A sovereignty push is colliding with infrastructure reality
Africa’s largest technology economies are becoming more explicit about a problem that has been building for years: they want greater control over their AI future, but much of the infrastructure needed to build it still sits in the hands of U.S. technology companies. According to a report from Rest of World, Nigeria, Egypt, and Kenya have released draft AI policies since January 2025 that identify dependence on major American firms as a strategic vulnerability, while South Africa reached a similar conclusion in a draft it published and then withdrew in April 2026 after AI tools used in drafting produced fake citations.
The core tension is not ideological. It is operational. Governments want stronger control over sensitive data, local capacity, and the terms on which foreign platforms operate, but they still depend on outside providers for computing power, funding, and expertise. That leaves African policymakers trying to build sovereignty on top of systems they do not fully own.
The challenge is substantial because AI development is inseparable from infrastructure. Model training, cloud access, data center capacity, and specialized hardware shape who can build, deploy, and govern advanced systems. On those measures, the continent starts from a thin base.
Big ambitions, limited capacity
The figures in the source material are stark. Africans account for 18% of the global population, yet the continent has less than 1% of global data center capacity, according to the World Economic Forum. The report also says the top five African markets combined have less capacity than France had in 2024, citing McKinsey.
That gap has direct consequences. It means governments that want to promote local AI ecosystems must often rent the underlying stack from foreign cloud and hardware companies. It also means public-sector ambitions can be constrained by commercial terms set elsewhere.
Some projects show momentum, but they also underline the dependency. Cassava launched what the report describes as Africa’s first AI factory in South Africa with Nvidia in March. East African data center provider iXAfrica is working with Oracle to bring Kenya its first public cloud region. Microsoft’s planned $1 billion data center project with G42 Kenya reportedly stalled after the government hesitated to commit to the computing purchases the companies wanted.
Even open-source African AI efforts are not fully insulated. Hilda Barasa, a Kenya-based senior policy adviser at the Tony Blair Institute for Global Change, told the publication that several open-source initiatives receive grants from Meta and run on Google Cloud. That illustrates the broader problem: open models or local policy goals do not automatically translate into independent infrastructure.
What sovereignty means in practice
The report suggests that “AI sovereignty” in the African context does not mean severing ties with global supply chains. Instead, it points toward a more pragmatic model centered on governance and bargaining power. Rachel Adams, founder of the Global Center on AI Governance, told Rest of World that digital sovereignty need not imply total independence. She described a more actionable version built around stronger control over sensitive data, better public procurement rules, investment in local infrastructure and skills, African-language datasets, and clearer accountability for foreign AI providers.
That framing matters because it shifts the debate away from symbolic independence and toward concrete state capacity. Governments may not be able to replace Google, Microsoft, Nvidia, or Meta in the near term. But they can try to write better contracts, set clearer rules for data use, support domestic technical talent, and reduce the extent to which AI adoption simply imports external dependencies.
Those measures are harder than publishing a strategy document. They require sustained investment, procurement discipline, and institutions capable of enforcing standards across public and private deployments.
The coordination problem
The report notes that Africa’s four biggest tech economies are not operating in a vacuum. They are competing for investment even while trying to build a more unified position. A proposed $60 billion fund and an AI council are presented as efforts to coordinate, but the underlying incentives remain complicated. Countries want data centers, financing, and partnerships now, which can weaken their leverage when negotiating longer-term terms of control.
That creates a familiar policy trap. The faster a country wants AI infrastructure, the more likely it may be to accept dependence as the price of entry. The more it emphasizes sovereignty conditions, the greater the risk that capital and projects shift elsewhere. The result is a patchwork of draft strategies confronting a market still dominated by foreign providers.
South Africa’s withdrawn draft adds another wrinkle: the governance systems shaping AI policy can themselves be brittle. A strategy document undermined by fabricated citations from AI tools is not just an embarrassment. It is a reminder that institutional readiness matters as much as political messaging.
Why this matters beyond Africa
The African debate is a compressed version of a wider global argument about AI power. Many countries want the economic and public-service benefits of AI without ceding control over data, infrastructure, or policy autonomy. Few have the domestic compute base, capital pool, and technical ecosystem to do that on their own.
What makes Africa especially important is that its choices will influence how a large, young, multilingual population is represented in the AI era. If the market develops mostly on foreign clouds, funded by foreign capital, and trained on data pipelines controlled elsewhere, local priorities could be constrained from the start. If governments can bargain for stronger rules, support local datasets and skills, and expand domestic infrastructure, they may secure more room to shape outcomes.
The report does not present a quick solution, and none is obvious. But it does show a notable policy shift: several of the continent’s biggest economies are no longer treating dependence on foreign AI firms as a background condition. They are naming it as a strategic problem. Whether that naming turns into durable leverage will depend less on declarations of sovereignty than on the harder work of building capacity, writing enforceable rules, and negotiating from a stronger institutional position.
This article is based on reporting by Rest of World. Read the original article.
Originally published on restofworld.org
