regions around the world, and when your budget can absorb card billing in dollars. Plenty of Kenyan teams run a mix: the customer-facing website on a local VPS paid with M-Pesa, and heavier workloads on a global cloud later. If you ever outgrow renting entirely, the next step up is [what colocation means](/articles/what-is-colocation-kenya), which is putting your own hardware into a rented data centre slot. ![Students learning ICT skills in a Kenyan classroom](/images/classroom-ict-training-kenya.webp) ## A short buying checklist Before you click pay on any plan, run through five quick checks. They take ten minutes and prevent almost every common regret. 1. **Payment, for real.** Confirm M-Pesa appears on the actual checkout page, not just in a brochure. If it is missing there, ask support. 2. **Resources.** Write down the memory, processor and disk you are promised, then compare against what the server actually reports after setup. 3. **Location.** Ask where the machine physically sits. Nairobi is what most buyers want for speed inside Kenya. 4. **Support.** Message their support once before buying, at a normal hour. How fast and how human the reply is tells you what help will look like at 2 p.m. on a launch day. 5. **Backups and refunds.** Ask how backups work and whether there is a money-back window. A provider that answers both clearly is showing you how it will treat you later. The M-Pesa generation of Kenyan hosting is quietly one of the most practical infrastructure stories in the country: world-class tools, priced in shillings, paid from a phone. Start with a plan you can afford to forget about, learn on it, and upgrade when the traffic tells you to. ## Frequently asked questions ### What is a VPS in simple terms? A VPS (virtual private server) is one slice of a big computer in a data centre. The provider splits a physical server into several private pieces using software called a hypervisor. Each piece gets its own operating system, memory and disk space, and behaves like a small dedicated computer you control over the internet. You can install anything on it, restart it whenever you like, and nobody else can see your files. ### Can I pay for VPS hosting with M-Pesa in Kenya? Yes. Several Kenyan providers accept M-Pesa directly. Hostnali states on its site that it is fully integrated with M-Pesa as well as Airtel Money (fetched 16 September 2026). MozOut lists M-Pesa, card and PayPal as payment options. Truehost, a Kenyan host, wrote in a February 2026 payment guide that most local hosts accept M-Pesa. Always confirm the payment option on the provider's own checkout page before you commit, because offers change. ### How much does a VPS cost in Kenya per month? Entry-level local VPS plans start from roughly KSh 900 to KSh 1,200 per month. MozOut advertises Linux VPS plans from KSh 913 per month and Truehost's managed cloud servers start at about KSh 1,120 per month (both fetched 16 September 2026). A 2026 Kenyan cloud pricing guide put basic entry plans at about USD 5 to 20 per month, which is roughly KES 700 to 3,000. Bigger plans with more memory and storage cost more. ### Is a local Kenyan VPS better than AWS or DigitalOcean? They solve different problems. A local Kenyan VPS wins on three things: you can pay in shillings with M-Pesa, you do not need a credit card, and your server sits close to Kenyan users, which keeps loading times low. Global clouds like AWS or DigitalOcean offer more services and scale better, but they usually want an international card, and their nearest big regions sit outside Kenya. Many Kenyan teams start local, then move or mix later as they grow. ### What should I check before buying a VPS? Five things. One, the payment method, confirm M-Pesa works on the actual checkout page. Two, the resources, how much memory, CPU and disk you get for the price. Three, the location, a server in Nairobi serves Kenyan visitors faster than one in Europe. Four, support, check whether there is a Kenyan phone number or chat that answers during your working hours. Five, backups, ask what happens if the disk fails, because a cheap VPS without backups can cost you your whole project. ## What Is 5G? From 1G to 6G, and Why It Matters in Kenya URL: https://data-centers-254.vercel.app/articles/what-is-5g-1g-to-6g-history Cluster: Internet & Connectivity Published: 2026-09-13 Updated: 2026-09-13 Summary: 5G explained in plain language: what the G means, the full history from 1G to 5G, why each generation mattered in Kenya, and the 6G future. Ask most people what 5G is and the answer stops at "faster internet." That is true but incomplete, and the incomplete version hides both why 5G took decades to arrive and why Kenya's data centre industry watches every spectrum announcement so closely. 5G is the fifth generation of mobile network technology (the radio system that connects your phone to a mast instead of a cable) and it is the first generation designed to do three jobs at once: move data very fast, respond almost instantly, and connect enormous numbers of devices in the same place. This guide walks through what each of those words actually means, how we got here through four earlier generations, what 5G has already changed in Kenya since Safaricom switched on the region's first commercial network in October 2022, and what the promised sixth generation will realistically deliver. ## What the "G" Actually Means The "G" stands for generation, and the numbering is more than marketing, it tracks formal technical standards. Every generation corresponds to a set of specifications published by international standards bodies, principally 3GPP (the partnership of regional telecom standards organisations) and the ITU (the UN's International Telecommunication Union), which sets the formal requirements under names like IMT-2000 for 3G, IMT-Advanced for 4G, and IMT-2020 for 5G. When a network operator says a phone is "on 4G," it means the phone and the mast are speaking a protocol that descends from those published specifications. Two things follow from this that surprise people. First, a "generation" is not a single technology but a family of them, 4G includes LTE and LTE-Advanced, and 5G includes both the low-band coverage layer and the millimetre-wave speed layer. Second, generations overlap for years: Kenya's 2G networks still carry SMS and M-Pesa sessions today, alongside 4G and 5G on the same masts. Each generation adds a layer; very few get switched off. ## 1G: The Analogue Beginning (1980s–1990s) The first generation was analogue radio for voice calls, and it began when NTT launched a commercial cellular network in Tokyo in 1979, followed by the Nordic NMT system in 1981 and AMPS in the United States in 1983. 1G had three defining limitations: calls were not encrypted (anyone with a scanner could listen), the handsets were bricks costing the equivalent of thousands of dollars, and there was no data at all, a 1G phone could not send so much as a text message, because texts did not exist yet. Kenya entered the cellular era during the analogue decades, when the state monopoly Kenya Posts & Telecommunications Corporation operated the country's early mobile service in the 1990s. Ownership was a badge of status, connections were dropped the way landline calls were, and the idea that mobile networks would one day carry a country's payments system would have sounded like science fiction. What 1G proved (and this is the through-line for everything after) is that people will pay a premium to communicate without a wire. ## 2G: Digital Voice, SMS, and the Technology M-Pesa Was Built On (1990s–2000s) The second generation switched from analogue to digital, and the change did far more than sharpen call quality. GSM (the standard that came to define 2G globally, first deployed in Finland in 1991) encrypted calls, introduced the SIM card that made phones portable between networks, and carried two quiet features that would define the next thirty years: SMS text messaging and circuit-switched data at dial-up-like speeds, later boosted by GPRS and EDGE add-ons. Kenya's modern mobile story starts here. Safaricom launched its GSM network in October 2000, and mobile ownership began its long collapse from luxury to necessity. The generation's Kenyan masterpiece arrived in March 2007, when Safaricom launched M-Pesa, a money transfer system that ran on 2G's SMS and USSD channels, deliberately requiring neither a smartphone nor mobile internet, because in 2007 almost nobody had either. It is worth sitting with that: Kenya's most celebrated digital innovation, now studied worldwide, rode the second generation of mobile technology. Data centres had nothing to do with it, 2G's genius was being clever with almost no bandwidth. ## 3G: The Mobile Internet Arrives (2000s–2010s) The third generation, standardised under the ITU's IMT-2000 umbrella, finally made the internet a mobile experience. NTT DoCoMo launched the world's first commercial 3G network in Japan in 2001, and speeds that started under half a megabit climbed past 40 Mbps as HSPA upgrades rolled through the late 2000s. For the first time, a phone could load a webpage, stream a video, or download an app over a cellular network at tolerable speed, which is precisely when the smartphone became conceivable. Safaricom became the first Kenyan operator to launch a 3G platform in 2008, a fact the company's own launch materials for its 4G network would later list proudly in its lineage (mobile data first in 2003, 3G in 2008). The significance for Kenya was the app economy: 3G is the generation that made Android phones affordable and useful, moved Kenyans from feature phones to touchscreens through the early 2010s, and gradually turned M-Pesa from an SMS service into an app. It also created the first real demand for local hosting of content, because data (unlike voice and SMS) needed servers, and that traffic initially rode on submarine cables and overseas hosting. ## 4G: The Smartphone Broadband Era (2010s) The fourth generation, certified under the ITU's IMT-Advanced programme, turned mobile data from something you tolerated into something you relied on. The first commercial LTE networks went live in Stockholm and Oslo in December 2009, and LTE's all-IP architecture (carrying everything, including voice, as data) is the reason a modern call between two 4G phones is really a data session. Speeds in the tens and hundreds of megabytes made video streaming, video calls, cloud backups, and app-heavy smartphones a mass-market reality. Kenya's 4G moment came on 4 December 2014, when Safaricom launched the country's first 4G network, Reuters reported it as Kenya's first fourth-generation service, and Safaricom billed it as the most advanced network in Africa at the time. The 4G era is when Kenya's digital economy visibly accelerated: video consumption, ride-hailing, e-commerce, mobile-money apps with full interfaces, and the first serious conversations about local data hosting because 4G traffic volumes had outgrown the habit of backhauling everything abroad. Every generation so far had made the network faster; 4G made it load-bearing for the economy, and exposed how much of the workload was landing in data centres, few of which were in Kenya. ## 5G: Three Promises at Once (2019–Present) The fifth generation is different from every predecessor in one structural way: it was designed from the start for three distinct promises, not one. The industry shorthand is eMBB, URLLC, and mMTC. Enhanced mobile broadband (eMBB) is the speed story, hundreds of megabits to multi-gigabit connections. Ultra-reliable low-latency communication (URLLC) is the response-time story, networks designed for round trips of a few milliseconds, fast enough for factory robots and remote-controlled machinery. Massive machine-type communication (mMTC) is the density story, support for up to one million connected devices per square kilometre, for the sensor-heavy world the Internet of Things promises. 4G improved all three modestly; 5G was engineered to deliver all three simultaneously. The first commercial 5G networks went live in South Korea and the United States in April 2019. Kenya's path ran through Safaricom's first trials in March 2021 (the first operator in Kenya to announce them) through a commercial commitment made in late 2021, to the commercial launch on 27 October 2022, which made Kenya the first market in East Africa with live consumer 5G. Airtel Kenya followed on 6 July 2023 as the country's second operator, launching with 372 sites. Rollout since then has been steady rather than spectacular (concentrated in Nairobi, Mombasa, Kisumu and progressively more counties) which is the normal shape of a 5G build, because mid-band 5G needs denser masts than 4G ever did. The honest status report for Kenya: 5G is real, live, growing, and still a minority experience, most Kenyan phones and most Kenyan neighbourhoods are on 4G and will remain so for years. That is not a failure; it is how generations have always landed. 2G took roughly a decade to saturate; 4G is still not universal eighteen years after Stockholm. The question that matters is not "when will everyone have 5G" but "what does the network already make possible", which is where the importance comes in. ## Why 5G Matters: Three Levels of Importance **For the individual user**, the visible difference is speed and home internet. 5G fixed wireless access (a 5G router at home instead of a fibre line) is quietly one of the technology's most important Kenyan use cases, because it offers fibre-class speeds without waiting for a fibre trench to reach the gate. For phone users on a good 5G connection, large downloads, HD video calls, and cloud-heavy apps stop feeling constrained. Those gains are real but incremental: most things a person does online in 2026 work fine on 4G. **For the economy**, the important capabilities are the two 5G features most users never see. Low latency makes real-time control possible, factory automation, precision agriculture, remote diagnostics, and the logistics systems that depend on machines talking to machines without human-perceptible delay. Device density makes sensor networks viable at scale: soil probes, fleet trackers, smart meters, and security cameras in numbers that 4G towers were never dimensioned for. These applications, not faster video streaming, are where 5G's economic case actually lives, and Kenya's agriculture, health, and logistics sectors are the natural early markets. **For the data centre industry** (the reason this site watches 5G so closely) the effect is architectural. 5G's latency promises physically cannot be kept by sending every packet to a distant server: light itself needs about 5 milliseconds to travel 1,000 kilometres through fibre and back. Keeping response times in single milliseconds forces compute capacity out of a few large central facilities and into many small ones near the users, Multi-Access Edge Computing, or MEC. Each MEC node is, functionally, a small data centre with real power, cooling, and security requirements. 5G does not replace the big facilities either; every 5G mast still backhauls over fibre, and the traffic it generates ultimately pools in core data centres, internet exchanges, and the submarine cable landing stations around Mombasa. Our detailed analysis of [how 5G drives data centre demand in Kenya](/articles/5g-networks-data-centre-demand-kenya) puts numbers on that effect (hundreds of projected edge nodes nationwide) and it is the single strongest structural link between a telecom generation and Kenya's facility build-out. ![Labelled diagram of a fibre patch panel with connector types](/images/dc-networking.webp) ## Predicting 6G: What Is Known, What Is Speculation Here is the verifiable part of the 6G story. The ITU (the same body that formally defined every previous generation) approved the 6G framework, Recommendation ITU-R M.2160 under the IMT-2030 name, in November 2023. Standards bodies including 3GPP have begun the multi-year specification work, following the pattern every generation has set: roughly a decade passes between a framework's approval and mature commercial networks. Putting that timeline on the wall gives a realistic window of around 2030 for the first commercial 6G services, with early deployments thin, expensive, and concentrated in wealthy markets first, exactly as 5G did in 2019. What 6G might do is the speculative part, and it is worth separating the credible from the loud. The credible candidates, drawn from the IMT-2030 framework and the research agenda around it: networks that use artificial intelligence natively to route and heal themselves; "integrated sensing and communication," where the radio network doubles as a sensor layer detecting objects and motion; and the use of much higher frequency bands to push peak data rates further still. The loud versions (terabit-per-second to every phone, holographic calls, "the network will think for you") should be read as research directions, not roadmaps. Prediction is not commitment; the 6G that ships in 2030 will be narrower than the 6G in today's keynotes, just as 5G's launch was narrower than its 2015 hype cycle. For Kenya, the practical reading is calibration, not urgency. The country is still mid-way through 5G adoption, and 6G's arrival date does not change any investment decision a Kenyan business should make this year. What it does change is the ceiling: if integrated sensing and AI-native networks materialise even partially, the data-centre implications multiply, sensing generates continuous data streams that need nearby processing, and AI-native networks need training and inference capacity that lives in facilities, not masts. The generation after next, like this one, will be won in buildings as much as on spectrum. ## The Pattern Worth Remembering Set the five generations side by side and a Kenyan pattern emerges that is easy to miss from inside any single decade. 2G gave Kenya M-Pesa because digital channels reached people who had no banks. 3G gave Kenya the app economy because the internet finally fit in a pocket. 