Document AI Infrastructure Readiness — Kenya
Status Public position — open for engagement
Prepared by Nancy W. Mukui, NEMA Lic. 6945
Classification Policy & engineering advisory

Does Kenya's AI policy account for the physical infrastructure AI actually requires?

A civil, structural and environmental engineer's reading of the draft Kenya AI & Emerging Technologies Policy — and what it still leaves ungoverned.

Direct answer

Not yet. Kenya's draft AI policy treats AI mainly as a digital system. At scale, it is also a physical one — large compute needs electricity, water, land, cooling and environmental approval, the same disciplines already applied to roads and dams. Without an explicit readiness framework, investment risks landing on unprepared grids, water-stressed sites and unassessed land.

Why this matters now

Kenya's ambition to become a leading African AI hub is achievable, and the draft policy is a real governance foundation. The Kenya AI Strategy 2025–2030 already names data centres, cloud infrastructure and connectivity as enabling foundations, and national energy plans prioritise expanding renewable generation. What's missing is the bridge between those two facts: a framework that coordinates energy, water, land, environment and climate sectors before investment commitments and site decisions are locked in.

This isn't an argument against data centres. It's an argument for attracting the right investment to the right locations, under conditions that protect Kenya's energy security, water resources, host communities and long-term competitiveness — the same logic already applied to every major road, dam and transmission project in the country.

AI's physical footprint, dimension by dimension

Table 1 — Where the draft policy is currently silent
DimensionWhy it mattersWhat's missing today
Electricity & gridAI compute creates large, continuous, sometimes variable loadsSite-specific transmission and distribution capacity isn't checked before commitments are made
Water & coolingCooling technology and local climate set direct water demandNo water-balance or drought-resilience requirement for water-stressed sites
Land & servicesFacilities need secure sites, roads, drainage, security, logisticsPoor siting shifts infrastructure cost onto counties and communities
Environmental & socialNoise, emissions, hazardous materials, e-waste, occupational healthAssessment isn't yet proportionate to AI-scale operations
Cumulative impactClustered facilities strain shared grids and catchmentsProjects assessed individually; cluster effects go unmeasured
Climate riskHeat, drought and flooding affect reliability and demandClimate screening isn't built into siting and design
Social licencePublic trust depends on transparency and fair benefit-sharingNo disclosure or benefit-sharing requirement yet exists

Eight provisions the final policy should add

Table 2 — Priority recommendations
#ProvisionWhat it requires
1Recognise AI infrastructure as national infrastructureDefine large-scale compute, data centres and AI-support facilities as strategic infrastructure needing cross-sector planning
2Create an AI Infrastructure Readiness FrameworkAssess power, water, cooling, connectivity, land, climate risk and community impact before major approvals
3Require early environmental & social screeningMove assessment upstream, before land acquisition, technology lock-in and power or water commitments
4Introduce resource-use disclosureSignificant facilities report projected electricity, water, cooling, backup generation, emissions and e-waste
5Address cumulative & regional impactsAssess clusters of facilities, not only individual projects, on grids, catchments and host communities
6Establish green, climate-resilient standardsRenewable baseload, efficiency, water-sensitive cooling, heat recovery, circularity, resilient design
7Build Kenyan professional capacityMultidisciplinary capability across power systems, civil and environmental engineering, EIA, and safeguards
8Create transparent monitoringPublished indicators, thresholds, responsible institutions and periodic national readiness reports

How it could be phased in

Table 3 — Proposed implementation matrix
ActionLeadTimingOutcome indicator
Interim screening guidanceICT Ministry, NEMA, Energy, Water, Lands, counties, KEBS0–12 monthsGuidance issued and applied to major proposals
National Readiness FrameworkICT Ministry with cross-sector taskforce0–18 monthsFramework approved; indicators published
Disclosure & assessment thresholdsNEMA, EPRA, WRA, KEBS, counties0–18 monthsThresholds gazetted or regulated
National project registerICT Ministry, NEMA, regulators12–24 monthsPublic register operational
Capacity programmeUniversities, professional bodies, industry, government12–36 monthsTrained public and private practitioners
Biennial readiness reportICT Ministry and taskforceEvery 2 yearsReport published with power, water, environment and investment indicators

Suggested policy language

Two clauses that could be inserted directly into the digital infrastructure or responsible-deployment pillar:

The Government shall establish and periodically update a National AI Infrastructure Readiness Framework to guide the planning, siting, approval, development and monitoring of large-scale compute, data-centre, cloud and other AI-enabling infrastructure — integrating electricity and grid capacity, renewable energy, water availability, cooling technology, connectivity, land-use planning, climate risk, environmental and social assessment, cumulative impacts, circularity, public participation, local professional capacity and community benefit.
Proposed core clause
Within twelve months of adoption, the responsible Ministry shall convene relevant national and county institutions to issue interim screening guidance and define thresholds for strategic assessment, project-level EIA/ESIA, resource-use disclosure and public reporting for AI-enabling infrastructure.
Proposed implementation clause

Related questions

Does a data centre in Kenya need an Environmental Impact Assessment?

Yes. Under the Environmental Management and Coordination Act and NEMA's EIA regulations, qualifying projects require an EIA before commencement. Read the full approvals brief →

Who regulates AI infrastructure siting in Kenya?

No single body — the ICT Ministry, NEMA, county governments, EPRA and the Water Resources Authority each hold part of the mandate, with no framework coordinating them yet. Read the full regulatory-map brief →

How much water and power does an AI data centre actually use?

It depends on scale, cooling technology and local climate, so figures vary widely by design. What matters for policy is a disclosure requirement, not a single number. Read the full disclosure brief →

Does Kenya require climate-resilient data centre design?

Not by name yet, but the Climate Change Act and National Adaptation Plan already apply to infrastructure generally — the link to data centres just hasn't been made explicit. Read the full climate-design brief →

About the author

Nancy W. Mukui is a civil, structural, environmental and public-health engineer, NEMA Lead Expert, construction practitioner and AI strategist with more than twenty years of infrastructure experience across East and Central Africa, delivered under World Bank, AfDB and JICA frameworks. Her current work sits at the intersection of artificial intelligence, physical infrastructure, environmental assessment and sustainable development.

NEMA
Lead Expert, Reg. 6945
EBK
Registered Engineer, Reg. B3429
IEK
Graduate Member, Reg. G2943
Education
M.Sc. Environmental & Public Health Engineering, University of Nairobi

Full profile & credentials →