The Open‐Source Paradox: Africa's Digital Sovereignty and the Structural Limits of Artificial Intelligence Autonomy

ABSTRACT Open‐source artificial intelligence is widely promoted as a democratising pathway to digital sovereignty for African states, offering access to frontier architectures without prohibitive capital investment. This paper investigates whether open‐source AI represents a credible route to autonomy or generates a new form of structural dependency. Drawing on the National Innovation System (NIS) theory and the political economy of cloud infrastructure, the paper argues that open‐source AI transfers model weights but neglects the structural foundations of capability: compute infrastructure, localised data and indigenous human capital. A structured narrative review demonstrates that Africa's research marginalisation, chronic infrastructure financing deficits and reliance on foreign cloud services collectively undermine the promise of open‐source AI's sovereignty. The analysis establishes that the compute layer, not the code layer, is the primary locus of power: consequently, adoption of foreign infrastructure relocates rather than resolves dependency. Three analytical contributions are advanced: a theoretically grounded critique of the open‐source paradox that integrates NIS and Big AI scholarship; an integrative framework that applies these resources to the African case; and a Digital Bandung collective action framework. This proposal presents the Digital Bandung framework as a heuristic device and an ideal type. By utilising this historical analogy, the paper illustrates a strategic logic for collective action rather than a rigid policy prescription, acknowledging that the original Bandung spirit must be modernised for the digital age. While treating Africa as a focal case, the study acknowledges that this approach risks obscuring significant intra‐continental variation.