Power and capital meet a global AI slowdown
Africa holds a small but fast-growing share of global compute. Africa Data Centres Association figures put active capacity at roughly 360MW — about 0.6% of global supply — with 238MW under construction and 656MW in the pipeline. McKinsey estimates that African data-centre power demand must rise from 0.4GW today to between 1.5GW and 2.2GW by 2030. That shift requires US$10bn–US$20bn in new facilities and supports a US$20bn–US$30bn revenue pool across the wider value chain.
Power is now the defining constraint. The Africa Data Centres Association identifies electricity availability and pricing as the principal bottleneck for new facilities. This mirrors a global pattern. International Energy Agency projections show data centres could consume around 950TWh of electricity by 2030, with power use at AI-focused sites expected to triple between 2025 and 2030. As regulators in the United States, Europe and parts of Asia weigh curbs on new AI campuses, those decisions will redirect capital and hardware — and Africa stands to benefit.
Recent deals signal that investors already see the opportunity. Amaco Energy Group has proposed a US$1.5bn off-grid AI data centre in Mombasa, backed by GE Vernova gas turbines. MTN Group has agreed a partnership with Dubai-based investor Tarek Al Ashram to develop roughly 150MW of AI-ready capacity through Africa Data Hub Holding, at a first-phase cost of US$3bn–US$6bn. Mastercard estimates Africa’s AI market could grow from US$4.5bn in 2025 to US$16.5bn by 2030 — which explains why operators are moving early to secure power and land.
As one Nairobi-based infrastructure analyst put it, Africa’s data-centre strategy is now a bet that power-led site selection can turn a global AI slowdown into a continental build-out of reliable electricity and compute capacity.
Skills, governance and the risk of a lost decade
The infrastructure push runs ahead of talent and policy. Africa holds roughly 18% of the world’s population yet less than 1% of global data-centre capacity. An IMF paper on sub-Saharan Africa finds that only about 5% of the region’s AI talent has reliable access to high-performance computing. Weak electricity networks, scarce digital infrastructure and limited technical skills compound that gap.
On governance, roughly 15 African states now have formal AI frameworks and more than 20 are drafting them. However, most score highest on vision and lowest on implementation. Budgets are rarely ring-fenced and oversight bodies remain nascent.
Wider infrastructure data stays mixed. The World Bank‘s World Development Report 2026 notes that three-quarters of firms in sub-Saharan Africa report power outages. The IMF’s AI Preparedness Index gives the region an overall score of 0.34, against 0.52 for Asia-Pacific and 0.63 for Europe, with digital infrastructure the weakest pillar. Those figures cut two ways. A slower global AI build cycle could give governments room to prioritise grid upgrades, fibre and education. By contrast, tighter regulation in advanced markets could redirect capital away from power-constrained regions entirely.
Cloud providers are working to close the skills gap. AWS reports investing more than US$819m in African infrastructure since 2018 and training over one million Africans in cloud and AI skills. Google has announced an applied AI lab in Ghana and a digital innovation centre in Soweto. These moves begin to link physical build-out with skills pipelines and local applications.
What does the AI slowdown mean for investors?
For long-term investors, the global AI slowdown debate is less about whether compute growth moderates and more about who uses the breathing space. If regulators in advanced economies cap high-density AI clusters, the relative appeal of African locations with scalable, low-carbon power could rise sharply. Knight Frank research suggests African data-centre demand could expand three to five times by 2030, implying US$10bn–US$20bn of new build. South Africa, Nigeria, Kenya, Egypt and Morocco anchor most announced projects.
The investment case rests on three questions. First, can sponsors secure reliable electricity through grid reform or private generation fast enough to meet demand? Second, will governments move from broad AI strategies to funded, enforceable governance regimes? Third, can training programmes lift Africa’s share of global AI talent so that new facilities host high-value workloads rather than mainly imported applications?
Investors should watch how quickly power-purchase agreements, permitting regimes and AI-governance frameworks evolve in core markets over the next three to five years — those signals will determine whether Africa’s data-centre moment becomes a structural boom or stalls at the grid.
Quick answers
The Africa Data Centres Association puts active capacity at roughly 360MW — about 0.6% of global supply — with a further 238MW under construction and 656MW in the pipeline.
Key investors include Amaco Energy Group, which has proposed a US$1.5bn facility in Mombasa, and MTN Group, which is developing up to 150MW of AI-ready capacity through Africa Data Hub Holding at a cost of US$3bn–US$6bn for the first phase.
Electricity availability and pricing are the principal bottleneck, according to the Africa Data Centres Association, compounded by limited AI talent — the IMF finds only about 5% of the region’s AI researchers have reliable access to high-performance computing.
