AI infrastructure, governance preparedness, and digital state systems in Africa

This paper investigates the relationship between AI infrastructure integration and governance across Africa using project-level data on 123 Chinese-financed AI-related infrastructure projects implemented in 40 countries between 2002 and 2022. The empirical analysis employs multivariate logistic regression, complemented by clustered, multilevel, grouped binomial, and fractional logit models to assess the robustness of the findings. Building on the concept of governance lag, three main findings emerge. First, integrated AI infrastructure is associated with weaker governance conditions. Second, integrated AI infrastructure is associated with larger project values and projects committed following shifts in the external policy environment for digital infrastructure. Third, electoral democracy and political freedoms display more consistent associations with infrastructure integration than data-protection legislation, suggesting that formal legal frameworks provide only a partial representation of the governance conditions under which integrated AI infrastructure is implemented. These findings remain robust across alternative estimation strategies, financing controls, sample restrictions, and country-level sensitivity analyses. Together, the findings provide project-level evidence on how AI infrastructure integration varies across governance environments, contributing to research on AI governance, digital public infrastructure, and institutional capacity.