When Strive Masiyiwa walked journalists through a chilled data hall on the outskirts of Johannesburg this year, the story was not the blinking lights. It was the promise behind them. His company Cassava Technologies had just announced a partnership with Nvidia to build what they call the first AI factory in Africa, a facility packed with thousands of advanced graphics processors designed to turn local data into intelligence at scale. For a continent that has mostly consumed artificial intelligence services built elsewhere, the move raises a sharp question. Is this simply a prestige project, or the foundation of a different digital power balance for African economies.
To understand the stakes, it helps to be precise about what an AI factory is. Nvidia chief executive Jensen Huang uses the term for a new kind of data centre that does not just store or transmit information. It is tuned end to end for AI workloads, from high performance chips and high speed networking to specialised software that trains and serves models. In his words, every company will eventually operate a physical factory that makes its products and an AI factory that produces the digital intelligence behind those products.
In practice, that means a facility where raw data enters and trained models, predictions and automated decisions come out, much as raw material enters a traditional plant and finished goods leave.
Masiyiwa’s play is to build that kind of capability inside Africa rather than at the edge of the continent in far away cloud regions. In March 2025 Cassava announced that it would upgrade its data centres with Nvidia AI supercomputers to create a powerful and secure facility in South Africa, with further sites planned in Nigeria, Kenya, Egypt and Morocco.
Public reports suggest a planned network of five AI factories with a total investment of around seven hundred and twenty million dollars and an initial installation of about three thousand Nvidia processors in Johannesburg, most of which are already reserved by developers and researchers.
The first force behind this project is the African compute gap. Most serious AI training today relies on large clusters of specialised processors located in North America, Europe or parts of Asia. African researchers and start ups have had to rent capacity in these distant regions, which drives up cost, increases latency and raises complex questions about data protection. Several studies and policy papers have warned that without local high performance
computing, African innovators will remain tenants in someone else’s digital real estate.
Cassava’s AI factory does not solve this gap by itself, but it begins to close it. By putting modern Nvidia infrastructure on the continent and selling access as a service, the company changes what is technically and financially possible for local teams that want to train language models, computer vision systems or recommendation engines on African data.
The second force is data sovereignty. African hospitals, banks, retailers and governments sit on sensitive information that cannot simply be shipped abroad for model training without regulatory and ethical concerns. The Cassava announcement emphasises that the AI factory will allow customers to keep data within African borders while still benefiting from cutting edge AI capacity.
A later partnership with the Rockefeller Foundation explicitly links this to civil society and social impact, with a plan to give non profit groups, universities and public sector bodies access to the same infrastructure for work in health, education and climate resilience. If the model works, African institutions gain a place to store and process their data under local legal regimes while still engaging global technical partners.
The third force is industrial strategy. AI factories are not just technical assets. They are geopolitical infrastructure, comparable to ports, power plants or cable landing stations. Nvidia is backing similar projects in Asia and other regions, often in partnership with manufacturers or cloud providers that want to anchor whole ecosystems around their facilities. (NEXTDC) In Africa, Cassava is positioning itself as the neutral digital backbone that ties together fibre networks, data centres and AI compute. That puts the company in direct conversation with global cloud giants and development finance institutions that are racing to fund digital public infrastructure across the continent.
What does all this mean for founders and executives in practical terms. It changes the constraints at the base of the stack. A health start up that wants to train a model to read chest X rays from Kenyan hospitals no longer needs to move those scans to a server farm in Europe, which simplifies compliance and may unlock new research partnerships. An agritech company can experiment with crop disease models that use local languages and agronomic data without worrying that every training run will burn through a budget priced in foreign currency. A bank can explore fraud detection models that rely on regional patterns while keeping customer data inside the continent. None of this is automatic. It depends on pricing, reliability and how effectively Cassava and its partners work with local developers. But the menu of realistic options becomes broader.
There are serious tensions to manage. Massive AI facilities consume power and water, and South Africa already faces grid instability. If the AI factory draws on fossil heavy energy, it risks deepening climate vulnerability in the name of digital progress. If capacity is locked up by large foreign platforms or a handful of corporate clients, the promise of democratised access will ring hollow. There is also the strategic risk that African firms become dependent on a single foreign supplier for core compute, in this case Nvidia, even as they try to escape dependence on distant cloud regions.
For African and diaspora decision makers, the smart response is neither uncritical celebration nor cynical dismissal. It is to treat the AI factory as a new piece of infrastructure and to plan accordingly. That means asking hard questions about governance, energy sources, access rules and pricing. It means building consortia of universities, start ups and corporates that can negotiate as serious customers rather than as scattered small buyers. Most of all, it means thinking clearly about what distinctive African problems and datasets can be turned into defensible products and services once high quality compute is closer to home.
The partnership between Strive Masiyiwa and Nvidia will not by itself decide whether Africa is a producer or a consumer in the age of artificial intelligence. It does, however, move the argument from aspiration to hardware and contracts. The leaders who read that signal early and align their strategy to the new reality will have a better chance of shaping the story rather than watching it unfold from the sidelines.
