Nvidia has announced a strategic partnership with Cassava Technologies in a deal valued at US$700 million, aimed at establishing AI-ready data centres across multiple African nations.
This marks Nvidia’s first direct infrastructure deployment on the continent and represents a significant step toward expanding access to advanced graphics processing units (GPUs) in regions historically lacking sufficient computational capacity.
The project exemplifies a departure from the predominantly state-led Chinese investments in Africa’s digital infrastructure. Instead, Nvidia’s approach emphasises private-sector-led development, with Cassava spearheading the design and construction of data centres tailored to support AI growth in sectors such as healthcare, agriculture, and financial technology.
Ziaad Suleman, CEO of Cassava Technologies, which operates in South Africa and Botswana, described the deal as a major milestone for African enterprise. He stated that Africa has often been limited to second-rate technology, but Nvidia’s involvement will help transform this dynamic by bringing high-performance computing capabilities to the continent’s burgeoning AI developers and data-driven industries.
The partnership’s initial phase launched in June with the deployment of 3,000 Nvidia GPUs to a new facility in South Africa built by Cassava. Over the next three to four years, the plan is to develop additional AI factories equipped with a further 12,000 GPUs across Egypt, Nigeria, Kenya, and Morocco.
In addition to infrastructure development, Cassava has signed a memorandum of understanding with the South African AI Association. This collaboration will provide more than 3,000 AI professionals with direct access to Nvidia GPUs, supporting efforts to cultivate human resources alongside technological advancements.
The initiative directly addresses the stark AI infrastructure gap in Africa, where the UN Development Programme reports that only 5% of AI practitioners have access to the necessary computational resources. Of those, only a fraction has on-premises GPU access, with most relying heavily on limited cloud budgets. The new data centres will offer high-bandwidth, low-latency connections suited for model training and inference, enabling local developers to perform complex AI tasks within Africa’s borders — a capability largely unavailable until now.
According to Omorogieva, access to Nvidia’s advanced chips and dedicated infrastructure could enable African nations to move beyond being mere consumers of AI technology. Instead, they could foster the development of indigenous AI systems, unlocking new economic and technological potential across the continent.