4G gave Kenya a digital economy heavy enough that hosting it abroad became the expensive habit. 5G is handing Kenya two things at once: fixed-wireless internet that competes with fibre for homes, and an edge-computing blueprint that pulls data centre jobs onshore for the first time. Each generation has mattered less for its headline speed than for the barrier it removed, and the barriers 5G removes are distance and delay, which is precisely what a data centre economy is built on. If the 6G predictions hold, the generation after next will push that logic further; the honest posture is interest with verification, which is the posture this site takes to every generation. Sources: Safaricom press releases (4G launch, 4 December 2014; 5G commercial launch, October 2022; trials first announced March 2021); Reuters (4 December 2014; 27 October 2022); Airtel Kenya (5G commercial launch, 6 July 2023); ITU Recommendation ITU-R M.2160 / IMT-2030 (November 2023). Timeline facts verified against multiple sources on 13 September 2026. ## What Is a Data Centre? A Plain-Language Guide URL: https://data-centers-254.vercel.app/articles/what-is-a-data-centre Cluster: Data Centres Published: 2026-08-25 Updated: 2026-08-27 Summary: A data centre is a specialised building housing the computers behind every app, website, and digital service you use. How they work, explained. A data centre is a specialised building (or purpose-designed space within a building) that houses thousands of networked computers called servers. These servers process, store, and transmit the data behind every app, website, streaming service, and digital payment system you use. When you send money on M-Pesa, search Google, or stream a YouTube video, a data centre is doing the work. Every single time. Think of a data centre as a factory for data. Instead of assembling physical products, it processes information at massive scale, 24 hours a day, 365 days a year. The computers inside never sleep. The power never stops. The cooling never turns off. Everything about a data centre is designed for one purpose: to keep servers running reliably and efficiently, no matter what. ## The Core Components of a Data Centre A data centre is far more than a room full of computers. It is an integrated system of interconnected infrastructure, each component depending on the others. Remove any one of these systems and the entire facility stops functioning. ![Labelled diagram of network rack components from router to PDU](/images/diagram-network-rack-components.webp) ### Servers and Storage Servers are the heart of any data centre. A single modern rack (the metal frame that holds the equipment) can contain 40 to 80 servers stacked vertically. Each server is essentially a powerful computer without a screen or keyboard, it exists to process requests and store data. Large facilities contain hundreds of racks, meaning tens of thousands of individual servers. Storage systems, often arrays of hard drives or solid-state drives, hold the actual data, databases, files, images, videos, application code. ![A data centre technician works beside UPS battery strings](/images/technician-ups-battery-room.webp) ### Power Systems Data centres consume enormous amounts of electricity. A single rack might draw 5 to 10 kilowatts. A facility with 200 racks could need 1 to 2 megawatts just for the servers, before adding cooling and other systems. The power infrastructure includes utility feeds from the national grid, [uninterruptible power supplies (UPS)](/energy) that provide instant backup during outages, diesel generators for extended outages, and power distribution units (PDUs) that deliver clean, stable electricity to every rack. ![Data centre power distribution and backup systems](/images/dc-power-systems.webp) ### Cooling Systems Servers generate heat, lots of it. A rack drawing 10 kilowatts produces as much heat as ten space heaters running continuously. If the temperature inside a server room rises above about 27 degrees Celsius, servers will throttle their performance or shut down entirely to prevent damage. Cooling systems, which can account for up to 40% of a facility's total energy consumption, use precision air conditioning (CRAC units), hot and cold aisle containment, and increasingly, liquid cooling to remove heat from the server environment. ![Data centre cooling and power distribution equipment](/images/dc-cooling-crac-2.webp) ### Networking Data centres need to connect to the outside world. This happens through fibre optic cables (typically multiple redundant links from different providers) that connect to [internet exchange points](/infrastructure) like KIXP in Nairobi, and through submarine cables that link Kenya to the rest of the world. Inside the facility, structured cabling connects every server and storage system to switches and routers that direct traffic between them and to the outside internet. ![Submarine cable coming ashore at a Kenyan beach landing](/images/mombasa-cable-landing.webp) ### Security Physical security is a critical differentiator. Enterprise data centres use multiple layers: perimeter fencing, 24/7 CCTV surveillance, biometric access control (fingerprint, iris, or facial recognition), mantraps (single-person entry portals), and on-site security personnel. The goal is to ensure that only authorised personnel can physically access the servers. This matters because the data stored inside (financial records, personal information, government databases) is valuable and sensitive. ![Security systems assessment during a facility audit](/images/dc-biometric-access-4.webp) ## Types of Data Centres Not all data centres are the same. The industry recognises several distinct types, each serving different purposes and operated by different kinds of organisations. **Enterprise data centres** are owned and operated by a single organisation for its own use. Banks, telecom companies, and large government agencies typically run their own enterprise facilities. Safaricom and Telkom Kenya, for example, operate data centres to support their mobile network and enterprise services. **Colocation data centres** (often shortened to "colo") are commercial facilities where multiple organisations rent space, from a single rack to a dedicated cage or private suite. The colocation provider supplies the power, cooling, networking, and physical security, while the customer provides and manages their own servers. [Africa Data Centres](/directory) and iXAfrica both operate significant colocation facilities in Nairobi. **Hyperscale data centres** are massive facilities, typically 10,000 square metres or more, built by or for the world's largest cloud and internet companies, Google, Microsoft, Amazon Web Services, Meta. These facilities are designed for efficiency at enormous scale, often in locations where power is cheap and climate conditions reduce cooling costs. Kenya's planned iXAfrica NBOX1.1 is East Africa's first facility designed to hyperscale standards. **Edge data centres** are smaller facilities located closer to end users, often in cities or neighbourhoods, designed to reduce latency for applications that require real-time response, things like autonomous vehicles, IoT systems, and 5G networks. While still emerging in Africa, edge computing is expected to grow as connectivity improves. ## Tier Ratings: Measuring Reliability The Uptime Institute's Tier classification system is the industry standard for rating data centre reliability. It defines four levels of infrastructure redundancy: - **Tier I (Basic)**: No redundancy. A single power or cooling failure can cause downtime. Suitable for small businesses with non-critical workloads. - **Tier II (Redundant Components)**: Some redundant components (N+1), but still a single distribution path. Slightly better availability than Tier I. - **Tier III (Concurrently Maintainable)**: Redundant power and cooling with multiple distribution paths. Any single component can be removed for maintenance without causing downtime. Guarantees 99.982% availability (maximum 1.6 hours downtime per year). This is the standard most commercial facilities in Nairobi target. - **Tier IV (Fault Tolerant)**: Fully redundant with 2(N+1) or greater redundancy. Designed to withstand any single failure without impact. Guarantees 99.995% availability (maximum 26 minutes downtime per year). Only the most critical facilities (military, financial trading, healthcare) justify the cost of Tier IV. ## Why Data Centres Matter for Kenya Kenya's economy is increasingly digital. M-Pesa processes transactions worth billions of shillings daily. Government services are moving online through e-Citizen and Huduma Centres. Banks, fintech companies, and mobile network operators all depend on data centre infrastructure to serve their customers. Without local data centres, every digital transaction in Kenya would have to be processed in a facility in Europe, South Africa, or the Middle East, adding hundreds of milliseconds of latency and significant cost. Kenya also has a structural advantage. The [submarine cables landing in Mombasa](/infrastructure) give the country direct connectivity to Europe, Asia, and the rest of Africa. This makes Nairobi an attractive location for data centres serving not just Kenya, but the broader East and Central African market of over 300 million people. The government's National Digital Superhighway Programme, the planned Microsoft-G42 AI data centre (currently stalled over power supply constraints), and the expansion of existing facilities by Africa Data Centres and iXAfrica all point to rapid growth. Understanding [Kenya's data centre licensing framework](/articles/kenya-data-centre-licensing-framework) is essential for anyone considering operating or investing in this space. For those interested in working in this industry, [data centre careers in Kenya](/careers) span roles from facility engineers and network architects to cooling specialists and compliance managers, positions that most Kenyans have never heard of but that are critical to the country's digital future. ## What Is a Server? The Machines Powering the Internet URL: https://data-centers-254.vercel.app/articles/what-is-a-server Cluster: Beginner Guide Published: 2026-08-28 Updated: 2026-08-28 Summary: A server is a specialised computer that serves other computers over a network. The types, how they differ from PCs, and why they live in data centres. Every time you open a website, send an email, make a mobile money transfer, or stream a video, a server somewhere is doing the work. Servers are the invisible machines behind every digital service you have ever used. They are the reason Google can find your search results in milliseconds, the reason M-Pesa can process your payment instantly, and the reason Netflix can start playing a movie within seconds of you clicking play. Yet most people who depend on servers every day have never seen one and could not describe what one actually is. ![GPU servers racked in a data centre](/images/ai-gpu-servers.webp) A server is, at its most fundamental level, a computer that provides services to other computers. The word "server" refers not to a specific type of hardware but to a role: a computer that serves. Any computer can be a server if it is running software that provides services to other computers over a network. Your laptop could be a server if you ran web server software on it and other computers connected to it to view web pages. In practice, however, when people say "server" they mean a purpose-built computer designed specifically for the demands of providing services reliably, continuously, and at scale. ## What Makes a Server Different from a Regular Computer Servers and regular computers (desktops, laptops) share the same fundamental architecture. Both have processors (CPUs), memory (RAM), storage (hard drives or SSDs), network interfaces, and power supplies. Both run operating systems. Both execute software. The differences are in the design priorities: servers are optimised for reliability, performance, and manageability, while regular computers are optimised for individual usability and cost. ### Reliability: Built to Run Forever A regular laptop is designed to be used for 3-5 years, turned on and off daily, and carried around in a bag. A server is designed to run continuously for 5-10 years or more without being turned off. This difference in design intent manifests in every component. Servers use **ECC memory** (Error-Correcting Code RAM), which can detect and correct single-bit memory errors that would cause a regular computer to crash. In a server running 24/7 processing thousands of transactions per second, a memory error could corrupt a database or drop an ongoing transaction. ECC memory prevents this, adding a layer of reliability that standard desktop memory does not provide. Servers have **redundant power supplies**, two or more power supply units (PSUs) in each server, each capable of powering the entire server independently. If one PSU fails, the other takes over seamlessly. This is why servers connected to data centre power systems (which already have redundant power paths) can achieve near-perfect uptime, there are multiple layers of redundancy between the grid and the processor. ![Rows of enterprise server racks in a data centre hall](/images/worldbank-dc-server-row.webp) Servers use **hot-swappable components**, drives, fans, and even power supplies can be replaced while the server is running, without shutting it down. In a data centre environment, shutting down a server to replace a failed drive would disrupt services. Hot-swap capability means a technician can slide out a failed drive and slide in a replacement while the server continues operating. ### Form Factor: Designed for Density Servers are designed to be mounted in standard 19-inch equipment racks. The most common form factor is the **rack server** (also called a 1U or 2U server), which is a flat, rectangular box that slides into a rack like a book into a bookshelf. The "U" stands for "rack unit", 1U is 44.45 millimetres (1.75 inches) tall. A standard 42U rack can hold 42 1U servers, though in practice, airflow and power requirements mean most racks hold 10-20 servers with space between them. The density of rack servers is remarkable. A single 42U rack, occupying less than half a square metre of floor space, can contain 20-40 servers, each one a powerful computer in its own right. A data centre hall with 200 racks can house 4,000-8,000 individual servers, processing millions of transactions per second for hundreds of different customers. Other server form factors include **blade servers** (compact server modules that slot into a chassis that provides shared power, cooling, and networking) and **tower servers** (standalone units that look like large desktop computers, typically used in small offices rather than data centres). ### Performance: Optimised for Workloads Server processors are designed differently from desktop processors. While a desktop CPU is optimised for single-threaded performance (making one task run as fast as possible), server CPUs are optimised for multi-threaded performance (handling many tasks simultaneously). A modern server processor may have 32, 64, or even 128 cores, allowing it to run hundreds of virtual machines or handle thousands of concurrent requests. Server memory is designed for capacity and reliability rather than speed. A typical server supports 256GB to 4 terabytes of RAM (compared to 8-32GB in a typical desktop), using ECC memory modules that can detect and correct errors. Server storage uses enterprise-grade SSDs and hard drives designed for 24/7 operation and higher write endurance than consumer drives. ## Types of Servers ### Web Servers A web server receives HTTP requests from browsers (like the one you are using now) and responds with web pages, images, and other content. When you visit datacentre254.com, your browser sends a request to a web server, which finds the page and sends it back. Popular web server software includes Nginx, Apache, and Microsoft IIS. A busy web server can handle thousands of requests per second. ### Database Servers ![Technician working with network cabling in a data hall](/images/fibre-splicing-technician-data-hall.webp) Database servers store, organise, and retrieve structured data. Every time you check your bank balance, search for a product on Jumia, or look up a contact on your phone, a database server is querying a database to find and return the relevant information. Database servers are typically the most resource-hungry type, requiring fast processors, large amounts of memory (to cache frequently accessed data), and fast storage (to read and write data quickly). Major database software includes PostgreSQL, MySQL, Microsoft SQL Server, and Oracle Database. ### Application Servers Application servers run the business logic of applications, the code that processes requests, applies business rules, and coordinates between web servers and databases. When you submit a loan application on a bank's website, the web server receives the form, the application server processes it (checking credit scores, calculating terms, applying rules), and the database server stores the result. ### Virtualisation Hosts In modern data centres, physical servers are rarely dedicated to a single application. Instead, **virtualisation software** (like VMware ESXi, Proxmox, or KVM) runs on the physical server and creates multiple **virtual machines** (VMs), each behaving like an independent server with its own operating system and applications. A single powerful physical server might run 10-50 virtual machines, each serving a different customer or application. This dramatically improves hardware utilisation and flexibility. ### File and Storage Servers File servers provide shared storage accessible over a network. They allow multiple users and applications to store and retrieve files from a central location. In enterprise environments, file servers have been largely replaced by **storage area networks** (SANs) and **network-attached storage** (NAS) systems, which provide more sophisticated storage management, redundancy, and performance. ## What Is Inside a Modern Server Open a modern 1U rack server and you will find a remarkably tightly packed arrangement of components. At the front, a row of hot-swappable drive bays (typically 2-8 drives) allows storage to be replaced without opening the case. Behind the drives, the motherboard holds the CPU(s), RAM slots, and expansion card slots. Cooling fans (typically 4-8 small, high-speed fans) pull air through the server from front to back, exhausting hot air out the rear. At the rear, you will find network ports (typically 2-4 Ethernet ports), management ports (for remote administration), and the power supply connections. A typical modern rack server for enterprise workloads might have two processors (each with 32-64 cores), 256GB-1TB of RAM, 4-8 SSDs or hard drives, and 2-4 network connections. It consumes 300-800 watts of power, generates significant heat, and costs between KES 400,000 and KES 2,000,000 depending on the specification. ## Why Servers Live in Data Centres [As explained in our data centre guide](/articles/what-is-a-data-centre), servers need a specific environment to operate reliably: stable power, precise cooling, physical security, fire protection, and high-speed network connectivity. A server sitting under a desk in an office might work for a small business, but it will not provide the reliability, security, or performance that a data centre environment delivers. ![Labelled diagram of a 42U server rack](/images/diagram-server-rack-42u.webp) Data centres exist because the collective needs of thousands of servers (for power, cooling, connectivity, and security) are best met by purpose-built facilities with dedicated infrastructure and expert staff. Every server you interact with online lives in a data centre, and understanding what servers are and how they work is the first step to understanding the entire digital infrastructure ecosystem that powers Kenya's economy. ## What Is Colocation? Renting in a Kenyan Data Centre URL: https://data-centers-254.vercel.app/articles/what-is-colocation-kenya Cluster: Data Centres Published: 2026-08-28 Updated: 2026-08-28 Summary: Colocation lets you rent space, power, cooling, and connectivity in a professional data centre. How it works in Kenya and what it costs. Imagine you run a growing technology company in Nairobi. You have ten servers humming in a converted storeroom in your office, connected to a single Safaricom fibre line, protected by a consumer-grade UPS that would last maybe 15 minutes during a power cut, and cooled by the building's air conditioning system that was designed for humans, not for machines that generate heat 24 hours a day. Every time there is a power blip, your IT team holds their breath. Every time the building management switches off the AC at night to save power, your server room temperature creeps up. Every time your single internet connection drops, your entire business goes offline. ![Blue-lit server racks in a data hall](/images/hero-server-hall.webp) This is the problem that colocation solves. Instead of housing your servers in an inadequate, improvised space, you rent space in a purpose-built data centre that provides professional-grade power, cooling, connectivity, and security. Your servers sit in a facility designed specifically for them, with redundant power supplies, precision cooling, multiple fibre connections from different providers, biometric access control, 24/7 security cameras, and fire suppression systems. You own the servers and the data on them. You manage the software and applications. But the physical infrastructure (the building, the power, the cooling, the connectivity) is provided by the colocation operator. Colocation is one of the oldest and most fundamental services in the data centre industry. It is the service that launched companies like Equinix, Digital Realty, and Teraco into multi-billion-dollar global businesses. And it is a service that is growing rapidly in Kenya as more organisations recognise the limitations of running their own server rooms. ## How Colocation Works Colocation is, at its core, a real estate and infrastructure service. The colocation provider builds and operates a data centre facility, and then rents space within that facility to multiple customers. The key principle is that multiple unrelated customers share the same physical facility, but their equipment, data, and network traffic are completely separated. ### Space Units Colocation space is typically sold in three units. A **rack** (also called a cabinet) is the standard unit, a metal frame approximately 600mm wide, 1,070mm deep, and 2 metres tall, with 42 vertical slots (called "U" for rack units) for mounting equipment. A standard 1U server is about 44mm tall, so a full rack can hold 42 1U servers (in practice, fewer because servers need airflow space). A **half-rack** provides 21U of space, and a **quarter-rack** provides approximately 10U. For larger deployments, customers can rent a **cage**, a wire mesh enclosure within the data centre that provides additional physical security and privacy. Cages are typically used by customers who need 10 or more racks. For the largest customers, some facilities offer **private suites**, essentially a room within the data centre that only that customer can access. ### Power Colocation pricing is closely tied to power allocation. Each rack comes with a specific power allocation, typically measured in kilowatts (kW). A standard rack allocation in Kenya is 4–6kW, which is sufficient for 10–20 modern servers. High-density racks, needed for GPU servers or dense compute platforms, may require 10–20kW per rack, which costs more because of the additional cooling and power distribution infrastructure required. ![Shore-based cable works on the Kenyan coast with a vessel offshore](/images/mombasa-cable-landing.webp) Power is delivered to the rack through power distribution units (PDUs) that are part of the facility's infrastructure. Most colocation racks are equipped with dual PDUs, each connected to a separate power path (A and B feeds), so that a failure in one power path does not affect the other. Servers with dual power supplies can connect to both PDUs, providing power redundancy at the server level. ### Cooling Colocation facilities maintain strict environmental conditions: temperature typically between 18–27°C and humidity between 20–80% relative humidity, in accordance with ASHRAE guidelines. Precision air conditioning systems, computer room air conditioning (CRAC) units or computer room air handling (CRAH) units, maintain these conditions continuously. In Nairobi's moderate climate, some facilities use free cooling (drawing in outside air when conditions permit) to reduce energy consumption, as explained in [our cooling systems guide](/articles/data-centre-cooling-systems-explained). ### Connectivity One of the most important features of a carrier-neutral colocation facility is connectivity choice. In a carrier-neutral facility, multiple network providers (Safaricom, Liquid Intelligent Technologies, Telkom Kenya, Jamii Telecommunications, and others) have equipment in the facility's meet-me room. Customers can connect to any or all of these providers, choosing based on price, performance, redundancy, or specific service requirements. This is fundamentally different from a non-neutral facility (like a telecom-owned data centre) where you can only connect to that operator's network. ## Colocation in Kenya: The Market Kenya's colocation market is served by several providers, each with different strengths and market positions. ### iXAfrica iXAfrica operates the NBOX1 and NBOX1.1 facilities along Mombasa Road in Nairobi. NBOX1 opened in 2023 as a purpose-built, carrier-neutral colocation facility designed to Tier III standards. It offers retail colocation (individual racks and half-racks) and wholesale colocation (dedicated rooms or cages for large customers). iXAfrica's facility is notable for its focus on sustainability and is designed to achieve low PUE values, leveraging Nairobi's climate for free cooling. ![KICC tower and the Nairobi skyline at night](/images/nairobi-skyline-night-kicc.webp) ### Africa Data Centres Africa Data Centres, part of the Cassava Technologies group, operates multiple facilities in Nairobi including the Sameer Business Park and Westlands sites. As Africa's largest colocation provider by footprint, ADC brings scale and standardised processes to the Kenyan market. Their integration with Liquid Intelligent Technologies' fibre network provides connectivity advantages, and their pan-African presence allows customers to deploy in multiple African countries through a single provider relationship. ### Safaricom Safaricom's data centres primarily serve the company's own needs, mobile network infrastructure, M-Pesa processing, and enterprise cloud services. However, Safaricom also offers colocation to enterprise customers, particularly those who want an integrated solution combining colocation with connectivity (Safaricom fibre and mobile) and cloud services. Safaricom's facilities are not carrier-neutral (customers primarily connect to Safaricom's network) which is a limitation for customers who want multi-provider connectivity. ## Pricing in the Kenyan Market Colocation pricing in Kenya varies by provider, space unit, power allocation, and contract terms. Here are indicative ranges based on current market rates. **Full rack (42U, 4-6kW)**: KES 60,000–120,000 per month. This typically includes the rack space, power allocation, basic cooling, 24/7 access, and shared internet connectivity (a base bandwidth allocation). Additional charges apply for extra power, cross-connects to specific providers, remote hands services, and premium support. **Half rack (21U, 2-3kW)**: KES 35,000–70,000 per month. Per-unit pricing is higher than a full rack (you pay a small premium for the smaller commitment), but the total cost is lower, making it accessible to smaller organisations. **Quarter rack (10U, 1-2kW)**: KES 20,000–45,000 per month. Best suited for small businesses or organisations with just a few servers that need professional infrastructure. **Dedicated cage (10+ racks)**: KES 50,000–90,000 per rack per month. Volume discounts apply for larger deployments, and pricing is typically negotiated individually. These prices are competitive by African standards, roughly 20–30% lower than equivalent colocation in Lagos and comparable to mid-tier facilities in Johannesburg. The main additional costs to budget for are cross-connect fees (KES 5,000–15,000 one-time per connection), remote hands charges (KES 2,000–5,000 per incident), and excess power charges if you exceed your allocated power. ![Security cameras monitoring a colocation facility](/images/dc-security-camera.webp) ## Choosing a Provider: What to Look For Selecting a colocation provider is a significant decision, your servers, your data, and potentially your entire business operation will depend on that provider's facility. Here are the key factors to evaluate. **Carrier neutrality**: Can you connect to multiple network providers? This is non-negotiable for most enterprise customers. Ask which providers have equipment in the meet-me room and whether there are any restrictions on connectivity. **Tier rating**: What tier of reliability does the facility target? Tier III (N+1 redundancy on power and cooling) is the minimum for enterprise workloads. Some facilities claim tier ratings without independent certification, ask whether the facility has been certified by the Uptime Institute or another recognised body. **Security**: What physical security measures are in place? Biometric access, 24/7 CCTV, mantraps, on-site security personnel, and individual rack locking should all be standard. Ask about the facility's security certifications and whether they have ISO 27001. **Power**: What is the power availability and reliability? Ask about the facility's total power capacity, the utilisation level (how much is already committed to existing customers), the generator fuel storage duration, and whether the facility has dedicated Kenya Power feeders. **SLA**: What does the Service Level Agreement guarantee? A professional colocation SLA should guarantee 99.99% or better uptime, with financial credits if the facility fails to meet this target. Read the SLA carefully, some providers exclude scheduled maintenance, force majeure events, or customer-caused outages from the SLA calculation. Colocation is not the right choice for every organisation. Small businesses with just one or two servers may be better served by cloud services. Organisations that do not want to manage any hardware at all should look at managed hosting or cloud. But for organisations that need control over their hardware, have compliance requirements that mandate data location, or have workloads that are not cost-effective to run in the cloud, colocation remains the most practical and cost-effective option in Kenya's growing data centre market. Ready to go deeper on the buying decision? Our [Nairobi colocation buyer's guide](/articles/colocation-data-centres-nairobi-buyers-guide) walks through what Tier III really means, which facilities are operational as of September 2026, and the questions to ask before you sign anything. ## What Is a Hyperscale Data Centre? URL: https://data-centers-254.vercel.app/articles/what-is-hyperscale-data-centre Cluster: Beginner Guides Published: 2026-08-28 Updated: 2026-08-28 Summary: What hyperscale data centres are, how they differ from enterprise and colocation sites, and why Kenya has not yet attracted AWS or Google. ## What Is a Hyperscale Data Centre? If you have ever streamed a Netflix show, searched on Google, or booked an Uber ride, your request was processed inside a **hyperscale data centre**, a gargantuan computing facility so large it can stretch across multiple football pitches and consume enough electricity to power a small Kenyan town. But what exactly makes a data centre "hyperscale," and why does it matter for Kenya's digital future? ![GPU servers racked in a data centre](/images/ai-gpu-servers.webp) In simple terms, a **hyperscale data centre** is a massive facility owned and operated by a single technology company to support its cloud computing, artificial intelligence, and digital services at a global scale. Unlike standard [what is a data centre](/articles/what-is-a-data-centre) facilities that serve multiple tenants, hyperscale data centres are purpose-built for one operator, typically one of the world's largest technology companies. ## How Hyperscale Differs from Enterprise and Colocation Facilities To understand hyperscale, it helps to compare it against the other two main categories of data centre. **Enterprise data centres** are built and operated by individual organisations for their own IT needs. A bank in Nairobi might operate a 500 kW facility to run its core banking systems. These are typically small, private, and designed for a single organisation's workloads. **Colocation data centres**, such as those operated by [Africa Data Centres](/articles/colocation-data-centre-kenya) and PAIX in Kenya, provide shared space, power, and cooling. Multiple customers rent rack space and bring their own equipment. Kenya's largest colocation facilities range from approximately 5 MW (PAIX Mombasa and Nairobi) to around 10 MW (Africa Data Centres' facilities). **Hyperscale data centres** exist on an entirely different scale: - **Power capacity**: 100+ megawatts of critical IT load, compared to Kenya's entire colocation market, which is estimated at roughly 20–30 MW combined. - **Server count**: Tens of thousands to hundreds of thousands of servers in a single facility. - **Physical footprint**: Often spanning hundreds of thousands of square metres across multiple buildings on a single campus. - **Power infrastructure**: Hyperscalers typically build their own **dedicated power substations** connected directly to the national grid, bypassing standard utility distribution. - **Ownership**: Single-tenant, the entire facility is owned and operated by one company (Amazon, Microsoft, Google, Meta, etc.). For context, a single hyperscale campus can consume more power than **Mombasa County's entire industrial demand**. This is why [data centre tier ratings](/articles/data-centre-tier-ratings-explained) matter enormously, hyperscale facilities are built to Tier III+ or Tier IV standards, ensuring 99.982% to 99.995% uptime. ## The Global Hyperscale Operators and Their African Footprint The hyperscale market is dominated by a handful of technology giants. As of 2025, the major hyperscale operators globally include: - **Amazon Web Services (AWS)** - **Microsoft Azure** - **Google Cloud Platform (GCP)** - **Meta (Facebook)** - **Alibaba Cloud** - **Oracle Cloud Infrastructure (OCI)** ### Where Are They in Africa? Africa's hyperscale presence is currently **concentrated entirely in South Africa**: - **AWS Africa (Cape Town) Region**: Launched in April 2020, this was the first AWS region on the African continent. It includes three Availability Zones and has been a game-changer for South African enterprises seeking low-latency cloud services. - **Microsoft Azure South Africa**: Microsoft operates two regions in South Africa, **South Africa North** (Johannesburg) and **South Africa West** (Cape Town), both launched in 2019. These were Microsoft's first cloud regions in Africa. - **Google Cloud Johannesburg**: Google launched its Johannesburg region in 2023, becoming the third hyperscaler with a physical African presence. The region serves customers across Sub-Saharan Africa. **Notice the pattern?