Strive Masiyiwa’s Cassava Technologies is putting high-performance AI computing infrastructure into its South African data centres, with Nigeria, Kenya, Egypt and Morocco next. The bigger question is whether local compute can give African companies more control over the economics of artificial intelligence.
When Cassava Technologies announced plans in March 2025 for what it called Africa’s first AI factory, much of the proposition was still on paper.
The company said it would install NVIDIA accelerated computing infrastructure in its South African data centres before extending the model to Nigeria, Kenya, Egypt and Morocco.
A year later, the project had moved into deployment.
Cassava had said NVIDIA has also added Cassava to its global network of AI cloud providers, giving Africa a locally based supplier of the specialised computing capacity required to train, fine-tune and run artificial intelligence models.
For Africa’s technology industry, the important part is not the “AI factory” label. It is where the computing power is situated, who controls access to it and what African businesses have to pay to use it.
“By using this secure, high-performance AI Factory, African businesses and governments can develop local solutions to local challenges,” the company said in a statement. “This infrastructure ensures AI deployment happens within a secure environment, aligned with both global and local data regulations.”
What Cassava is actually building
An AI factory is essentially a data centre built specifically for intensive artificial intelligence workloads.
Traditional data centres handle a broad range of computing and storage tasks. AI infrastructure is designed around large numbers of specialised processors, particularly graphics processing units, or GPUs, working together to train models and process large volumes of AI requests.
NVIDIA describes AI factories as infrastructure designed for training, fine-tuning and running AI models at scale. Cassava is using NVIDIA accelerated computing systems and software under NVIDIA’s Cloud Partner architecture.
African developers have long been able to access AI through international cloud providers. Cassava is betting that there is a market for providing more of that computing capacity from within Africa.
DON’T MISS THIS: Meet the Nollywood actors behind the high-grossing movies in 2024/2025
“Our AI Factory provides the infrastructure for this innovation to scale, empowering African businesses, startups, and researchers with access to cutting-edge AI infrastructure to turn their bold ideas into real-world breakthroughs — and now, they don’t have to look beyond Africa to get it,” Masiyiwa said.
The business is compute
Artificial intelligence may look like a software industry from the consumer side, but the economics begin with physical infrastructure.
Large AI models require expensive chips, electricity, cooling systems, data centres, fibre connections and the capital needed to keep all of those assets running.
Whoever controls that infrastructure controls the AI value chain.
NVIDIA remains central because its hardware and software underpin much of the global AI computing market. Cassava’s role is different. It is building the African infrastructure through which companies and institutions can access that computing capacity.
The arrangement makes Cassava an infrastructure operator rather than an African rival to NVIDIA.
Its advantage comes from assets it already owns or controls across telecommunications, fibre and data centres.
That existing network gives Cassava something many new AI infrastructure companies would have to build from scratch – physical locations, connectivity and enterprise relationships across several African markets.
“Collaborating with NVIDIA gives us the advanced computing capabilities needed to drive Africa’s AI innovation while strengthening the continent’s digital independence,” Masiyiwa added.
Why keeping African data in Africa matters
Cassava has made data sovereignty a central part of its pitch.
Running AI workloads locally can allow companies and governments to keep more sensitive information within national or regional jurisdictions rather than sending it to computing infrastructure elsewhere.
That can matter to governments, banks, telecommunications companies, healthcare organisations and other businesses handling regulated or sensitive information.
It does not automatically make an AI system sovereign.
The hardware remains supplied by NVIDIA. Many underlying AI models may still come from American, Chinese or European companies. Software dependencies also cross borders.
Local computing infrastructure, however, gives African organisations greater choice over where their data is processed and where some of the economic activity associated with that processing takes place.
That is a more concrete measure of digital sovereignty than simply having African users of global AI products.
Cassava is also building a software layer
The company is no longer limiting the strategy to GPUs and data centres.
In 2025, it launched the Cassava AI Multi-Model Exchange, or CAIMEx, which provides access to different AI models and tools through a common platform.
Cassava says developers can use it to build, fine-tune and deploy applications using NVIDIA software, models and microservices.
The company has since introduced products running on that infrastructure, including an autonomous network system designed for African telecommunications operators.
That begins to reveal the commercial model.
Cassava is not simply installing machines and waiting for developers to rent them. It is trying to build a stack that runs from data centres and fibre through computing infrastructure to platforms and enterprise AI applications.
If it works, the company could capture revenue at several points in the AI supply chain.
“AI is helping innovators solve our greatest challenges in agriculture, healthcare, energy, and financial services. As an NVIDIA Cloud Partner, Cassava is providing essential infrastructure and software to help pioneering companies and organisations accelerate AI development across the continent,” Zuiderveld said.
Is there demand?
Installing GPUs is only the first part of the business.
AI infrastructure is expensive, and its economics depend heavily on utilisation. The machines have to spend enough time processing paying workloads to justify the capital tied up in them.
NVIDIA itself describes utilisation, energy efficiency and cost per unit of AI output as central to AI-factory economics. Cassava therefore needs more than an African AI narrative.
It needs businesses, governments, researchers and developers willing to pay for local computing capacity at sufficient scale.
The company says its target customers include organisations in financial services, telecommunications, healthcare, mining, energy, retail and the public sector. It is also working with developer communities and research organisations.
Whether those customers generate enough sustained demand will determine whether the South African deployment becomes the foundation of a continental business or an expensive pool of underused computing capacity.
The ownership question
There is another reason the Cassava project deserves attention.
Africa’s role in previous technology waves has often been strongest at the consumption end of the chain. Consumers buy devices, use platforms and generate data while much of the infrastructure, intellectual property and profit sits elsewhere.
AI creates the possibility of repeating that pattern.
Cassava cannot change the ownership structure of the global semiconductor industry. NVIDIA still supplies the core computing technology. But locally operated AI infrastructure moves one part of the value chain onto the continent.
If the planned expansion reaches Nigeria, Kenya, Egypt and Morocco, African companies would have greater access to specialised compute without every workload being routed through foreign data centres.