Africa has nearly a fifth of the world's people and under 1% of its data-centre capacity, which means the infrastructure of artificial intelligence on the continent is largely owned by others.
On 26 June 2026 Al Jazeera asked the question directly: who will control Africa's AI infrastructure, and at what cost. The figures behind the question are stark. Africa accounts for roughly 18% of the world's population and less than 1% of global data-centre capacity, and the continent's five largest data-centre markets combined hold less capacity than France alone.
Compute is the ground on which artificial intelligence is built. A continent renting the ground from others does not yet own its own intelligence — and the contest now under way over who builds, powers and keeps Africa's data centres will shape the terms for a generation.
The capacity gap is the whole story
The headline numbers describe a structural dependency, not a temporary lag. When the data centres that train and serve artificial intelligence sit overwhelmingly outside the continent, African data travels abroad to be processed, African developers build on foreign infrastructure, and the economic value of African AI accrues to whoever owns the racks.
A response is taking shape. 49 countries and the African Union have endorsed the Africa Declaration on Artificial Intelligence, a continental attempt to set common terms rather than negotiate one country at a time. The instinct is right: a bloc has leverage that a single nation, bargaining alone against a trillion-dollar firm, does not.
💡The figure that frames the decade
Less than 1% of global data-centre capacity, for 18% of the world's people. The number is not a measure of African capability — it is a measure of where the world has chosen to build, and of who has been treated as a market for intelligence rather than a maker of it.
Kenya as the test case: a billion-dollar data centre and its power bill
The clearest live example is a proposed $1 billion data centre in Kenya backed by Microsoft and the Emirati firm G42. The project promises capacity the continent badly needs, and President William Ruto has welcomed the investment while naming its catch: infrastructure on that scale demands substantial additional power generation.
The energy point is not a detail. A data centre is, in the end, a machine for turning electricity into computation, and a continent where hundreds of millions still lack reliable power must decide how much new generation to devote to foreign-owned racks versus homes, clinics and schools. The compute question and the energy-justice question are the same question.
Alongside the commercial deals runs a diplomatic template. A recent Carnegie analysis proposes the US-Kenya "Technology Prosperity Deal" — a seven-pillar bilateral framework spanning infrastructure, governance, skills and financing — as a model for durable US-Africa AI partnerships. The analysis notes plainly that no African country has yet signed such a deal, even as the openings multiply.
A new scramble, and the room it leaves Africa
The reason the openings multiply is competition. The United States is courting African AI partnerships precisely because it is not the only suitor: China pledged some $50 billion, artificial intelligence included, at the 2024 Forum on China-Africa Cooperation, and Gulf capital is moving fast through firms like G42. The rivalry is real, and for once the fragmentation works in Africa's favour.
African states will indeed be provided with greater room for manoeuvre on AI and data infrastructure, precisely due to how contested and fragmented this industry is amongst global leaders.
— Priyal Singh, geopolitical analyst, Signal Risk
Singh's point is the hopeful one: a contested field is a negotiable field. With the United States, China and Gulf investors all competing to build, African governments can play terms against terms in a way they could not when there was a single buyer or a single technology to adopt. The leverage is genuine, if it is used deliberately and used together.
But leverage at the level of states is not the same as benefit for citizens, and that is the warning Afrobarometer's chief executive, Joseph Asunka, sounds. With a working-age population heading toward a billion people by 2050, the stakes of getting the deal structure wrong are generational.
These negotiations should not just be conducted at the elite level and dumped on citizens.
— Joseph Asunka, CEO, Afrobarometer
A dignity-first reading: from market to maker
Emergent Intelligence (EI) — the dignity-first lens through which I read artificial intelligence — begins from Ubuntu: a person is a person through other persons, and a people come into their own through what they build together. Applied to infrastructure, the principle is exact. A continent that only consumes intelligence built elsewhere is, in a precise sense, not yet fully an author of its own future.
I have argued that the deeper risk is not exclusion from AI but inclusion on bad terms — a digital Berlin Conference, in which the continent is partitioned anew, this time by platform and protocol. Data-centre dependency is the physical layer of the partition. Whoever owns the compute sets the defaults, captures the value, and holds the off-switch.
