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, general manager for sub-Saharan Africa at AWS, told the AWS Summit in Johannesburg that the Cape Town cloud region now carries 154 services, and framed the company’s strategy around enabling local developers to build for regional markets rather than consuming technology developed elsewhere.
The commitment is broadly comparable to Microsoft’s African position. 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.
The leapfrog argument, and its gap
Ball’s central case is that the current AI cycle offers Africa another chance to bypass development stages followed by wealthier economies.
Her precedent is mobile money. “Africa did not wait for conventional banking infrastructure to become widespread before developing large-scale mobile payment systems,” she said, pointing also to South Africa’s move from limited mobile connectivity in the early 1990s to more than 100 million connections.
The analogy is appealing and incomplete. 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. Africa holds roughly 1% of the world’s data centres against about 15% of global population. McKinsey projects capacity must rise from around 0.4 gigawatts to as much as 2.2 gigawatts by 2030, requiring $10 billion to $20 billion.
That is the argument Dr Francois du Plessis put in these pages last week from the opposite direction — that South Africa should use AI extensively while relying on established international data centres, rather than competing to host facilities its water and power systems cannot comfortably support. Ball’s leapfrog framing and Du Plessis’s position both accept that compute can sit elsewhere. They differ on whether that is a problem to solve or a constraint to work within.
Where the examples hold up
Ball cited three companies. Two were not named.
A local fintech serving the informal retail economy has used a generative AI know-your-customer compliance engine to make previously uneconomic customer segments viable. An East African insurer moved a product from concept to production in 20 days using an AI agent. Neither company was identified, which limits what readers can assess.
The third was Aerobotics, and it is the strongest case. The Cape Town agritech uses Amazon SageMaker to combine drone, satellite and mobile imagery, monitoring the health of 65 million trees, and has expanded into 18 countries including the United States.
That is an African company using AI to export a product into developed markets — the outcome every African AI strategy names as the objective, and one of relatively few concrete instances of it.
The market case
Ball cited Mastercard’s projection that Africa’s AI market will reach $16.5 billion by 2030, growing at 27%, and the IMF’s assessment that AI could add four percentage points to African GDP if the right foundations are established.
The IMF’s own framing carries a qualifier Ball did not mention: under current baseline conditions, without infrastructure and skills investment, the productivity gain shrinks to roughly 0.2%. The four points are conditional, not projected.
The line everyone is now using
AWS is also expanding skills training, with the stated aim of increasing African developers’ capacity to create and control AI products.
“When I look forward, what excites me most is Africans on the continent not just consuming AI, but building AI and owning AI,” Ball said.
That formulation has become standard. Microsoft’s incoming East Africa lead Angela Nganga said this month that enabling East African companies “to become producers as well as consumers of AI innovation is essential.” Nigeria’s Bosun Tijani, Lagos governor Babajide Sanwo-Olu, Zambia’s Percy Chinyama and South Africa’s Deputy Minister Mmapaseka Letsike have each made versions of the same argument this year.
The tension is unresolved rather than hypocritical. Nigeria, Kenya, Egypt and South Africa have each identified dependence on US technology companies as a strategic risk in their draft AI strategies, while relying on those companies for the compute their AI ambitions require. Hyperscalers advocating African AI ownership are advocating something that currently runs on their infrastructure.
Ball said AWS intends to keep expanding capacity and bringing services closer to developers and customers. “It’s our job at AWS to make sure nothing stands between you and what you’re trying to build.”
