The Real Price of an African AI Data Centre Is Electricity

Everyone talks about Africa's AI race in terms of chips, models and talent. The evidence says the real constraint is something far more basic, whether the lights stay on.

In 2026, Kenya's President William Ruto made this collision explicit and public. Microsoft and G42's proposed 1 billion dollar data centre near Naivasha, geothermal-powered and initially sized at 100 megawatts but expandable to a full gigawatt, ran straight into the limits of Kenya's national grid. Ruto warned that the project at full scale would require enough electricity that the country would effectively have to switch off power for half the nation to run it. This in a country with one of the cleanest, most dispatchable grids on the continent, roughly 90 percent renewable, led by geothermal. Even Kenya's genuinely well-run power system is not large enough to absorb a single hyperscale AI campus without political controversy.

The scale of the underlying gap is interesting. Africa holds close to 20 percent of the world's population and roughly 0.6 percent of global data-centre capacity. The International Energy Agency puts the continent's data-centre electricity consumption at under 1 kilowatt-hour per capita, the lowest of any world region, rising only to about 2 kilowatt-hours by 2030 on current trends. McKinsey estimates African data-centre power demand will grow from roughly 0.4 gigawatts today to somewhere between 1.5 and 2.2 gigawatts by 2030, requiring 10 to 20 billion dollars in construction investment before a single additional server rack is switched on.

Two genuinely different national strategies are emerging in response, and both are visible right now. South Africa, with 55 gigawatts of installed capacity, the largest single grid on the continent, is leaning on renewables and grid diversification. Teraco's 120 megawatt solar wheeling arrangement, moving power from the Free State to Johannesburg and Cape Town data centres via the Eskom grid, is the first of its kind on the continent and a genuine template. Microsoft has committed 5.4 billion rand to South African infrastructure by 2027; AWS has pledged 30.4 billion rand over several years. Fifty-eight commercial data centres already operate there, more than double Nigeria's count.

Nigeria is betting on gas instead. With Africa's largest proven natural gas reserves, over 200 trillion cubic feet, and roughly 61 percent electricity access nationally, Nigerian developers are building dedicated, oversized gas plants specifically to guarantee uptime for individual data-centre clients. One Ogun State facility pairs a 20 megawatt data centre with a 100 megawatt gas plant, five times the load it actually serves. As NJ Ayuk of the African Energy Chamber puts it, no one questions Microsoft's balance sheet, and that changes the financing equation for gas projects that were unbankable on domestic demand alone.

The country with arguably the weakest position in this race is not the one making headlines. Zambia's electricity access rate sits at roughly 31 percent, as low as single digits in rural areas, despite hosting three data centres and sitting at the centre of a major mineral-corridor rivalry between Western and Chinese-backed rail investment. Access this thin is a hard ceiling on AI ambition, whatever else a country has going for it.

What all three cases share is the same underlying lesson: power availability, not compute or connectivity, is what actually determines where global AI infrastructure investment lands on this continent. Site-selection decisions worth billions of dollars increasingly turn on a single question that has nothing to do with software at all.

Africa INSIGHTS Pro subscribers get the full Energy & Power Market Brief, comparing twelve African countries on generation capacity, grid reliability, and AI-hosting readiness, with the complete data behind the Kenya, South Africa and Nigeria cases, eighteen sourced statistics, and the companies and government officials worth watching as this plays out over the next eighteen months.

Upgrade to Pro to unlock the full Energy Behind Africa's AI Economy brief and our complete Market Brief library.

Evaluating a specific site, market or power-purchase structure for an infrastructure investment? Our team builds custom research and advisory projects around exactly these decisions. Get in touch.