We do not own the hardware, and owning it is the opportunity.

Africa holds roughly eighteen per cent of the world’s population and under one per cent of its data-centre capacity. The five largest African markets combined hold less capacity than France held in 2024. Said as a complaint that is a familiar line. Said as a position it is an underbuilt asset class in a market where the regulator has just guaranteed that demand has to be served onshore.

Capital has already noticed. Cassava has committed up to 720 million dollars to deploy twelve thousand NVIDIA processors across African markets. Nigeria’s next phase of data-centre development is costed at between 240 and 250 million dollars for AI-ready infrastructure alone. Microsoft and G42 put a reported one billion dollars into Kenya. None of that is ours, and most of it will be rented back to Nigerian institutions who will have no alternative after the first of January.

The binding constraint is not processors, it is power. African data-centre electricity demand is projected to rise from about 0.4 gigawatts to 2.2 gigawatts by 2030, on a continent where roughly half of sub-Saharan Africa still lacks reliable supply. Whoever pairs compute with power will own the decade. That is a capital allocation question rather than a technology one, and capital allocation is what this Group is actually built to do.

So the objective is ownership, and the route to it is staged rather than heroic. The fleet we already have is the beachhead: it earns immediately, it teaches us which workloads matter, and it is compliant on day one. Owned local nodes next, sized to what the Group actually runs rather than to a vendor’s reference architecture. Then a compute position held deliberately as an asset, operated by the AI and Intelligence Company the memorandum mandates, and sold to the institutions around us who are in precisely our position and have not started.