Study maps $12.7B AI infrastructure investment across Africa, reveals control imbalance

A new paper from academics including Kai-Hsin Hung, Sumaya Nur Adan, and others examines the financial and governance dynamics of AI infrastructure across Africa. Using a value chain framework, the authors systematically mapped 46 publicly announced projects worth $12.7 billion between 2019 and 2025. Their findings show a highly concentrated landscape: global data center operators, hyperscale cloud providers, and development finance institutions dominate, with investments clustering in just four countries — South Africa, Kenya, Nigeria, and Egypt. Physical infrastructure (land, buildings, power) accounts for 73% of total funding, but control over the compute layer — the actual hardware and software that runs AI — remains concentrated among a small number of global technology firms.

The paper coins the term "asymmetrical interdependence" to describe this structural condition. While African nations attract significant capital for AI infrastructure, they lack meaningful governance capacity over the compute resources themselves. The authors argue that current AI governance debates focus too much on geographic access and not enough on ownership, investment flows, and financial control. Infrastructure presence alone is insufficient for compute equity. The study, currently under review at Data and Policy journal, calls for a shift in compute governance frameworks to account for these power dynamics, urging policymakers to consider who controls the AI stack — not just where the data centers sit.