The dignity-first response is not to refuse foreign investment, which Africa needs, but to insist that investment build capacity that stays. The test for any deal — Microsoft's, G42's, a US Technology Prosperity Deal, a Chinese facility — is whether it leaves behind owned infrastructure, trained people and local power, or whether it merely rents the continent its own future back at a margin.
💡Why compute is a dignity question
Sovereignty over compute is not nationalism; it is the material form of self-determination. A people that cannot run its own models cannot fully govern its own data, its own languages, or its own account of itself. That is why the data-centre map is a dignity question, not merely an economic one.
What a good deal would actually require
Turning leverage into ownership needs more than goodwill. It needs terms written down and negotiated as a bloc: local ownership stakes that grow over time, mandatory skills transfer, data-residency guarantees, and dedicated power generation so that compute does not come at the expense of households. The Africa Declaration on AI is the right vehicle; the work is to give it teeth.
Asunka's warning is the procedural heart of it. A deal negotiated only among presidents and chief executives, then handed down to citizens, repeats the old colonial form even when the technology is new. Legitimacy requires that the people whose data, land and electricity are committed have a real say in the committing.
Owning the ground, not renting it
Africa will not close a 99-to-1 capacity gap in a single investment cycle, and the choice is not between total self-sufficiency and total dependence. The choice is about direction: whether each new data centre moves the continent toward owning its compute or deeper into renting it.
A dignity-first frame asks every party to the scramble a single question. Does this deal make Africa a maker of artificial intelligence, or keep it a market for someone else's? The honest answers will differ, and the continent's job — bargaining together, with citizens in the room — is to take the deals that build and decline the deals that merely extract.
Who will control Africa's AI infrastructure is still an open question on 26 June 2026, and that is the opportunity. The contest is real, the leverage is real, and the terms are not yet set. The task is to use the moment to build infrastructure that stays, power that serves people first, and an African intelligence that the continent owns — not one it rents back, at a margin, from the powers building the racks.
Frequently Asked Questions
The questions below address the most common queries about who controls Africa's AI infrastructure, drawn from the June 2026 reporting and analysis.
Why does Africa have so little AI data-centre capacity?
As of June 2026, Africa accounts for roughly 18% of the world's population but less than 1% of global data-centre capacity, and its five largest data-centre markets combined hold less capacity than France. The gap reflects decades of investment flowing elsewhere, alongside constraints such as limited reliable power, financing and skills — not a lack of African capability or demand.
What is the Africa Declaration on Artificial Intelligence?
The Africa Declaration on Artificial Intelligence is a continental framework endorsed by 49 countries and the African Union to set common terms for AI development and infrastructure. Its purpose is to let African states negotiate as a bloc — over data centres, data sovereignty and partnerships — rather than bargaining individually against far larger foreign firms and governments.
What is the Microsoft and G42 data centre in Kenya?
It is a proposed $1 billion data centre in Kenya backed by Microsoft and the Emirati firm G42, intended to expand the continent's AI compute capacity. President William Ruto has welcomed the investment but warned that infrastructure at that scale would require substantial additional power generation, raising the question of how new electricity is balanced between data centres and citizens' needs.
What is the US-Kenya Technology Prosperity Deal?
The US-Kenya Technology Prosperity Deal is a proposed seven-pillar bilateral framework — covering infrastructure, governance, skills, financing and more — analysed by the Carnegie Endowment in June 2026 as a possible template for durable US-Africa AI partnerships. No African country has yet signed such a deal, even as the United States, China and Gulf investors compete for African technology partnerships.
What is the Emergent Intelligence view of African AI sovereignty?
Emergent Intelligence (EI) is a dignity-first, Ubuntu-grounded reading of artificial intelligence. It holds that owning compute infrastructure is the material form of self-determination, and that foreign investment should be judged by whether it leaves behind owned infrastructure, trained people and local power — not whether it rents Africa its own future back at a margin. Citizens, not only elites, must have a real say in the terms.
Sources and Further Reading
Cover image: high-voltage power lines crossing an African landscape at golden hour — via Pexels.