** All three hyperscalers chose South Africa, not Kenya, not Nigeria, not Egypt. This is a critical point for understanding the barriers Kenya faces. ![Server racks in a modern data centre hall](/images/whats-inside-ai-data-center-wide.webp) ## Why No Hyperscale Data Centre in Kenya, Yet Kenya is widely recognised as East Africa's technology hub, home to a vibrant startup ecosystem, a mature fintech sector (M-Pesa, and hundreds of mobile lending platforms), and a growing digital economy. So why haven't hyperscalers built here? ### 1. Power Reliability Concerns Hyperscalers design their facilities for **99.999% uptime** at the infrastructure level. While Kenya Power (KPLC) has made significant improvements, the grid still experiences **frequency instability, voltage sags during heavy rain seasons**, and unplanned outages. A hyperscale facility consuming 100+ MW cannot afford even momentary power quality issues. The cost of building fully redundant power systems (including massive diesel generator farms and battery backup) adds significantly to the capital expenditure in markets where grid power is unreliable. ### 2. Market Size South Africa has Africa's most mature enterprise IT market, with large financial services, mining, and manufacturing sectors that are heavy cloud consumers. Kenya's cloud market, while growing at approximately 25–30% annually, is still smaller in absolute terms. Hyperscalers typically require a minimum addressable market of several hundred million dollars in annual cloud spend before committing to a $200–500 million facility build. ### 3. Regulatory Environment Kenya's regulatory landscape for data centres is still evolving. The **Data Protection Act 2019** introduced data localisation requirements that could drive demand, but the licensing framework, tax incentives (or lack thereof), and the approval process for large-scale power connections remain less investor-friendly than South Africa's. Hyperscalers look for **regulatory certainty**, clear, stable rules that won't change after they commit hundreds of millions of dollars. ### 4. Fibre Connectivity While Kenya has excellent international connectivity through submarine cables landing in Mombasa (including EASSy, TEAMS, and SEACOM), the **inland fibre backbone to landlocked neighbours** (Uganda, Rwanda, Burundi, South Sudan, and eastern DRC) is less robust. South Africa's fibre network connects more thoroughly to Southern African Development Community (SADC) countries, making Johannesburg a more attractive regional hub. ## The Economic Impact Hyperscale Facilities Bring When a hyperscaler builds a data centre, the economic ripple effects are substantial. The **AWS Cape Town region** provides the closest African example: - **Direct employment**: Hundreds of construction jobs during the build phase, followed by permanent engineering, operations, and security roles. - **Indirect employment**: Thousands of jobs created in the local ecosystem, from fibre network construction to managed service providers, system integrators, and cloud consulting firms. - **Tax revenue**: Significant contributions to corporate tax, property tax, and customs duties on imported equipment. - **Ecosystem development**: The presence of AWS, Azure, and Google Cloud in South Africa has spawned a thriving community of cloud-native startups, training academies, and partner networks. - **Foreign direct investment**: Each hyperscale facility represents $200–500 million (or more) in capital investment, sending a strong signal to other technology companies. For Kenya, attracting even a single hyperscale facility could **transform Nairobi into the undisputed cloud capital of East Africa**, creating an ecosystem that benefits the entire region. ## Kenya's Potential to Attract Hyperscale Investment Despite the current barriers, Kenya has several compelling advantages that could attract hyperscale investment in the medium term: **Growing enterprise cloud adoption**: Kenyan banks, telecom companies (Safaricom, Airtel), and government agencies are rapidly migrating to cloud. The Huduma Namba digital identity project and the e-Citizen platform are driving government cloud consumption. **Geothermal energy advantage**: Kenya is a global leader in geothermal energy, with the Olkaria geothermal complex producing over 800 MW. For hyperscalers increasingly focused on renewable energy and carbon neutrality, Kenya's geothermal resources are a genuinely unique selling point. See our guide on [geothermal energy for Kenyan data centres](/articles/geothermal-energy-kenya-data-centres) for a deeper analysis. **Strategic location**: Nairobi's position as a gateway to East Africa (a market of over 300 million people) is attractive. If fibre connectivity to neighbouring countries improves, Nairobi becomes a natural regional hub. **Young, tech-savvy workforce**: Kenya produces thousands of IT graduates annually, and the country's engineering talent is well-regarded across the continent. ### What Needs to Change To convert this potential into reality, several things must happen: 1. **Power cost and reliability must improve**: Kenya's commercial electricity tariffs, while lower than many African countries, are still higher than the global average for data centre operations. KPLC needs to demonstrate consistent power quality at the scale hyperscalers require. The development of [solar power solutions](/articles/solar-power-data-centres-kenya) for data centres is a step in the right direction. 2. **Regulatory certainty**: The Kenyan government, through the [Communications Authority](/articles/kenya-data-centre-licensing-framework) and the Kenya Investment Authority, needs to create a clear, attractive incentive framework, potentially including tax holidays, duty-free importation of data centre equipment, and streamlined power connection processes. 3. **Fibre backbone expansion**: Investment in cross-border fibre to Uganda, Rwanda, and beyond would dramatically increase Kenya's appeal as a regional hub. The [comparison between Nairobi and Mombasa](/articles/nairobi-vs-mombasa-data-centre-locations) highlights the importance of connectivity infrastructure. 4. **Market development**: Continued growth of Kenya's enterprise cloud market (driven by digital transformation in banking, government, healthcare, and education) will eventually reach the scale that justifies hyperscale investment. ## Understanding Regions and Availability Zones Hyperscale cloud providers organise their infrastructure into **regions** and **availability zones (AZs)**, concepts that are central to how hyperscale data centres operate. A **region** is a geographical area where a cloud provider has clustered data centres. For example, "AWS Africa (Cape Town)" is a region. Each region typically contains two or more availability zones. An **availability zone** is one or more discrete data centres within a region, each with independent power, cooling, and networking. The AZs within a region are connected through low-latency fibre links but are physically separated (often by several kilometres) so that a fire, flood, or power outage at one AZ does not affect the others. This architecture is why hyperscalers build **campuses** of multiple data centres rather than single massive buildings. When AWS launched in Cape Town, it opened with three AZs, meaning at least three separate data centre facilities (likely more, since each AZ can include multiple buildings). For Kenya, this means that attracting a hyperscale investment would likely involve not just one data centre, but a **campus of two to four facilities**, representing billions of shillings in investment. ## Comparing Hyperscale to Kenya's Current Data Centre Landscape | Feature | Kenyan Colocation | Hyperscale | |---------|-------------------|------------| | **Power capacity** | 1–10 MW per facility | 100–500+ MW per campus | | **Tenancy model** | Multi-tenant | Single-tenant | | **Typical operators** | Africa Data Centres, PAIX, Liquid Intelligent Technologies | AWS, Microsoft, Google, Meta | | **Number of servers** | Hundreds to low thousands | Tens of thousands+ | | **Primary customers** | Banks, telcos, government, enterprises | Cloud service users globally | | **Power infrastructure** | Utility grid + backup | Own substation + utility + backup | | **Investment scale** | KES 2–10 billion per facility | KES 30–75 billion+ per campus | Africa Data Centres' facilities in Nairobi (including the campus along Mombasa Road) represent the largest data centre investments in East Africa at approximately 10 MW. PAIX operates facilities in both Nairobi and Mombasa at around 5 MW each. Liquid Intelligent Technologies also maintains a significant presence. But even combined, these facilities are a fraction of the scale of a single hyperscale campus. ![Aerial view of a data centre campus with rooftop solar](/images/dc-environment-sustainability.webp) ## The Road Ahead for Kenya The question is not whether hyperscale data centres will come to Kenya, but **when**. The convergence of several trends, AI-driven demand for computing capacity, Africa's rapidly growing digital economy, Kenya's renewable energy advantages, and the natural saturation of South Africa as Africa's primary cloud hub, all point toward East Africa as the next frontier. Kenya's government has signalled its ambitions through the **Kenya Digital Masterplan 2022–2032**, which explicitly targets data centre development as a national priority. The **Konza Technopolis** project, while primarily focused on IT services and BPO, includes data centre infrastructure plans that could eventually attract hyperscale interest. For businesses and IT professionals in Kenya, understanding the hyperscale landscape is essential. Even without a local hyperscale presence, Kenyan organisations are already consuming AWS, Azure, and Google Cloud services, routed through the Cape Town and Johannesburg regions. This introduces additional latency (typically 40–60 milliseconds from Nairobi to Cape Town) that a local region would eliminate. ## Frequently Asked Questions ### What is the minimum power capacity of a hyperscale data centre? A hyperscale data centre typically operates at **100 megawatts (MW) or more** of critical IT load power. For comparison, most Kenyan colocation facilities currently operate between 1 MW and 10 MW. This massive power requirement is why hyperscalers often build their own dedicated power substations connected directly to the national grid. ### Why hasn't AWS, Google, or Microsoft built a data centre in Kenya yet? Several factors explain this. First, **power reliability** remains a concern, hyperscalers require near-perfect uptime, and while Kenya Power has improved, frequency stability and outage management still lag behind hyperscale standards. Second, Kenya's enterprise cloud market, while growing fast, is smaller than South Africa's. Third, **regulatory certainty** around data localisation, licensing, and taxation is still evolving. Finally, Kenya lacks the deep fibre backbone connectivity to landlocked East African countries that would make Nairobi a natural regional hub on par with Johannesburg. ### How does a hyperscale data centre differ from a colocation facility? A colocation facility rents out space, power, and cooling to multiple tenants who bring their own servers. A hyperscale data centre is owned and operated by a single company (such as Amazon, Google, or Microsoft) to run its own cloud services at massive scale. Hyperscale facilities typically have over 5,000 servers and 100+ MW of power, while colocation facilities in Kenya like Africa Data Centres and PAIX serve many customers in shared environments of 1–10 MW. ### What is the economic impact of a hyperscale data centre on a local economy? The AWS Cape Town region, launched in 2020, provides a useful reference point. It created **hundreds of direct and indirect jobs** during construction and operations, generated significant tax revenue, stimulated demand for local fibre network expansion, and created an ecosystem of partners, managed service providers, and training programmes. Similar investment in Kenya could catalyse the broader tech ecosystem and position Nairobi as the cloud capital of East Africa. ### What would Kenya need to do to attract hyperscale investment? Key requirements include **guaranteed power reliability** with redundant grid connections and competitive tariffs, a clear and stable regulatory framework for data centres, improved fibre connectivity to neighbouring countries, investment-ready land with access to large-scale water and power infrastructure, and a larger enterprise cloud market. The Communications Authority of Kenya and Kenya Investment Authority would need to work closely with hyperscalers to create incentive packages comparable to what South Africa has offered. ## Why Kenya's Data Centres Cluster in Nairobi and Mombasa URL: https://data-centers-254.vercel.app/articles/why-data-centres-cluster-nairobi-mombasa Cluster: Market Analysis Published: 2026-08-30 Updated: 2026-09-16 Summary: 23 of Kenya's 27 tracked data centres sit in two cities. The three forces behind the cluster: where cables land, where the grid is, where demand lives. Ask where Kenya's data centres are and the honest answer is still essentially two words: Nairobi, Mombasa. The DC254 directory currently tracks 27 facilities across the country, and 23 of them sit in just two cities: 19 in Nairobi, including announced projects that have yet to break ground, and 4 in Mombasa, from iColo's MBA1 and MBA2 to Telkom Kenya's Telephone House and the SEACOM cable landing station. The four outside the pair are the government's Konza National Data Centre and three Nairobi-metro edge sites in Limuru, Ruiru, and Thika. No facility exists in Kisumu, Nakuru, Eldoret, or any of Kenya's other major towns. For a country of roughly 55 million people and an economy that runs on mobile money, that is a strikingly narrow geographic footprint, and it is not an accident. This concentration is usually presented as a statistic. It is more useful to treat it as a story about three forces that decide where digital infrastructure gets built: where the international cables come ashore, where the power grid can actually deliver large blocks of electricity dependably, and where the customers with money to spend on computing are located. In Kenya, all three forces point at the same two cities, and until at least one of them changes direction, the map is unlikely to look much different. This article examines each force in turn, then asks what it would take for a third city to appear on Kenya's data centre map. ## First, the facts: 27 facilities, two cities Start with what the verified data actually shows. Of the 27 facilities in the DC254 directory, 19 are located in Nairobi County, from hyperscale halls along Mombasa Road like iXAfrica NBOX1 and the Africa Data Centres campus, to enterprise and telecom facilities in Westlands such as Safaricom's Waiyaki Way data centre, to smaller colocation rooms scattered through the industrial area. Four operate in Mombasa County: iColo's MBA1 in Miritini, MBA2 in Nyali, Telkom Kenya's Telephone House exchange, and the SEACOM cable landing station. The remaining four sit just outside the two-city frame: the government's Konza National Data Centre and three Nairobi-metro edge sites (Safaricom Red Hill in Limuru, Nxtra by Airtel at Tatu City, and Safaricom Thika). Disclosed operational IT load across the eight facilities that publish figures totals roughly 10.5 megawatts (DC254 database), modest by global standards, and almost entirely consumed within Nairobi's metropolitan economy; several operators do not disclose figures, so true installed capacity is higher. Two clarifications matter before going further. First, "two cities" is still slightly generous to Mombasa: Nairobi holds the overwhelming majority of capacity, and Mombasa's four tracked facilities remain modest in scale, valued primarily for where they stand rather than how much they hold. Second, announced projects complicate the picture but do not change it. The Microsoft–G42 joint venture, nominally a $1 billion facility, remains at announcement stage and is expected to site near Nairobi or the Olkaria geothermal fields, either way, still within the same power-and-demand corridor this article describes. ## Force 1: The cables land in Mombasa Kenya's connection to the global internet is physical, and it arrives on the ocean floor. Seven live submarine cable systems now land on or near the Mombasa coast: SEACOM, TEAMS, EASSy, LION2, DARE1, PEACE, and 2Africa, the 45,000 km consortium system live in Kenya since June 2024 and fully completed in November 2025. Africa-1 landed at Mombasa in 2024 and awaits full service, and Meta's Daraja, a 4,108 km Oman-to-Mombasa route with Safaricom as landing partner, is in development (announced October 2025; Business Daily, September 14, 2026). Every international phone call, every Netflix stream cached locally, every cloud query routed to a European or American region travels through one of a handful of landing stations on that coastline. The economics follow the infrastructure: bandwidth is cheapest and most dependable within reach of the landing stations, because backhauling traffic hundreds of kilometres inland before it reaches the world adds cost, complexity, and a failure point. ![Map of submarine cables landing on the East African coast](/images/submarine-cables-map.webp) This explains why Mombasa matters in the data centre story. iColo's MBA1 and MBA2, Telkom's Telephone House, and the SEACOM landing station exist primarily because of their location: peering, transit, and disaster-recovery sites near the landing stations, letting operators and carriers reach the cables directly without depending on the corridor to Nairobi. They are facilities whose value comes from geography rather than scale. The same logic explains why the landing stations themselves, which are technically small data facilities, cluster along a few kilometres of coast: the cables come ashore at specific beaches determined by the seabed route, and the buildings must follow. But cable access alone does not make a data centre market. If it did, Mombasa would tower over Nairobi. In practice, the coast hosts a tiny share of Kenya's capacity because cables are an input, not a demand source, and