“For Africa, AI has the potential to significantly impact the attainment of Agenda 2063 Aspirations and the Sustainable Development Goals (SDGs). It is seen as a driving force for positive change, socio-economic transformation and cultural renaissance.” – African Union.

Artificial Intelligence (AI), in the above quote, is everything at once: engine, catalyst, and cultural lodestar. To lend this ascription a veneer of precision, the African Union (AU), in its first and only continent-wide AI strategy, turns to a familiar source of authority: an estimate of generative AI’s impact on global GDP by McKinsey. Somewhat arbitrarily, the AU extracts a neat 5 percent share for Africa and concludes that generative AI could add between $110 to $220 billion to African GDP a year. A striking figure, certainly. A very elastic one too. To dwell on the elasticity of these projections is, perhaps, to miss the point. Whether the figure is $110 to $220 billion is secondary to the conditions under which Africa’s AI gains might be realised.

Internal Constraints & External Challenges

The internal constraints on Africa’s AI uptake are well-rehearsed. An analysis of oft-cited inhibitors reveals a problem not of absence, but of disarticulation. Talent pipelines are expanding: between 2019 and 2024, Africa recorded a 39% increase in access to computer science (CS) education, outpacing Asia (33%) and Europe (25%). In relative terms, then, Africa is catching up. In absolute terms, it is not – students in African countries remain the least likely to have access to CS education, a disparity rooted in chronic underinvestment and reflected in the continent capturing just 0.02-0.03% of global AI investment in 2025.

As low levels of investment impede the development of data centres, data storage facilities, and high-performance computers, the continent is experiencing an exponential increase in data generation. What is lacking is access: data remains siloed, poorly structured, and legally fragmented, inhibiting its use for common goods. These constraints converge at the level of governance, where regulatory frameworks are partial and uneven, limiting standard-setting capacity and eroding public trust.

Internal constraints on Africa’s AI uptake only tell part of the story. AI is not insulated from geopolitics and, as it has gained salience on the political agenda, it has become embedded in the strategic calculus of major powers – most notably, the US and China. Manifestations of this calculus increasingly extend beyond US-China frontier model competition, metastasizing into the global diffusion of AI systems. Both the US and China proceed on the assumption that exporting open models will cultivate dependency among middle powers on their AI systems and technological stacks, thereby consolidating global market share. For African states, this creates a risk not of exclusion from the emerging AI order, but of incorporation into it on asymmetric, dependency-inducing terms. This risk operates along two dimensions:

First, horizontal lock-in: this is where dependence emerges at the level of applications and economic integration. As firms and governments adopt particular model families and standards, their digital economies – and AI-enabled services – become entrenched in and contingent on the underlying ecosystem of the exporting power. Second, vertical lock-in: here, dependence is embedded deeper in the technological stack. In this dimension, adopting an open-source model entails adopting the infrastructure on which it runs: cloud platforms, data pipelines, and deployment infrastructure, inter alia. Over time, as the cost of reconfiguring tech ecosystems becomes prohibitive and alternative ecosystems become harder to integrate, this creates path dependencies for the importing state.

Current AI Initiatives in Africa

With a view to addressing these challenges, policymakers at national, continental, and international levels have begun to articulate a range of initiatives aimed at improving Africa’s AI uptake and closing the AI equity gap. Notable among these are the AU’s 2024 Continental Artificial Intelligence Strategy and the AI Task Force launched by South Africa during its 2025 G20 presidency. These initiatives reflect a clear and increasingly coordinated diagnosis of Africa’s structural constraints in AI, as well as a recognition of the need for cross-border collaboration. What they omit, however, is a consideration of how such cooperation might be institutionalised in a way that enables African states to pool their capabilities, shape markets, and exercise collective leverage in the emerging AI order.

As external actors look to consolidate their positions at the apex of the emergent AI order, leveraging open-source model exports to create strategic dependencies, African states remain institutionally fragmented. While current initiatives underline the importance of coordination, they conceive of it in terms of alignment and capacity-building, a focus which distracts from the advantages to be gained at the level of scale – through shared data infrastructure, interoperable standards, pooled data, and aggregated market demand. What is missing is not cooperation per se, but the institutionalisation of cooperation within a framework capable of aggregating capabilities across borders and converting Africa’s growing inputs – talent, data, and digital adoption – into strategic autonomy in the emerging AI order. What follows from this diagnosis is not simply the need for more coordination, but for a particular kind of institutional arrangement.

