Sovereign AI in the Developing Global South: Volume II – Architectures, Open Silicon, and Sectoral Integration

Sovereign AI in the developing Global South has moved from data localisation to a broader mandate of intelligence sovereignty: control over algorithms, silicon and the energy infrastructure that turn local data into strategic intelligence. This second volume examines how developing states are pursuing "pragmatic sovereignty", choosing which dependencies are acceptable while localising the capabilities that secure strategic autonomy. It reviews a development-first reference architecture built on community data trusts and data exploitation licences, and the trilemma linking compute, energy and data representation. The article then traces the democratisation of the compute layer through RISC-V open silicon, with India's DIR-V programme and the SHAKTI and VEGA processors as the main case; sectoral deployments in agriculture, civic infrastructure and health, including Farmer.CHAT and Sunbird AI; fiscal and procurement instruments such as Brazil's PBIA; legal and regulatory architectures in Bangladesh and the African Union; and divergent regional paths in Latin America, through Latam-GPT, and in Sub-Saharan Africa. It concludes that the Global South cannot out-spend frontier laboratories but can build a sovereign intelligence architecture by combining open silicon, curated indigenous data and regional alliances.