Microsoft and G42's $1 Billion Olkaria AI Data Centre Stalls Over Kenya's Power Grid Limits

A flagship $1 billion AI data centre project planned by Microsoft and Abu Dhabi-based G42 near Kenya's Olkaria geothermal field has run into delays, as the country's electricity transmission infrastructure struggles to keep pace with the project's substantial power requirements.

The Olkaria site was chosen specifically for its proximity to Kenya's geothermal power fields, which supply a significant share of the national grid and offered the developers a rare opportunity to pair large-scale AI compute with genuinely renewable baseload power — an increasingly important selling point as hyperscalers face growing scrutiny over the carbon footprint of AI data centres globally.

Grid capacity, not power generation, is the constraint

Industry sources familiar with the project say the core bottleneck is not Kenya's geothermal generation capacity itself, which remains among the most reliable renewable baseload sources on the continent, but the transmission and substation infrastructure needed to move that power reliably to a data centre campus at the scale AI training clusters require. Modern AI training facilities can draw tens to hundreds of megawatts continuously, a demand profile that differs substantially from the intermittent, distributed load patterns Kenya's grid was originally built to serve.

Kenya Power and the Kenya Electricity Transmission Company have both been named in regional reporting as working with the project partners on an upgraded transmission plan, though no public timeline for resolving the bottleneck has been confirmed. The delay illustrates a challenge increasingly facing African data centre and AI infrastructure projects more broadly: capital for construction and hardware has become comparatively easier to secure than the grid capacity needed to actually energise large facilities, particularly where existing transmission networks were not designed with gigawatt-scale digital infrastructure in mind.

For Kenya, the Olkaria project remains strategically significant regardless of the delay. It would represent one of the largest single foreign direct investments in the country's technology sector and a marquee validation of Kenya's standing as an East African AI hub, alongside Nairobi's growing GPU colocation capacity. Microsoft and G42 have both continued to describe the project as active, with the parties working through the power infrastructure sequencing rather than reconsidering the investment itself.

Source: TechRadar Pro.

The infrastructure angle matters because AI adoption is constrained by more than software. Reliable electricity, connectivity, storage, cooling, cybersecurity and affordable access all determine whether a model can be used outside a demonstration. New capacity can strengthen local sovereignty and reduce dependence on overseas systems, but only if it is commercially accessible to startups, universities, public agencies and smaller businesses as well as large enterprises.

The practical questions now are operational. When will the capacity be available, who will be able to use it and under what pricing and governance arrangements? Readers should also watch for evidence on energy resilience, data protection and local skills. Infrastructure projects often attract attention at the announcement stage; their long-term value is established through uptime, customers, local jobs and the quality of services they enable. Those measures will show whether this is a durable part of Africa's digital foundation.