Sovereign Intelligence or Perpetual Rental: Financial Engineering, Dual-Core Capital Architectures, and Governance De-Risking for Artificial Intelligence Equity in the Global South
For two decades, digital-development finance treated telecommunications connectivity as the terminal threshold of inclusion. Frontier artificial intelligence has overturned that assumption: without indigenous compute, connectivity becomes an extraction conduit through which Base-of-the-Pyramid markets supply data and demand while renting foreign cognitive infrastructure. This article reconstructs the structural failures of grant-dependent digital finance—pilotitis, asymmetric value leakage, foreign-exchange drain, the institutional-void cost-of-capital paradox, and bilateral procurement fragmentation—and sets out a Dual Core Finance architecture that capitalises a Sovereign AI Development Fund through concessional anchor capital and community-operated velocity capital. It develops Algorithmic Impact Bonds as ex-ante surety instruments, Complementary Asset Portfolios grounded in fiduciary community data trusts, governance diagnostics that compress risk premia, and Minimum Viable Innovation Engines for frugal public delivery. Drawing on practitioner evidence from Bangladesh's Ministry of Foreign Affairs Neural Network (AI) Lab and Sundarbans deployments, and on comparative programmes in India, Brazil, the African Union and Latin America and the Caribbean, it advances five policy recommendations for the post-WSIS+20 and Global Digital Compact agenda. Written in a personal capacity; the views expressed are those of the author and do not necessarily represent the views of the Ministry of Foreign Affairs or the Government of Bangladesh.
Authors
- Syed Muntasir Mamun (ORCID: https://orcid.org/0000-0001-6845-2853)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-10-29
- DOI
- https://doi.org/10.5281/zenodo.23767816
- Primary Topic
- ICT Impact and Policies
- Type
- article
- Field-Weighted Citation Impact
- 0.00