the demand lives 500 kilometres up the road, in a capital that already has fibre corridors connecting it to the landing stations. ## Force 2: The power sits between the Rift Valley and Nairobi The second force is electricity, and here Kenya's geography produces an unusual advantage with a narrow focus. The country's electricity mix is one of the greenest on Earth (geothermal, hydro, and wind supply the large majority of generation) but the single most important source for data centres is geothermal, and nearly all of it comes from one place: the Rift Valley belt that runs through Olkaria, Naivasha, and adjacent fields. Kenya's geothermal capacity is concentrated within roughly 100 kilometres of Nairobi, and the transmission corridors that move that power feed, above all, the Nairobi metropolitan load centre. ![High-voltage pylons on the backbone grid](/images/kenya-transmission-pylons.webp) For a data centre operator, this matters because compute facilities are essentially power-conversion machines: they take grid electricity and turn it into heat and computation, continuously, at a scale measured in megawatts. Kenya's roughly 14 MW of installed data centre IT load sounds small, but the next wave is not, the Microsoft–G42 project alone has been discussed at a scale of 100 megawatts, and its reported stalling illustrates the point precisely: Kenya Power cannot currently deliver that magnitude of power at a single site within the existing grid's comfortable limits. Large data centres go where large, dependable power can be contracted, and in Kenya that means sites with good transmission access, overwhelmingly the corridor between the geothermal fields and Nairobi. The consequence is subtle but important: Kenya's green power advantage is real, and it is a genuine marketing asset for attracting AI and cloud investment. But because that power is generated in the Rift Valley and consumed in Nairobi, the advantage accrues to a specific corridor rather than to the whole country. A town without a strong transmission connection to the grid sits far from the advantage no matter how green the national statistics are. ![Map of Kenya's Rift Valley geothermal sites](/images/kenya-geothermal-map.webp) ## Force 3: The demand lives in Nairobi The third force is the least technical and the most decisive: the customers are in Nairobi. Kenya's banking sector, its fintech and mobile money platforms, its largest telco, its government systems, its insurance and retail groups, and the growing community of cloud-consuming enterprises are all headquartered within a few square kilometres of each other. Nairobi is also where the Kenya Internet Exchange Point (KIXP) sits, allowing local networks to swap traffic directly instead of paying to haul it to Europe and back. Data centres are real estate for computing, and like all real estate they command a premium where their tenants cluster. ![Westlands commercial district in Nairobi](/images/nairobi-westlands.webp) The demand concentration is self-reinforcing. Banks put their systems in Nairobi facilities because latency to their branches and customers is low and their operations teams are nearby. Cloud and content platforms deploy caches and regions in Nairobi because that is where the users are. Startups build on local cloud because their enterprise customers are there. Each decision deepens the pool of interconnection, skills, and spare capacity in the capital, which in turn makes the next facility more likely to be built there. Mombasa's port economy and coastal businesses generate real IT demand, but not enough of the right kind (dense, always-on, compliance-driven workloads) to anchor a facility ecosystem on the same scale, though analysts cited by Business Daily (September 14, 2026) point to the port, the Dongo Kundu Special Economic Zone, and the city's gateway role for landlocked neighbours as demand pulls that keep strengthening. There is also a skills dimension that is easy to underestimate. Running a Tier III facility requires a small cadre of specialised engineers (electrical, mechanical, network, and facilities specialists) who are scarce anywhere in the world. Nairobi's universities, telcos, and existing facilities have created the only meaningful pool of this talent in the country. A new facility in a town without that pool either relocates the staff at cost or staffs thinly and accepts operational risk that no operator selling uptime guarantees will take. ## Why not Kisumu, Nakuru, or Eldoret? None of this means a data centre cannot be built elsewhere in Kenya. It means the economics currently do not favour it, and it is worth being precise about why. Kisumu, the western region's largest city, sits far from both the cable landing points and the geothermal corridor, served by transmission lines and fibre routes that pass through rather than concentrate there. Nakuru and Eldoret sit closer to the Rift Valley power belt, but neither has yet attracted the dense enterprise demand that fills halls, their local economies, while significant, do not host the banks, telco cores, and government workloads that justify dedicated facilities. In each case, at least one of the three forces is missing, and usually two. What upcountry towns do attract (and will increasingly attract) is edge infrastructure: small, often prefabricated sites that cache content and terminate 4G and 5G traffic close to users, linked back to the core facilities in Nairobi. Edge sites need a fraction of the power and staff of a real data centre, which makes them viable in towns where full facilities are not. The realistic near-term map of Kenyan compute, then, is not "a data centre in every county" but a two-tier structure: core capacity in Nairobi with a transit node in Mombasa, and a growing dusting of edge sites across regional towns. ## What would change the map Maps like this one change when one of the underlying forces moves. The most discussed possibility is power: if a project of the Microsoft–G42 scale secures dedicated geothermal supply at Olkaria, it would establish that Kenya can site hyperscale compute next to generation rather than next to demand, a genuinely new pattern for the country, and one that could, over a decade, pull subsequent investment toward the Rift Valley. Konza Technopolis, the planned smart city southeast of Nairobi, represents a second path: a purpose-built district with modern power and fibre provisioned in advance, effectively manufacturing the conditions that elsewhere emerged organically. Both remain promises rather than operating reality, and both sit within or beside the existing corridor rather than beyond it. The most concrete new test of that logic arrived in August 2026. Amaco Energy Group, a Greek multinational, announced a proposed $1.5 billion (Sh194 billion) AI data centre for Mombasa, with Dongo Kundu and Kilindini identified as candidate sites near the port and the Special Economic Zone, and GE Vernova engaged to supply gas turbines for the firm's Hercules power barge, an LNG-fired platform designed to generate electricity and handle cooling without depending on Kenya's national grid (Business Daily, September 14, 2026). The proposal matters precisely because it attacks the two forces that have always capped Mombasa: it would self-supply power rather than wait for the grid, and it bets on port-and-SEZ demand rather than Nairobi's enterprise base. It is still only a proposal, with no announced capacity, timeline, or government approval, but it is the first credible attempt to found a third force on the coast. We now track this proposal in its own article: [Amaco's Mombasa AI data centre, explained](/articles/amaco-hercules-mombasa-ai-data-centre), including the GE Vernova power barge agreement and the open questions. The quieter transformation is the growth of edge computing, which does not move the core facilities but steadily extends the infrastructure footprint outward. And regulatory choices, county investment incentives, data centre provisions in building codes, or localisation requirements that force government workloads into specific regions, could nudge decisions at the margin. But the honest baseline is this: cables, power, and demand all currently agree on Nairobi and Mombasa, and infrastructure geography changes slowly. The two-city map is not a failure of ambition; it is the rational response to where Kenya's inputs actually are. ## Frequently asked questions ### Why are Kenya's data centres concentrated in Nairobi and Mombasa? Because all three forces that decide data centre locations point the same way. Seven live submarine cable systems land on the Kenyan coast (SEACOM, TEAMS, EASSy, LION2, DARE1, PEACE, and 2Africa), with Africa-1 landed at Mombasa and Meta's Daraja in development, making international bandwidth cheapest near the landing stations. The grid's strongest corridors carry geothermal power from the Rift Valley into Nairobi, where large blocks of dependable electricity are easiest to contract. And the demand (banks, telcos, fintech, government systems, and the KIXP exchange) is concentrated in the capital. Facilities follow inputs and customers, and in Kenya both point to the same two cities. ### Could a data centre be built in Kisumu, Nakuru, or Eldoret? A small edge facility, yes, any town with grid power and fibre backhaul can host one. What those towns lack is the combination that makes larger facilities economic: multiple fibre routes, high-capacity dependable grid connections, dense enterprise demand, and the specialised workforce that Tier III operations require. Without a change in one of those factors, upcountry sites will remain edge deployments rather than full data centres. ### Does Mombasa's cable access make it better than Nairobi for data centres? Not by itself. Proximity to the landing stations matters most for international transit and disaster recovery, which is precisely the niche Africa Data Centres' Mombasa facility occupies. But cables are an input, not a demand source, the workloads that fill data centres sit in Nairobi, which is already connected to the coast by fibre. Mombasa is strategically important for connectivity; Nairobi is where capacity is consumed. ### What would it take for data centres to spread beyond the two cities? The most likely catalysts, in order: a solved power constraint, dedicated geothermal supply proving that hyperscale loads can sit next to generation rather than demand; continued transmission grid expansion; and the growth of edge computing pushing modular facilities into regional towns. Purpose-built districts like Konza and county-level incentives could accelerate the shift, but power and demand remain the decisive variables. ## Why Kenya's Data Centre Market Estimates Disagree URL: https://data-centers-254.vercel.app/articles/why-kenya-data-centre-market-estimates-disagree Cluster: Market Analysis Published: 2026-09-08 Updated: 2026-09-08 Summary: Six research firms value Kenya's data centre market from $266M to $606M. We fetched every report page and explain exactly why the numbers diverge. Ask five research firms for the size of Kenya's data centre market and you will get answers that cannot all be true at once. For the 2025 base year, published estimates run from **$266 million** to **$606 million**, a factor of 2.3 between the smallest and the largest. Growth forecasts stretch from **4.67% to 32%** CAGR, a seven-fold spread. And against all of them, the most methodologically transparent public source sizes Kenya's commercial colocation revenue at roughly **$25 million a year**. Somebody is measuring something very different from everybody else. This article is the one nobody else in this market writes: we fetched the actual report pages, every one on the same day, 8 September 2026, and explain, definition by definition, why the numbers disagree, and how a serious reader should use each one. This matters beyond curiosity. These figures get cited in investor decks, government policy briefs, and journalist headlines, usually without the base year, the definition, or the firm's method attached. An investor reading "$606 million market" and an operator reading "$25 million colocation revenue" will make very different decisions about the same country. Both would cite "research". Only one of them would be looking at the market as it physically exists, the one you can verify facility by facility in our [DC Directory](/directory). ![Conflicting market estimates for Kenya](/images/dc-challenges.webp) ## The estimates, side by side Here is every Kenya data centre market estimate we could verify as of 8 September 2026, with the firm, the figures, and the date, the way they should always be cited: | Firm | 2025 size | Forecast | CAGR | Last updated | |---|---|---|---|---| | Arizton | $266M | $805M by 2031 | 20.27% | June 2026 | | Ken Research | $266.0M | $805.0M by 2031 | 20.3% | July 2026 | | 6Wresearch | $369M | $622M by 2032 | 11.3% | July 2026 | | Credence Research | $509.01M (2024) | $733.34M by 2032 | 4.67% | undated | | DCMarketInsights | $606.07M | $1,240.61M by 2035 | 7.29% | undated | | Mordor Intelligence | 15 MW (capacity) | 25 MW by 2030 | 10.76% | Dec 2025 | | Xalam Analytics (D4D Hub) | ~$25M/yr colo revenue | ~5x capacity growth | 32% (2025–30) | January 2026 | Look closely at the table and several things jump out. Ken Research's headline numbers are identical to Arizton's, same $266M, same $805M, same CAGR to the decimal, which means they are one data point wearing two logos, whether through licensing, a shared upstream dataset, or circular citation. Credence and DCMarketInsights publish no verifiable dates on their free pages. Mordor is not even measuring the same quantity: it sizes capacity (15 MW growing to 25 MW) rather than revenue. Only Xalam's numbers, roughly 15 MW live, $25 million a year of colocation revenue, 55 MW under construction and ~25 MW more in pipeline, roughly five-fold growth at 32% CAGR, come from a study whose facility-level method is publicly visible, because it was published openly by the EU's Digital Investment Facility under the D4D Hub in January 2026. ## Reason 1: Nobody agrees on what "the market" is The single biggest driver of divergence is definitional, and almost none of the summaries state their definition. At least four different things get called "the Kenya data centre market": - **Colocation revenue**, what operators earn from renting racks, cages, suites, and power. This is the tightest, most defensible definition, and it is what Xalam measures: roughly $25 million a year. You can sanity-check it: ~15 MW of live commercial capacity at even a generous $150–200 per kW per month yields $27–36 million at full occupancy, and actual occupancy runs around 60–65%. - **The total data centre economy**, colocation plus enterprise self-build capex, connectivity and managed services, cloud on-ramps, and professional services. This definition gets you into the hundreds of millions, which is where Arizton's $266M and 6Wresearch's $369M plausibly live. - **Investment value**, the capital cost of building facilities, sometimes cumulative over multiple years. DCMarketInsights' $606M and Credence's $509M read like this: a single 100 MW AI campus would blow past them, but a market measured in "everything spent on data centres, recently" fits. - **Capacity**, megawatts, not dollars. Mordor's 15 MW→25 MW is honest about being this; capacity estimates at least have the virtue of being checkable against facilities. None of these definitions is wrong. But adding them together in a headline, or worse, comparing them against each other, produces exactly the confusion the Kenyan market suffers from. When we carry market numbers in our own [market-in-numbers piece](/articles/kenya-data-centre-market-numbers), we label each figure with firm, date, and what it appears to measure, because the label is most of the information. ## Reason 2: Circular citation and the illusion of consensus Market research has a supply chain, and Kenya sits at its thin end. The global firms maintain deep models for the US, Europe, and China; African markets get modelled from fewer inputs, and smaller research houses often license or adapt figures from larger ones. The result is a phenomenon worth naming: **two reports citing each other's number is not independent confirmation**. Ken Research quoting Arizton's exact $266M→$805M at the same CAGR is the clearest documented case in this market, but the pattern is general, search for any Kenyan market stat and you will find the same figures echoing across aggregator sites with the original source stripped away. The practical rule for readers: count sources, not reports. Before you describe a number as "confirmed by multiple studies", check whether the studies share a lineage. In our coverage, we try to keep at least one methodology-visible source (currently Xalam via the D4D Hub) and at least one definitional outsider (a capacity-based count like Mordor's) in the same conversation, because agreement between different methods is evidence, while agreement between copies is just repetition. ## Reason 3: Base years, forecast windows, and the moving target The Kenyan market is small enough that a single facility launch moves it by double digits. Digital Realty's 6.4 MW NBO2, opened 7 September 2026, increased the country's verified commercial colocation capacity by roughly 40% in one day, on paper, at least, until occupancy fills in. A model with a 2024 base year (Credence), a 2025 base year (Arizton, 6Wresearch), and a mid-2025 cut-off (Xalam's January 2026 publication of its Kenya briefing) are therefore measuring visibly different markets. Add different forecast windows (2030 vs 2031 vs 2032 vs 2035) and identical underlying data would still produce different CAGRs and headline "future sizes". This is also why the same firm's numbers move between editions. Arizton's Kenya page was updated in June 2026; its earlier editions carried different totals. That is not falsification, it is a small market repricing as real capacity lands. But it does mean any citation without a date is close to meaningless, and it is why our [methodology](/methodology) attaches a fetch date to every claim we publish, including the ones in this article. ## Reason 4: Paywalls, summaries, and what journalists actually read Almost none of the people citing these reports have read them, because almost all of them are behind paywalls costing thousands of dollars. What circulates publicly is the free executive summary (typically one paragraph and a chart) plus secondary reporting that quotes the summary. The definitional subtleties that explain the divergence (what revenue is counted, what is excluded, how the base year is constructed) live in the paid body of the report, not the summary. So the public conversation runs on numbers whose definitions are invisible by construction. The honest response is not to buy six reports; it is to read the free layer for what it is (directional and named) and to anchor on what can be verified for free. PeeringDB's facility records, operator disclosures, Uptime's certification directory, and registry data cost nothing, which is why our directory builds on them and why a publicly released study like the Xalam/D4D Hub briefing is worth so much: it is the only estimates document in this stack whose method a stranger can audit. ## How to read any Kenya market number (the working checklist) After walking through the divergence, here is the checklist we apply before repeating any market figure, and that we suggest you apply to ours: 1. **Name the firm and the date.