Africa’s AI Sovereignty: A Strategy of Leverage, Not Independence
The conversation around AI sovereignty in Africa is often framed by a vocabulary problem. The term “sovereignty” evokes images of independence, self-reliance, and a decisive break from foreign infrastructure. Yet, this idealized vision is far from the current reality. What is unfolding instead is a more nuanced and pragmatic approach: Africa’s AI sovereignty is being shaped as a deliberate strategy of remaining within the global Big Tech supply chain while negotiating better terms and greater control within it.
This apparent paradox is not a contradiction awaiting resolution—it is the plan itself.
The Structural Challenge: 18% of Population, Less Than 1% Data Centres
To understand the complexity, start with the stark numbers. Africa is home to roughly 18% of the world’s population, yet it holds less than 1% of global data centre capacity, according to the World Economic Forum. McKinsey’s research highlights that the top five African markets combined have less data centre capacity than France had in 2024. This is not merely a gap; it is a chasm that underscores the continent’s infrastructural limitations.
Without local data storage capacity, localization of data is impossible. Sovereign AI models require local compute power, which is currently scarce. Negotiating from a position of strength is difficult when the alternative to existing deals is essentially “no deal” at all.
Therefore, the most accurate framework to view Africa’s AI sovereignty efforts is not independence, but leverage.
Negotiating Dependency: The Big Four and the GPU Backbone
In April 2025, during the inaugural Global AI Summit on Africa in Kigali, African leaders unveiled a $60 billion AI fund. Notably, the largest hardware investment is 12,000 Nvidia GPUs allocated to Nigeria, Egypt, Kenya, South Africa, and Morocco. The fastest route to AI sovereignty is through these Nvidia chips, supported by partnerships with Google Cloud and Microsoft data centres.
Critically, the continent’s four largest tech economies openly acknowledge their dependency on US tech giants like Google, Microsoft, Nvidia, and Meta. Their national AI strategies explicitly identify reliance on these companies as a security risk, not just a procurement hurdle.
Rachel Adams, founder of the Global Center on AI Governance, succinctly captured this reality: “Africa’s push for digital sovereignty cannot mean total independence from global AI supply chains. But it can mean stronger control over sensitive data, better public procurement rules, investment in local infrastructure and skills, African language data sets, and clearer accountability for foreign AI providers.”
The Kenya Signal: Pushing Back on Unfavorable Terms
A clear example of emerging negotiation strength came with Microsoft’s $1 billion data centre deal in Kenya, announced with UAE’s G42. The project stalled when Microsoft and G42 demanded annual capacity payments that President William Ruto found untenable, noting the energy demands would force Kenya to “switch off half the country” to maintain the facility. The infrastructure was available, but the terms were not acceptable.
This setback is significant because it demonstrates that African governments are increasingly willing to reject deals that impose unsustainable conditions. While the cost of refusal includes delayed infrastructure and potential capital flight, the long-term precedent of demanding better terms is invaluable.
Similar resistance is visible in health data agreements. Ghana, Zimbabwe, and Zambia have recently walked away from US-linked health data-sharing deals, citing concerns over data sovereignty after USAID restructuring raised fears of commercial exploitation of African health data. This refusal often means forfeiting much-needed funding, illustrating the difficult trade-offs governments face. Nigeria’s contrasting approach in signing the US America First Global Health Strategy highlights the diversity of positions within the continent.
Parallel to these refusals, positive initiatives are emerging. iXAfrica’s collaboration with Oracle to deliver Kenya’s first public cloud region and Cassava Technologies’ launch of Africa’s first AI factory in South Africa, featuring Nvidia supercomputers, underscore a strategy of selective acceptance—embracing partnerships that build local capacity without ceding control.
Tay PeiChin of the Tony Blair Institute warns that partial infrastructure can be hollow: “We have seen in some other countries in North Africa, where they build data centres, put their data in, but then they outsource the management of the data centre to a third-party provider, who can just lock up the thing and throw away the keys. So it’s not just about having control, but having meaningful control.”
The $60 Billion Fund: Ambition Meets Realism
The $60 billion Africa AI Fund is the continent’s most concrete response to AI sovereignty, encompassing infrastructure, talent development, and startup support. Yet, the fund’s largest line item—12,000 Nvidia GPUs—reveals the structural reality. Sovereignty here is not technological ownership but a strategic purchase of negotiating leverage and time.
This fund provides the option to reject poorer deals in the future and to develop African language datasets on hardware physically controlled within the continent, even if designed abroad. It reflects a shift in understanding: sovereignty in 2026 does not mean owning the entire AI stack—which remains the domain of a few US and Chinese giants—but having sufficient leverage to walk away from disadvantageous agreements without catastrophic consequences.
However, coordination remains the greatest challenge. As Hilda Barasa of the Tony Blair Institute notes, no single African country’s workload justifies standalone large-scale infrastructure; regional cooperation is essential. Yet trust between governments is fragile, often overshadowed by stronger ties to Washington or Beijing than to one another. This fragmentation plays into Big Tech’s hands, enabling bilateral deals that fracture continental unity—a dynamic exemplified by Nigeria’s divergence in health strategy from its neighbors.
The African Union’s July 2024 Continental AI Strategy and Smart Africa’s November 2025 Africa AI Council are critical steps in building the institutional framework necessary for regional coordination. Their goal is less about creating frontier AI models and more about establishing the “plumbing” of common standards, shared procurement, pooled talent, and mutual data governance recognition. This foundational work will determine whether the $60 billion fund translates into genuine leverage or simply reinforces existing dependencies under new branding.
Balancing Global Powers: Navigating Between the US and China
Priya Vora, CEO of the Digital Impact Alliance, highlights that African nations seek deeper digital market integration while maintaining cautious distance from dependency on both China and the United States. This balanced wariness marks a departure from past decades’ binary framing, which forced a choice between US-led or Chinese-led digital orders.
Current strategy documents reject this dichotomy, emphasizing the need for optionality from both global powers. The $60 billion fund, then, is the price of keeping multiple doors open.
Yet, this optionality is asymmetrical. The hardware backbone is predominantly American, as are cloud credits and frontier AI models. While Chinese alternatives exist in financing and connectivity, they play a minimal role in the $60 billion fund’s compute architecture. The procurement realities thus favor the US tech ecosystem, and this imbalance will persist until Africa cultivates substantial domestic compute capabilities.
In effect, the optionality is real but narrow and time-bound. The fund provides a window—estimated between five and ten years—to build the institutions, regional coalitions, and talent base necessary for a stronger negotiating position in the next phase of AI development.
If this window is leveraged effectively, the Kigali announcement will be remembered as the moment Africa’s leverage strategy began to compound. If not, the 12,000 Nvidia GPUs will stand as costly evidence that naming dependency is not equivalent to overcoming it. The strategy is defensible; the proof lies in execution.
