The Forecasting MachineRead report

Loading shared forecast

The Forecasting MachineOpen interactive report

By December 31, 2030, which African country will have the largest operational AI-focused data-center capacity?

Africa is undergoing rapid investment in digital infrastructure, cloud computing, and AI/HPC workloads. While South Africa historically holds the largest operational share, countries like Egypt, Nigeria, Kenya, and Morocco are expanding local capacity quickly. This question tracks which African country will lead in operational AI-focused data-center capacity by the end of 2030. Qualifying Capacity: - Must be operational and commercially or institutionally usable by December 31, 2030. - Announce, planned, financed, or under-construction projects are excluded. - Must be documented for generative AI, GPU/accelerator clusters, accelerated computing, HPC, or large-scale machine learning. - Ordinary cloud/colocation capacity does not count unless explicitly designated for AI/HPC. - Capacity is attributed to the country where the physical infrastructure is located. Tie-breaking rules: 1. Documented installed AI accelerator/GPU computing capacity. 2. Total operational high-density data-center capacity capable of supporting AI/HPC. 3. Forecasting platform standard tie-breaking procedure.

Forecasted on

Forecast probabilities

  • South Africa: 70.5%
  • Egypt: 12.5%
  • Nigeria: 6.5%
  • Kenya: 2%
  • Morocco: 6.8%
  • Another African country: 1.8%

Reasoning

South Africa is favored to retain the lead, with a 70.5% probability; its installed infrastructure and near-term AI hardware give it a substantial edge over projects that still need to become operational. Egypt is the clearest challenger at 12.5%, while Nigeria (6.5%), Morocco (6.8%), Kenya (2.0%), and another African country (1.8%) each have a less likely path.

South Africa — incumbency with real hardware behind it (70.5%)

South Africa’s advantage is not just a long list of announced projects: it has a deep data-center ecosystem, established hyperscaler presence, and AI compute that is operational or close to it. The country accounts for more than 60% of the continent’s operational data-center capacity, and its largest reported facility, Teraco Isando, has 70 MW [e6]. Cassava’s Nvidia GPU deployment and Stratos Lab’s planned B300 cluster provide a more direct basis for counting AI-focused capacity than general-purpose colocation alone [a1, a12, a13]. Google has also described AI-infrastructure expansion around its Johannesburg cloud region, while Continent 8 has launched a Johannesburg facility [e12, e20].

The lead is not immune to friction. South African organizations have called for a pause on new projects over electricity, water, land-use, and environmental concerns, and an appeal suspended approval of Equinix’s proposed Cape Town hyperscale project [e22, e25, e26]. That makes it unwise to count its prospective capacity as assured. But the country’s established base and multiple routes to further investment mean one delayed project need not derail the overall lead. The government’s stated interest in developing the country as a digital-infrastructure hub adds to that supportive backdrop [e1, e21].

Egypt — the most credible project-driven challenge (12.5%)

Egypt has a plausible route to the top if its sovereign-backed AI project scales on schedule. The headline ambition is 200 MW, but the initial phase is much smaller—20 MW—so the key question is not the announced endpoint but how much qualifying capacity is usable by 2030 [a6, a15, a17]. Egypt also has an existing data-center base: the East Cairo campus is reported at 25 MW, though that figure alone does not establish that the facility qualifies as AI-focused [e6]. A national push to license data centers and a separate proposed facility signal policy momentum, but do not yet settle the delivery question [a5, a14].

Egypt’s probability is therefore meaningful, but below South Africa’s. A large project can change the ranking; it still has to turn from a target into operating AI capacity while South Africa continues to add infrastructure.

Nigeria — strong demand, difficult scaling conditions (6.5%)

Nigeria has genuine momentum. It has a sizeable operating data-center sector, and new data-localization rules are likely to increase demand for local infrastructure [e11, e17]. MTN and an investment partner have also said they will focus initial development on South Africa and Nigeria, while Nigeria’s government is working on policies and investment to support AI development [e2, e3, e18]. Nigeria and South Africa have both crossed the reported $5 billion milestone in specialized AI and data-center investment [e8].

But demand and investment are not the same as qualifying capacity. Nigeria’s largest reported operational site is 36 MW, and the national grid’s limited scale makes rapid expansion of power-intensive AI facilities challenging [e6, a19]. The country has a credible path to growth, but a less certain path to surpassing South Africa or Egypt by the deadline.

Morocco — a large target with limited confirmation (6.8%)

Morocco’s announced Casablanca AI project, with a stated target of scaling from 40 MW to 500 MW, is large enough to make the country a serious wildcard [a19]. A new agreement with Vertiv covers infrastructure for AI and HPC, including power and cooling, and Oracle has established a second research and development hub in the country [e16, e38]. These developments point to interest and ecosystem-building, but do not establish that the proposed capacity will be financed, built, and operating by the end of 2030.

That gap matters. A 500 MW target could reorder the field if delivered, but the evidence for its execution is thinner than for South Africa’s existing infrastructure or Egypt’s sovereign-backed project. The forecast gives Morocco a notable chance without treating an ambitious announcement as completed capacity.

Kenya — demand is not enough to overcome the infrastructure hurdle (2.0%)

Kenya has strong AI adoption and growing policy activity, but neither automatically produces local AI compute capacity [e31, e32]. The clearest caution is the stalled Microsoft/G42 data-center project, which was shelved after concerns that it would require roughly a third of the country’s national grid capacity [a8]. Kenya could still develop smaller or differently powered projects, but overtaking the leading countries would require a substantial conversion of plans into operational, qualifying infrastructure.

Another African country — a small residual possibility (1.8%)

A country outside the named contenders could benefit from a project that is currently less visible or from faster-than-expected delivery. But Africa’s market remains concentrated, and the available evidence points to the named countries—especially South Africa, Egypt, and Nigeria—as the more developed centers of investment and infrastructure [e6, e8]. This outcome remains possible, but there is little specific evidence for another country overtaking them by 2030.

Why the probabilities favor South Africa

The decisive distinction is between operating capacity and announced ambition. South Africa has a large general data-center base, identifiable AI hardware deployments, hyperscaler investment, and several paths to expansion; its approvals and power constraints keep the lead from being certain. Egypt’s sovereign project is the strongest threat, while Nigeria and Morocco have credible but less proven routes to scale. Kenya and other countries face an even steeper conversion challenge. That balance supports the forecast probabilities: South Africa 70.5%, Egypt 12.5%, Nigeria 6.5%, Kenya 2.0%, Morocco 6.8%, and another African country 1.8%.

Cited evidence