** "Arizton, June 2026 edition", not "research shows". 2. **Identify the quantity.** Revenue, investment, or capacity? Colocation-only or the total economy? If the summary does not say, the number is decoration. 3. **Check for siblings.** Identical figures at two firms usually mean one shared source, not two confirmations. 4. **Prefer the auditable.** A study whose method is public (Xalam/D4D Hub) beats a bigger number you cannot inspect. Free registry data beats both when it exists. 5. **Reconcile with physical reality.** Any revenue claim should roughly cohere with live megawatts, occupancy, and pricing; any capacity claim should cohere with the facility list. Our directory exists precisely so this check takes minutes, not weeks. Run the $266M and $606M figures through that checklist and they stop contradicting each other: they were never measuring the same thing. Run the "$25 million colocation market" through it and it becomes the most credible single number in circulation, the one consistent with fifteen-odd megawatts of live commercial capacity at regional pricing. That is not a smaller story than "$606 million"; it is a true one, and for an emerging market the true version is more useful: it tells you where the real upside is, from a small, verified base growing five-fold, not from a large number that was never measurable to begin with. ## Wingu Africa: Kenya's Edge Data Centre for Cloud URL: https://data-centers-254.vercel.app/articles/wingu-africa-data-centre-kenya Cluster: Operators Published: 2026-08-29 Updated: 2026-08-29 Summary: Wingu Africa operates a carrier-neutral edge data centre in Nairobi built for cloud-on-ramp, CDN, and enterprise workloads. What it offers. Not every data centre in Kenya needs to be a multi-megawatt hyperscale campus. As the country's digital economy matures, a significant and growing portion of infrastructure demand comes from organisations that need something different: proximity to users, fast access to cloud platforms, flexible deployment terms, and carrier-neutral connectivity without the cost and complexity of a large-scale facility. Wingu Africa was built to serve precisely this segment, positioning itself as Nairobi's dedicated edge data centre for cloud on-ramp, content delivery, and enterprise workloads. ![Server racks in a modern data centre hall](/images/whats-inside-ai-data-center-wide.webp) This profile examines Wingu Africa's origins as a Kenyan-born company, its edge computing focus, how it differentiates from larger operators in the market, and its role in the broader Kenyan digital ecosystem. ## Who Is Wingu Africa? Wingu Africa is the data centre operating arm of the Wingu Group, a Kenyan technology company founded and headquartered in Nairobi. The name Wingu is a Swahili word meaning cloud, a deliberate choice that signals the company's core mission of bridging local enterprises and content providers to global cloud platforms. Unlike several of Kenya's data centre operators that are backed by international private equity or are subsidiaries of multinational groups, Wingu Group was established by Kenyan entrepreneurs with deep experience in the local technology and connectivity market. This domestic origin is significant. It means Wingu's approach to facility design, pricing, customer engagement, and service delivery is shaped by an understanding of the Kenyan market that comes from operating within it, not from importing a playbook developed for other continents. The Wingu Group encompasses several related businesses spanning connectivity, cloud services, and managed IT, with Wingu Africa focused on the physical data centre infrastructure layer. This integrated structure gives the company an advantage in understanding the full stack of customer requirements, from network provisioning to cloud interconnection to ongoing management. ## The Nairobi Edge Data Centre ![Server racks with structured patch cabling](/images/server-rack-patch-cabling.webp) Wingu Africa operates a carrier-neutral edge data centre facility located in Nairobi. The facility is designed and positioned as an edge computing node rather than a large-scale campus, meaning it prioritises low-latency access to end users, flexible deployment sizes, and rapid provisioning over the massive power capacity and rack counts that define hyperscale facilities. An edge data centre in the Nairobi context serves a specific function. Kenya's internet users generate traffic that is best served from infrastructure within the city rather than routed through facilities further afield or from platforms hosted in Europe or South Africa. By placing compute and content caching infrastructure in Nairobi, Wingu Africa reduces round-trip times and materially improves the end-user experience for mobile banking, streaming, and SaaS applications. The facility's design reflects this edge-oriented philosophy. Key characteristics include: - **Carrier-neutral operation:** Multiple network providers can deliver connectivity to the facility, ensuring tenants are not locked into a single telecom operator. This is particularly important in Kenya, where the choice of connectivity provider directly affects both cost and performance. - **Cloud-on-ramp capability:** The facility provides direct, low-latency connectivity to major cloud platforms. For Kenyan enterprises migrating workloads to AWS, Microsoft Azure, Google Cloud, or Oracle Cloud, Wingu offers a local point of presence that eliminates the need to backhaul cloud traffic to distant interconnection points. - **Flexible deployment sizes:** Rather than requiring customers to commit to large rack deployments or multi-year contracts for significant power allocations, Wingu accommodates smaller footprints. This makes the facility accessible to SMEs, startups, and content providers whose infrastructure needs do not justify a hyperscale deployment. - **Content delivery optimisation:** The facility is specifically designed to host content delivery network (CDN) nodes and caching infrastructure, making it a natural home for content providers and media companies that need to serve the Kenyan and East African market with minimal latency. ## Edge Computing and Cloud-On-Ramp Focus Wingu Africa's positioning around edge computing and cloud on-ramp services reflects a deliberate strategic choice. Globally, the growth of edge computing, driven by real-time applications, IoT, 5G, and content delivery, has created demand for smaller facilities positioned closer to users. Kenya is experiencing a compressed version of this evolution. The country is simultaneously building its first hyperscale facilities, like iXAfrica's NBOX1, and developing demand for edge infrastructure. These are complementary rather than competing approaches. Hyperscale facilities handle large, centralised workloads including AI training and regional cloud regions. Edge facilities like Wingu handle the last-mile compute, content caching, and cloud access that directly serve end users. The cloud-on-ramp function is particularly important in the Kenyan context. Enterprises are migrating to cloud platforms, driven by the need for scalability, reduced capital expenditure, and regulatory requirements around data protection and sovereignty. The Kenya Data Protection Act, enacted in 2019, creates a concrete regulatory incentive to use local infrastructure rather than hosting workloads in foreign jurisdictions. Wingu's facility serves as the local entry point to global cloud platforms, allowing enterprises to maintain a physical presence in Kenya while accessing the full range of cloud services. This hybrid model, where compute and data reside locally but connect to global cloud platforms, is becoming the dominant architecture for Kenyan enterprises navigating performance and regulatory requirements. ## CDN and Content Delivery Role Content delivery is a significant and growing use case for data centre infrastructure in Kenya. As streaming services, digital media platforms, and e-commerce sites expand across East Africa, demand for locally cached content has increased. Content served from a Nairobi facility reaches Kenyan users in a fraction of the time required to fetch it from Europe or South Africa. Wingu Africa has positioned itself to capture this demand. The facility is designed to host CDN edge nodes, streaming caches, and other content delivery infrastructure. This makes it an attractive proposition for: - **Global CDN operators** looking to establish or expand their East African presence with a local point of presence in Nairobi. - **Streaming and media companies** serving the Kenyan and wider East African market that need low-latency content delivery. - **E-commerce platforms** whose performance depends on fast page loads and responsive application behaviour for users across the region. - **Over-the-top (OTT) service providers** delivering video, music, and other media content to mobile subscribers in Kenya and neighbouring markets. The content delivery role is closely linked to Wingu's carrier-neutral model. CDN operators benefit from connecting to multiple networks within a single facility, ensuring cached content reaches users regardless of which telecom operator they subscribe to. In Kenya's competitive telecom market, where Safaricom, Airtel, and others each serve significant subscriber bases, this multi-network accessibility is essential. ![Corridor inside a data centre hall](/images/hero-server-hall.webp) ## Target Market: SMEs, Cloud Providers, and Content Companies Wingu Africa's customer base is distinct from the hyperscale-focused operators in the Kenyan market. The company targets three primary segments. **Small and medium enterprises.** Kenya has a vibrant SME sector that accounts for a significant share of economic activity. These businesses are increasingly digitising their operations, adopting cloud applications, and requiring reliable infrastructure to host their workloads. However, most SMEs do not need and cannot justify the scale or cost of deployment at a hyperscale facility. Wingu's flexible, smaller-scale offering addresses this gap directly, providing enterprise-grade infrastructure with terms and deployment sizes that are accessible to mid-market organisations. **Cloud service providers and systems integrators.** Kenya's growing cloud ecosystem includes local and regional cloud service providers, managed service providers, and systems integrators that need infrastructure from which to deliver their own services. These companies often require carrier-neutral facilities where they can connect to multiple upstream providers and deliver services to their own customers across different networks. Wingu's model is well-suited to this channel, serving as the physical infrastructure platform for a broader ecosystem of cloud and IT service providers. **Content providers and media companies.** As discussed, the content delivery segment represents a natural fit for Wingu's edge-oriented facility. Content providers benefit from Nairobi's position as the connectivity hub for East Africa and from the low-latency access to end users that an edge facility provides. ## How Wingu Differs from iXAfrica and Africa Data Centres Understanding Wingu Africa's position in the Kenyan market requires recognising how it differs from the two largest alternative colocation providers. **Scale and capacity.** iXAfrica's NBOX1 campus targets 16 MW of IT load capacity across multiple phases, with plans for an additional 53 MW at a second campus. Africa Data Centres operates multiple facilities across Nairobi and Mombasa as part of a Pan-African network of 40-plus facilities. Wingu, by contrast, operates a smaller, purpose-built edge facility. The scale difference is not a deficiency but a reflection of a different market segment. Wingu is not competing for the same hyperscale deployments that iXAfrica and ADC target. **Power density.** iXAfrica has engineered its facility for rack densities of 40 to 50 kW per rack, designed to serve GPU clusters and AI workloads. Wingu's edge facility is designed for standard enterprise and content delivery workloads that typically require 3 to 10 kW per rack. These are the workloads that make up the vast majority of enterprise and CDN deployments in Kenya. **Capital structure.** iXAfrica is backed by Actis and Helios Investment Partners, while Africa Data Centres is a subsidiary of Cassava Technologies, listed on the London Stock Exchange. Wingu Group is a Kenyan-born company with local ownership and investment. This gives it operational flexibility and a decision-making structure that can respond quickly to local market conditions without the governance overhead of multinational ownership. **Target customer.** iXAfrica is actively pursuing global hyperscale cloud providers and large enterprise customers. Africa Data Centres serves a broad enterprise market across its Pan-African platform. Wingu focuses on the mid-market: SMEs, cloud providers, content companies, and regional enterprises that need reliable, flexible colocation without hyperscale overhead. These differences make Wingu a complementary player in the ecosystem. The Kenyan market is large and diverse enough to support multiple facility types, and a dedicated edge operator expands the overall market rather than cannibalising demand from larger facilities. ## Carrier-Neutral Model Carrier neutrality is a foundational principle of Wingu Africa's offering. In a carrier-neutral facility, tenants choose their own connectivity providers rather than being limited to the facility operator's preferred network. This model has become the global standard for professional colocation, and its importance in Kenya cannot be overstated. Kenya's connectivity market features multiple competing providers, including Safaricom, Airtel Kenya, Jamii Telecommunications, WIOCC, SEACOM, Liquid Datapoint, and others. Each provider has different network characteristics, pricing structures, and geographic strengths. An enterprise customer's choice of connectivity provider affects latency to different destinations, resilience through diverse path routing, and total cost of ownership. Wingu's carrier-neutral model means customers select the provider or combination of providers that best meets their requirements. A content delivery company might connect to multiple telecom networks to ensure cached content reaches users on every mobile network. A cloud service provider might choose one provider for international connectivity and another for local last-mile delivery. This flexibility is valuable in Kenya, where interconnection quality between networks can vary and where regulatory and commercial dynamics create distinct advantages for different providers. ## Sustainability Approach Sustainability in Kenya's data centre sector benefits from a structural advantage that is unique among African markets: the country's electricity grid is approximately 93 percent renewable, dominated by geothermal energy from the Olkaria geothermal fields in the Rift Valley. This means any data centre operating on Kenya's national grid, including Wingu Africa's facility, starts with a significantly lower carbon footprint than equivalent facilities in markets dependent on coal, gas, or diesel-generated electricity. Wingu has incorporated sustainability through practical efficiency measures rather than aspirational commitments. Facility-level efficiency remains important for controlling operating costs and demonstrating environmental responsibility to customers who increasingly include sustainability criteria in their vendor evaluations. The company employs energy-efficient cooling systems appropriate for its edge facility and power management practices that minimise waste. As international customers impose more stringent ESG requirements on their supply chains, Wingu's position on Kenya's predominantly green grid provides a natural sustainability advantage. ## Partnerships and Ecosystem Wingu Africa operates within a broader ecosystem of technology and connectivity partners. The Wingu Group's integrated structure, spanning connectivity, cloud services, and managed IT alongside the data centre operation, creates natural partnerships with technology vendors, network operators, and cloud platforms. Connectivity partnerships with multiple network providers enable the carrier-neutral model. Technology partnerships with cloud platforms and infrastructure vendors ensure the facility supports modern enterprise workloads. Relationships with systems integrators and managed service