The Institutional Construct

Although much of the work required to enable Africa’s effective uptake of AI will occur at the national level, regional cooperation is essential if AI is to bring about broad benefits across African societies. The cross-border character of the digital ecosystem render strategies that are purely national inefficient and costly, while regional cooperation unlocks efficiency gains across research, data, and infrastructure. Buttressing this practical imperative for regional cooperation in AI is a geopolitical one: whereas individual African states lack the market power to shape the terms of technological integration and resist lock-in to superpower stacks, regional cooperation changes this equation. By aggregating demand and coordinating interoperable standards, African states can pool their purchasing power and exert collective influence over the AI market.

The institutionalisation of a regional AI strategy championed by select African middle powers confers two advantages. Endogenously, African states will be better placed to overcome the internal constraints facing the continent’s AI uptake by, for example, harmonising governance standards and enable cross-border data pooling and transforming talent pipelines through shared training programmes and regional labour mobility. Exogenously, by transforming fragmented continental engagement with AI into a consolidated, regional structure, African states can establish regional procurement frameworks that prioritise the modularity of tech stacks and open standards, mitigating the risk of lock-in and subordination to the AI ecosystem of any exporting power.

Cooperation must therefore be regional and capable of generating leverage. Importantly, however, it must do so within a policy context shaped by states’ perception of AI as a strategic, uncertain, and competitive technology, which constrains the range of viable policy tools. Accordingly, reviews of existing supra-national AI initiatives observe that while binding international treaties that govern the development and use of AI seem far-fetched at this stage, influence can be exercised through softer forms of power, such as convening power and norm- and agenda-setting. As Lewin Schmitt observes:

“It may well be that governments [decide] to transfer authority to IOs only as long as they deal with rather abstract principles or soft governance, but would withdraw or stall as soon as work proceeds towards more regulatory, hard governance.”

If what remains of international cooperation in AI is soft, flexible, and contingent, then the case for highly formalised structures weakens. In this context, informal intergovernmental institutions (IIGOs), precisely because of their informality, are better suited to coordinate action without triggering resistance of withdrawal.

Emerging Middle Powers

Although certain barriers, such as misaligned incentives, lay in the path to strategic cooperation on AI, the current geopolitical environment offers an aperture through which Africa can chart a course to continental autonomy in AI may be chartered. More specifically, as the Western-led liberal order erodes and competing visions of international organisation begin to take root, the more competitive geopolitical environment gives Africa ‘exit options’ from the strictures of the liberal order.

To seize these exit options and advance Africa’s strategic autonomy in an increasingly bifurcated world order, the continent’s emerging middle powers – South Africa, Nigeria, and Egypt, for instance – must exercise their technical capacities and economic weights to convene and sustain institutionalised regional cooperation on AI.

Three Policy Levers for an African AI Coalition

Drawing on the epigraph, for AI to serve as the engine, catalyst, and cultural lodestar that the African Union envisions, the institutionalisation of a regional AI strategy is critical. This requires more than mere alignment; it demands an institutional vehicle – most plausibly an informal intergovernmental organisation – through which African states can pool capabilities, shape markets, and bolster strategic autonomy. Underpinning Africa’s approach to its regional AI strategy should be three key policy tenets: first, the establishment of a regional data commons; second, the prioritisation of interoperability and competition through coordinated procurement frameworks; and third, the development of modular, open AI infrastructure through coordinated regional investment and standards-setting.

Questions for Further Reflection

How can the gains from a regional AI strategy be distributed in a manner that sustains political buy-in among African states?

Under what conditions might an informal intergovernmental institution evolve into a more formalised structure?

To what extent does the pursuit of modular, AI ecosystems genuinely expand strategic autonomy?

Recommended Readings

‘Strategic Cooperation on AI: Core Functions’ RAND. 10 March, 2024.

Schmitt, Lewin. ‘Mapping Global AI Governance: A Nascent Regime in a Fragmented Landscape’. AI Ethics. 17 August, 2021.

‘Open Source: How Middle Powers Can Build Influence in the Age of AI’. Tony Blair Institute for Global Change. 9 February, 2026.