Discussion of AI governance in Africa is dominated by data protection, algorithmic accountability and national AI strategies. Electricity rarely features. Yet no AI application functions without it, and the decisions that will determine whether AI infrastructure strengthens or strains African power systems are crucial. Through connection agreements, tariff determinations, power procurement, infrastructure investment and siting approvals, these discussions are live policy questions.
On this basis, we are introducing the “Powering AI Series”, a collection of short and focused articles for policy stakeholders to explore the energy infrastructure, regulatory and policy questions that will shape how AI develops across African power systems. This matters because the deployment and growth of AI depends on physical infrastructure including data centres, generation facilities, transmission networks, distribution systems and power substations. Generation capacity and reliable electricity supply are also key determinants of where AI infrastructure can be located, how quickly it can scale and how its growth can be sustained.
As African countries seek to attract investments in data centres and cloud infrastructure, energy availability, affordability and grid capacity will increasingly shape where these systems can be built and who can access them. At the same time, decisions about how new large electricity loads are connected, priced and prioritised can have consequences well beyond individual projects, particularly in power systems that already face supply constraints, high costs or uneven access. Energy planning must therefore be embedded in AI governance.
For Africa, energy policy should become a core pillar of AI governance and not a parallel policy conversation. Appreciating this connection is essential for the continent because expanding digital ambitions must go hand in hand with strengthening energy systems.The central argument is that the relationship between AI and electricity runs in two directions, and that most public discussion considers only one half. AI is a growing claim on power systems: data centres are large, continuous loads that require high levels of reliability, and they arrive in specific places on specific grids.
However, AI is also becoming a tool for running those same power systems by forecasting demand, optimising networks, detecting faults, integrating renewable generation, managing distributed energy resources and improving system efficiency. Policy approaches that treat AI only as a burden on the grid will miss what it can return. Similarly, a policy approach that sees AI only as an opportunity will eventually confront the realities of the power system, often beginning with the connection queue.
This series explores all these by moving from the physical to the contractual to the political. It begins with what AI infrastructure actually consumes and where that demand is likely to emerge in Africa. It then turns to the grid itself. Particularly, it looks at why available generation is only part of the equation, why transmission and distribution capacity matter, and why grid capability rather than investor appetite ultimately determines what can be built and where.
From there, the series examines the instruments that determine the terms on which AI infrastructure enters the power system, including tariffs, power purchase agreements, connection agreements, connection priority and decisions about who pays for network reinforcement. It looks at what exists in Africa today rather than what has simply been announced, distinguishing between proposed AI and data-centre investments and the infrastructure that is actually being built and connected.
The practical question follows. When a large new electricity load connects to a system where households and businesses may still face unreliable or unaffordable supply, how should that connection be governed? Who bears the cost of the additional infrastructure? Who benefits from new investment? And what obligations, if any, should accompany preferential access to scarce power or grid capacity?
The series closes by examining what these developments mean for the governance of electricity systems as AI infrastructure is developed across the continent. If governments want to attract AI infrastructure while expanding reliable electricity access and maintaining resilient power systems, AI and energy policy cannot be developed as separate conversations. The choices being made about power today will shape where AI can develop tomorrow and, in turn, how AI may reshape Africa's power systems.