providers create a channel through which Wingu's infrastructure reaches end customers who may not directly engage with a colocation provider. This ecosystem approach suits the Kenyan market, where enterprise technology adoption often involves intermediaries including systems integrators, value-added resellers, and managed service providers. By positioning itself as the infrastructure platform these intermediaries use, Wingu extends its market reach beyond customers who would directly engage with a colocation provider. ## Future Plans Wingu Africa's growth trajectory will follow the demand it serves. As Kenya's digital economy expands, the need for edge infrastructure in Nairobi and potentially other Kenyan cities will increase. The continued migration of enterprises to cloud platforms will sustain demand for cloud-on-ramp infrastructure. The growth of content consumption, driven by affordable smartphones, expanding broadband access, and new streaming services, will increase demand for CDN and content caching infrastructure. Regulatory developments around data protection and potential data localisation requirements could create additional demand for local data centre capacity. Wingu's natural expansion path would involve increasing capacity at its existing Nairobi facility and potentially establishing additional edge nodes in other locations within Kenya or the wider East African region. The edge computing model is inherently distributed, meaning growth for Wingu is more likely to take the form of additional, smaller facilities in strategic locations rather than a single massive campus expansion. ## Role in Kenya's Digital Ecosystem Wingu Africa occupies a specific and valuable niche in Kenya's data centre landscape. It is not trying to be the largest facility or to attract the biggest hyperscale deployments. Instead, it serves as the accessible, flexible, locally-owned infrastructure option for the broad middle of the Kenyan digital market. This role matters. It provides infrastructure access to SMEs and mid-market enterprises that are the backbone of Kenya's economy but are often underserved by facilities designed for large corporate and hyperscale customers. It provides a local platform for cloud service providers and content companies that need Nairobi-based infrastructure without hyperscale overhead. As a Kenyan-born company, it contributes to local expertise and ownership in a sector increasingly attracting international capital. The presence of operators like Wingu, alongside iXAfrica and Africa Data Centres, signals the maturation of Kenya's data centre market. A healthy ecosystem requires providers at different scales, serving different customer segments, with different ownership structures and strategic orientations. Wingu Africa's edge-focused, carrier-neutral, locally-rooted approach is a necessary and complementary part of that ecosystem. For a full list of data centre operators and facilities in Kenya, visit our [data centre directory](/directory). To learn more about the demand driving edge infrastructure, see our article on [edge computing in East Africa](/articles/edge-computing-east-africa). If you are new to colocation, our guide to [colocation in Kenya](/articles/what-is-colocation-kenya) covers the fundamentals. ## Women in Data Centres: Kenya's Technical Workforce URL: https://data-centers-254.vercel.app/articles/women-in-data-centres-kenya Cluster: Careers Published: 2026-08-28 Updated: 2026-08-28 Summary: Gender diversity in Kenya's data centre industry: current participation, barriers, career pathways, and the initiatives driving change. ## The Gender Gap in Kenya's Data Centre Industry Kenya's data centre industry is growing rapidly, driven by the country's position as East Africa's digital hub and the increasing demand for cloud services, colocation, and digital infrastructure. Yet behind the gleaming server racks and sophisticated cooling systems lies a persistent challenge: **the significant underrepresentation of women in technical data centre roles**. ![A packed tech community session in Nairobi, women building careers in Kenya's digital economy](/images/tech-community-event-nairobi.webp) While comprehensive data centre-specific workforce statistics are not publicly available in Kenya, broader Kenyan tech sector data provides a reliable indicator. According to multiple industry surveys, **women make up approximately 20-30% of Kenya's technology workforce**. Within data centres specifically (where roles span electrical engineering, mechanical systems, networking, and facilities management) the representation is estimated at the lower end of this range, around **20-25%**. This is not merely a diversity metric. Kenya faces a genuine **talent shortage in data centre operations and engineering**, with demand for qualified professionals consistently outstripping supply. Excluding half the population from this talent pipeline is not just an equity issue, it is an economic constraint that limits the industry's growth potential. ## Understanding the Barriers ### Perception of Data Centres as Male-Dominated Spaces Data centres have historically been associated with heavy engineering, electrical systems, HVAC, physical security, and rack-level hardware work. These fields have long been male-dominated in Kenya and globally, creating a **self-reinforcing perception** that data centre careers are not for women. This perception is particularly strong in Kenya's technical training institutions, where data centre-specific courses are rare and students in electrical and mechanical engineering programmes are predominantly male. Without visible female role models in the industry, the perception persists that these careers are not open or welcoming to women. ### Gaps in the STEM Pipeline Kenya's education system shows a well-documented gender gap in STEM (Science, Technology, Engineering, and Mathematics) subjects at the secondary and tertiary levels. While girls' enrolment in STEM subjects has been improving (supported by government policy and NGO interventions) the pipeline from STEM education to data centre careers remains weak. The challenge is not just about getting women into STEM, but about connecting STEM graduates to infrastructure careers. Many young women completing degrees in electrical engineering, computer science, or IT in Kenya are unaware that data centres represent a viable and often lucrative career path. ### Workplace Culture and Retention Even when women enter the [data centre careers in Kenya](/articles/data-centre-careers-kenya), retention can be a challenge. Data centre operations often require shift work, on-call responsibilities, and time spent in physically demanding environments (server rooms with high noise levels, raised floors, and controlled temperatures). Without supportive workplace policies (including flexible working arrangements, adequate maternity leave, and childcare support) some women find it difficult to sustain long-term careers in the sector. ### Limited Mentorship and Networking The [career progression in African data centres](/articles/career-progression-african-data-centres) often depends on informal networks and mentorship relationships. In a male-dominated industry, women may have fewer opportunities to build these relationships, limiting their access to career advancement opportunities, project assignments, and industry knowledge. ## Women Leading Change in Kenya's Tech Infrastructure ![Instructor leading an ICT training class in Kenya](/images/classroom-ict-training-kenya.webp) Despite the barriers, a growing number of women are making significant contributions to Kenya's data centre and broader tech infrastructure sector. While the industry does not always celebrate these contributions publicly, several women have emerged as leaders and role models: **In corporate data centre operations**, women are increasingly holding senior positions in project management, client relations, and strategic planning at facilities operated by companies like **Africa Data Centres** and **Liquid Intelligent Technologies**. These leaders are shaping how data centres are designed, built, and operated in Kenya. **In the telecom and connectivity space**, women hold key engineering and management positions at **Safaricom**, **Airtel Kenya**, and **Kenya Data Centres**, overseeing the infrastructure that underpins the country's digital economy, including the fibre networks, submarine cable landing stations, and interconnection facilities that connect Kenyan data centres to the global internet. **In the startup and innovation ecosystem**, women-founded and women-led technology companies are emerging in areas like edge computing, IoT management, and data analytics, all of which have direct relevance to the data centre industry. The presence of these women in visible leadership roles is critical. As the [McKinsey Diversity Wins report](https://www.mckinsey.com/featured-insights/diversity-inclusion/diversity-wins-how-inclusion-matters) demonstrated, companies in the top quartile for gender diversity are **25% more likely to achieve above-average profitability**. This is not a coincidence, diverse teams bring different perspectives that improve problem-solving, innovation, and decision-making. ## Initiatives Driving Change in Kenya ### Women in Tech Africa (Kenya Chapter) **Women in Tech Africa** is one of the continent's largest communities of women in technology, with an active Kenya chapter that runs networking events, mentorship programmes, and skills development workshops. While the organisation covers the broader tech sector, it has increasingly focused on infrastructure careers, including data centres, cloud computing, and networking. ### Women in Engineering Kenya (WIEK) **Women in Engineering Kenya** focuses specifically on women in engineering disciplines, including electrical, mechanical, and civil engineering, all of which are directly relevant to data centre design and operations. WIEK runs career guidance programmes in secondary schools and universities, helping young women understand the range of engineering careers available, including those in the data centre sector. ### AkiraChix **AkiraChix** is a Kenyan organisation that has been pivotal in training and mentoring young women in technology. Founded in 2010, the organisation has trained hundreds of women in coding, hardware, and technology entrepreneurship. Several AkiraChix alumni have gone on to careers in IT infrastructure, networking, and cloud computing, skills that are directly transferable to data centre environments. ### Microsoft Africa Development Centre (ADC) Microsoft's **Africa Development Centre** in Nairobi has implemented comprehensive diversity programmes that include targeted hiring of women engineers, mentorship pairings, and leadership development for high-potential female staff. While Microsoft ADC is not a data centre operator per se, its programmes are creating a pipeline of skilled women technologists who may choose data centre careers. ### Safaricom's Gender Initiatives **Safaricom**, Kenya's largest telecommunications company and a major data centre operator, has published gender diversity targets and runs programmes including the **Safaricom Women in Technology** initiative. These programmes include technical training, leadership development, and return-to-work programmes for women who have taken career breaks. ## Career Pathways with Growing Female Participation ![Executives cutting the ribbon at the NBO2 data centre launch in Nairobi](/images/nbo2-launch-ribbon-cutting.webp) ### Project Management Project management is one of the data centre career pathways with the highest and fastest-growing female participation in Kenya. Data centre construction, expansion, and upgrade projects require skilled project managers who understand both the technical and business dimensions of infrastructure delivery. Certifications like **PMP** and **PRINCE2** are well-established in Kenya, and the structured nature of project management makes it accessible to professionals transitioning from other sectors. ### Compliance and Governance As Kenya's data protection regulatory framework matures (driven by the **Data Protection Act 2019** and the Office of the Data Protection Commissioner) demand for compliance and governance professionals in data centres is growing. Roles in this pathway include data protection officers, compliance managers, and audit specialists. This pathway has seen strong female participation, partly because it draws on legal and business backgrounds where women are better represented. ### Sustainability and ESG Roles Environmental, Social, and Governance (ESG) considerations are becoming increasingly important for data centre operators, particularly as international clients and investors demand sustainability reporting. Roles in sustainability management, energy efficiency optimisation, and environmental compliance are attracting women professionals with backgrounds in environmental science, engineering, and corporate sustainability. ### Network Engineering Network engineering is a core technical discipline in data centres, responsible for designing and managing the complex network infrastructure that connects servers, storage systems, and external connectivity. While traditionally male-dominated, this pathway is seeing growing female participation as organisations like **Cisco Networking Academy** and **Huawei ICT Academy** in Kenya actively recruit women into their programmes. ### Operations Management Data centre operations management (overseeing the day-to-day running of facilities, managing teams, and ensuring service level agreements are met) is a career pathway that combines technical knowledge with management skills. Women in this pathway often progress from technical roles or enter from related fields like facilities management or IT operations. ## What Companies Are Doing ### Diversity Hiring Targets Several of Kenya's major data centre operators and technology companies have established formal diversity hiring targets. While these targets are often company confidential, the public commitments from organisations like **Africa Data Centres** (part of the Remgro group) and **Liquid Intelligent Technologies** indicate a target of **30-40% female representation** in technical roles within the next three to five years. ### Mentorship Programmes Formal mentorship programmes pair experienced professionals with women who are earlier in their careers. These programmes provide guidance on technical skills development, career planning, and navigating workplace challenges. Companies like **Safaricom**, **Microsoft ADC**, and **Amazon Web Services** (which has a presence in Kenya) run structured mentorship programmes that include data centre and infrastructure professionals. ### Flexible Working Arrangements The shift towards cloud-managed infrastructure and remote monitoring tools has made it increasingly possible for data centre professionals to perform many of their duties remotely. Companies that offer **hybrid working models**, flexible hours, and remote management capabilities are finding it easier to attract and retain women in data centre roles. This is particularly important for mid-career women who may have family responsibilities. The COVID-19 pandemic accelerated the adoption of remote management tools in Kenyan data centres, and many operators have retained these capabilities as a competitive advantage in talent attraction. ## The Business Case for Diversity Beyond the moral and social arguments for gender diversity, there is a compelling **business case** that is particularly relevant in Kenya's data centre industry: **Talent shortage mitigation**: Kenya's data centre industry faces a significant shortage of qualified professionals. The [UN Women Africa report on women in technology](https://africa.unwomen.org/en/digital-economy) highlights that closing the gender gap in Africa's tech workforce could add billions of dollars to the continent's GDP. For data centre operators competing for a limited talent pool, tapping into the full workforce is not optional, it is a competitive necessity. **Improved performance**: Multiple studies, including McKinsey's research, have consistently shown that diverse teams outperform homogeneous ones. In the complex, high-stakes environment of data centre operations, where decisions about reliability, security, and efficiency have significant financial implications, diverse perspectives lead to better outcomes. **Client requirements**: International clients (particularly hyperscale cloud providers, multinational enterprises, and financial institutions) increasingly require their service providers to demonstrate diversity and inclusion commitments as part of procurement processes. Data centre operators that cannot meet these requirements risk losing business. ## Practical Steps for Women Entering Kenya's Data Centre Industry For women considering a career in Kenya's data centre industry, several practical steps can help navigate the entry process: ### Understand the Entry Points The data centre industry has multiple entry points that suit different backgrounds and skill levels. [Learning how to get a job in a Kenyan data centre](/articles/how-to-get-job-kenyan-data-centre) requires understanding that roles span from facilities and electrical engineering to networking, cloud operations, and management. Identify which pathway aligns with your existing skills and interests. ### Pursue Relevant Certifications Certifications are particularly valuable in Kenya's data centre industry because they provide standardised proof of competence. Start with foundational certifications like **CDCP** (Certified Data Centre Professional) or **CompTIA Network+**, then progress to specialisations based on your chosen career pathway. Many certification providers offer online study options and some, like the Uptime Institute, have begun offering scholarships for women in emerging markets. ### Join Professional Networks Joining organisations like **Women in Tech Africa**, **Women in Engineering Kenya**, and **AkiraChix** provides access to mentorship, job opportunities, and peer support. Industry events like the **Africa Data Centres Summit** and **KENYA ICT WEEK** also offer networking opportunities. ### Seek Out Companies with Strong Diversity Records When evaluating potential employers, research their diversity and inclusion policies. Companies that publicly publish diversity data, have active employee resource groups for women, and offer mentorship programmes are more likely to provide supportive working environments. ### Salary Expectations Understanding salary expectations is important for effective negotiation. Entry-level data centre roles in Kenya typically offer **KES 60,000-120,000 per month**, mid-level roles command **KES 150,000-350,000**, and senior positions can reach **KES 400,000-800,000 or more**. Research market rates through industry surveys, recruitment agencies, and professional networks to ensure fair compensation. ## Kenya's Position in the African Context Kenya's approach to gender diversity in the data centre sector reflects broader African trends but with some distinctive features. **South Africa** leads the continent in data centre gender diversity initiatives, driven by a more mature industry and stronger regulatory requirements for workplace equity. **Nigeria** has a large and growing tech ecosystem with active women-in-tech communities. **Rwanda** has made gender equality a national policy priority, with notable results in the tech sector. Kenya's strengths include its vibrant tech startup ecosystem, strong telecommunications infrastructure, and a growing number of organisations specifically focused on women in technology. The challenge is translating this ecosystem energy into concrete representation in the data centre industry, which has historically been less visible and less accessible than software development or digital services. ## Conclusion The data centre industry in Kenya is at an inflection point. Rapid growth in demand for digital infrastructure, combined with a persistent talent shortage, creates both the imperative and the opportunity to build a more gender-diverse workforce. The barriers are real but increasingly well-understood, and a growing ecosystem of organisations, programmes, and role models is working to dismantle them. For women considering a career in this field, the timing has never been better. The industry needs talent, it values competence, and the pathways to entry are more accessible than ever. For companies operating in the sector, investing in gender diversity is not just the right thing to do, it is a strategic imperative that will determine competitive advantage in the years ahead. ## Frequently Asked Questions ### What percentage of data centre technical roles in Kenya are held by women? While precise data centre-specific figures are not publicly available, broader Kenyan tech sector data indicates that women hold approximately 20-30% of technical roles. In data centres specifically, the figure is estimated at the lower end of this range, around 20-25%, with higher representation in project management, compliance, and sustainability roles, and lower representation in hands-on engineering and facilities operations. ### What certifications should women pursue for data centre careers in Kenya? Key certifications include: CDCP (Certified Data Centre Professional) for general knowledge, CDCS (Certified Data Centre Specialist) for design and operations, CompTIA Network+ and CCNA for networking roles, Uptime Institute's ATD or AOS for operations and management, and PMP or PRINCE2 for project management. For those entering the field, CompTIA A+ and CDCP provide excellent foundational knowledge without requiring extensive prior experience. ### Which organisations in Kenya support women in tech and data centre careers? Notable organisations include Women in Tech Africa (with an active Kenya chapter), Women in Engineering Kenya (WIEK), AkiraChix (focused on young women in tech), the Microsoft Africa Development Centre's diversity programmes, and Safaricom's Women in Technology initiative. Additionally, professional bodies like the ICT Authority and the Kenya ICT Business Alliance (KICTA) have begun incorporating gender diversity into their workforce development agendas. ### What is the salary range for women entering data centre roles in Kenya? Entry-level data centre roles in Kenya typically offer KES 60,000-120,000 per month for technical positions such as junior network engineers or operations technicians. Mid-level roles in project management, compliance, or sustainability command KES 150,000-350,000. Senior positions such as data centre managers or head of infrastructure can earn KES 400,000-800,000 or more. These figures are comparable across genders, though the gender pay gap in Kenya's tech sector means women may start 5-15% below male counterparts in some organisations. ### How is the African data centre industry addressing gender diversity? Across Africa, major operators including Africa Data Centres, PAIX, and teraco have implemented diversity hiring targets and mentorship programmes. The African Data Centre Association (ADCA) has identified workforce diversity as a strategic priority. Countries like South Africa and Nigeria lead in gender diversity initiatives, but Kenya is gaining momentum through a combination of grassroots organisations, corporate programmes, and regulatory encouragement from bodies like the Communications Authority and the ICT Authority. ## World Bank AI Compute Plan: Rent or Build for Kenya? URL: https://data-centers-254.vercel.app/articles/world-bank-shared-ai-compute-kenya Cluster: Cloud Computing Published: 2026-10-24 Updated: 2026-10-24 Summary: The World Bank's new initiative lets governments rent pooled GPU compute. For Kenya, the question is whether shared AI infrastructure serves or sidesteps local build-out. ## What the Bank Announced In Seoul on 20 October 2026, the World Bank Group launched its first digital strategy and, alongside it, the AI Infrastructure and Readiness Initiative: a programme to give governments in Africa and Latin America shared, pay-as-you-go access to GPU computing. The Bank frames it as "a practical alternative to standalone national data centres," with Cassava Technologies and NVIDIA named as knowledge partners. President Ajay Banga tied the package to what the Bank calls "small AI, AI for the people," arguing that digital technology helps countries create jobs and make scarce resources go further. Read that framing carefully, because it contains a quiet provocation. The World Bank, the institution that has financed power grids and undersea cables for decades, is telling governments there is a path into AI that does not begin with building a national data centre. For a continent holding roughly 0.6 percent of global data centre capacity against 18 percent of the world's population, that is either the most sensible shortcut on offer or a detour around problems that still have to be solved on the ground. For Kenya specifically, the answer turns on rules the Bank's announcement never mentions: where the GPUs sit, and what Kenyan law already requires to stay home. ## The Numbers Behind the Push The Bank's own figures explain the urgency. Sub-Saharan Africa accounts for about 18 percent of the world's population but holds roughly 0.6 percent of global data centre capacity, and only around 5 percent of existing African facilities are ready for AI workloads, according to figures the Bank has circulated with the announcement. Measured differently, Africa averages about 0.33 MW of data centre capacity per million people, against roughly 15 MW globally. The readiness picture is just as uneven as the infrastructure. The Bank's Africa Economic Update, published a day after the announcement, finds AI adoption concentrated in a small set of leading economies, with Nigeria, South Africa and Kenya ahead of the pack, and most African countries using AI below the levels seen in lower-middle-income economies. More than 2 billion people were offline in 2025, and internet use in low-income countries stood at 23 percent against 94 percent in high-income ones. Those are the conditions the shared-compute proposal is designed for: governments that need results before they can finance concrete. ## The Rent-versus-Build Question The initiative's premise deserves a fair hearing before a verdict. Building a national AI data centre means solving power first, then cooling, connectivity, financing and finally demand. Kenya's own pipeline shows how slow that chain runs: the Konza expansion is only now entering [PPP preparation funding](/articles/konza-ppp-data-centre-funding), years after entering the pipeline. For a ministry that wants crop disease detection for farmers or an inventory tool for clinics, waiting for a national facility to reach financial close is not a plan. Renting pooled capacity turns a capital problem into an operating expense and lets useful work start this fiscal year. ![Illustrative AI-generated image of African engineers and data scientists training an AI model in a compute lab](/images/worldbank-ai-training-illustrative.webp) But renting has a catch that the announcement leaves entirely open, and early reporting noticed. As written, the initiative does not say how governments will access the infrastructure, what it will cost, which countries participate first, or, most importantly for this audience, where the shared GPUs would physically be located. Every one of those unknowns sits upstream of the decision the Bank is asking governments to make. [The compute-gap story we covered in September](/articles/aws-1-5bn-africa-compute-gap-kenya) made a related point about private capital: announcements are easy, the terms sheet is where reality lives. ## What Kenya's Own Rules Already Decide Here is the part of the debate that happens in Nairobi rather than Seoul. Kenya's Data Protection Act and the public-sector hosting expectations built around it already require much government and personal data to remain in-country. That legal fact does most of the work in the rent-versus-build argument before economics enters at all. ![Corridor between server racks inside a modern data centre](/images/worldbank-dc-corridor.webp) A shared GPU pool whose location is undisclosed cannot host a workload the law says must live in Kenya. It cannot process citizen records for a ministry, back a government cloud, or touch the datasets that localisation rules protect. What it can serve is the wide upper layer of AI work that carries no residency constraint: model experimentation, non-sensitive inference, prototyping for agriculture and small business tools. Kenya's [data sovereignty push](/articles/kenya-data-sovereignty-localisation) is not a mood or a preference; it is a compliance perimeter. Any government computing plan splits cleanly along that line, and the Bank's initiative, as announced, only serves one side of it. That is also why the rent-versus-build framing is a false binary for Kenya. The country is not choosing between a GPU rental account and a national data centre. It is building the local layer because law and anchor demand force it, the [compute hub plan](/articles/kenya-data-centre-compute-hub-plan-explained) says as much, and commercial capacity is arriving regardless: Cassava, one of the initiative's own knowledge partners, has said its NVIDIA-powered AI factory is deploying in South Africa with expansion planned for Nigeria, Kenya, Egypt and Morocco. A government that rents pooled compute for bursty small-AI work while local capacity matures is running a sensible on-ramp, not abandoning the build. ## The Wrinkle in the Partner List Cassava and NVIDIA's role deserves scrutiny, because the Bank has handed its credibility to two parties with a commercial roadmap of their own. Cassava is not a neutral consultancy; it is an operator building paid AI capacity across the continent, with Kenya on its expansion list. If the initiative's shared pools end up riding on commercial infrastructure like Cassava's AI factories, the programme becomes a demand-aggregation vehicle that channels government spending into private facilities. That would be defensible, arguably elegant, but it should be said plainly. It would also mean the GPU location question answers itself: the capacity would sit where the operators build, including, eventually, in the region. ![Close-up of NVIDIA GPU chips mounted on a circuit board](/images/worldbank-nvidia-gpu-board.webp) The optimistic read and the cynical read converge on the same watch item. If the shared pools sit in-region, the initiative seeds utilisation for exactly the local facilities DC254 tracks and lowers the unit cost of government AI. If they sit outside Africa, the initiative becomes an off-ramp that quietly ships commodity AI workloads abroad while the hard local problems of power and financing remain unsolved. The announcement, as it stands, supports either interpretation, which is precisely why the GPU location and data residency terms are the details to watch. ## What Shared Compute Cannot Do Strip the debate to workload shapes and the limits become obvious. Small AI, the Bank's own target, means intermittent, modest inference jobs: identifying a crop disease from a photo, drafting a clinic summary, forecasting stock for a kiosk. Those jobs fit rented pools well. The workloads that anchor a national digital economy do not: identity platforms, tax systems, health records, government cloud, anything under a residency rule, and any training run large enough to need sustained power contracts. Those need facilities with metered power, certified operations and contracts, the things [data centre operators](/directory) actually sell. An economy that rents all of its compute has outsourced its leverage over pricing, uptime and jurisdiction, and no development lender has answered for that yet. There is also the constraint under every constraint. Power, not GPUs, is the binding input for local capacity in most of the region, which is why the Bank's own energy programming runs in parallel. Compute rentals do not build generation or transmission; they simply make demand productive wherever electricity already exists. ## What to Watch Five markers will settle whether this initiative helps Kenya's infrastructure or routes around it: 1. **GPU location and data residency terms.** The single most consequential undisclosed detail. In-region capacity makes the initiative a complement; out-of-region capacity makes it a workaround. 2. **The access mechanism and price list.** Pay-as-you-go against whose invoice, settled in which currency, with what commitment? Ministries will want comparisons before signing anything. 3. **The first cohort.** Which governments enrol, and whether Kenya, already ahead on adoption per the Bank's own update, is among them. 4. **The knowledge partners' footprint.** Whether Cassava's Kenya expansion and the Bank programme develop into one pipeline, with the initiative steering demand toward commercially built regional capacity. 5. **Uptake against the local plan.** Whether ministries rent first and build later, or rent instead of building. The first pattern accelerates Kenya's compute hub ambitions; the second starves them. ## The DC254 View Our read, stated plainly: for Kenya, the initiative is best treated as an on-ramp and a demand signal, not an alternative. Localisation rules decide the base case; the market, power first, decides the build. A government that rents burst capacity while Konza's PPP, the Cassava expansion and the compute hub plan mature loses nothing and buys time. A government that reads "alternative to standalone national data centres" as policy direction would be trading a solvable infrastructure problem for a dependency nobody has priced. The Bank has offered a bridge. Kenya should walk it without mistaking it for the destination. ## Frequently Asked Questions ### What is the AI Infrastructure and Readiness Initiative? A World Bank Group effort announced in Seoul on 20 October 2026, offering governments in Africa and Latin America shared, pay-as-you-go access to GPU computing as a practical alternative to building standalone national data centres. Cassava Technologies and NVIDIA are named as knowledge partners. It sits inside the Bank's first digital strategy. ### Does shared compute mean countries no longer need their own data centres? No. The initiative targets small, bursty AI workloads like crop disease detection or inventory tools. Sustained, compliance-bound workloads such as government cloud, identity systems and health records still need in-country capacity, especially where data protection law requires local residency. Shared compute is an on-ramp, not a substitute for the infrastructure behind sovereignty rules. ### Where would the shared GPUs be located? The announcement does not say. Neither the press release nor early coverage discloses where the pooled capacity would physically sit, how governments would access it, or what it would cost. GPU location is the decisive detail for countries with data residency rules, because compute outside the country cannot host locally-resident data. ### How does this interact with Kenya's data localisation rules? Kenya's Data Protection Act and public-sector hosting expectations require much government and personal data to remain in-country. Workloads covered by those rules cannot simply move to a shared pool abroad. Rented compute can serve compliant inference tasks, but the base layer of government platforms stays a local build case. ### How is this different from the AWS $1.5 billion Africa commitment? AWS's commitment is commercial: private capital building cloud regions and infrastructure for profit across African markets. The World Bank initiative is a development-financing programme lending governments a shared public-interest route into AI capacity. One adds commercial supply; the other tries to organise demand that markets have not reached yet.