AI in Emerging Markets: An ID30 Reading of the New IFC and World Bank Handbook
The new IFC and World Bank Group handbook, “Accelerating Artificial Intelligence Investment in Emerging Markets” (May 2026), offers a framework rather than a forecast. It gives governments, investors, and builders two lenses to judge where AI investment can compound, and where it stalls. Here is how we read it at ID30.
Two lenses, read together
The Ecosystem lens looks at who builds and enables AI. The Structural lens looks at the preconditions that make scaling last: data, digitization, energy, and construction. The authors are clear that either lens alone misleads. Pilots dazzle, then stall on structural reality. Read together, they reveal a feedback loop where early ecosystem wins justify infrastructure bets, and infrastructure upgrades unlock the next wave of solutions.
DPI sits at the foundation
The handbook places digital public infrastructure at the base of the AI stack: identity, payments, data exchange, and registries, the shared rails that make AI deployment possible at national scale. As models and tooling commoditize, durable advantage shifts away from raw compute toward proprietary and trusted local data, verified performance in compliance heavy workflows, and integration with national payment and identity systems. Access to a powerful model alone is not enough to compete.
The ID30 lens
Here is the shift that matters most for our work. For years, the hardest constraint on open source digital public goods was not the code. It was the team. AI drastically reduces the large skilled teams these solutions once required, and it eases the brutal retention battle that follows. That gives open source DPGs a second life, and gives local ownership a real chance.
It does not remove the human. People stay on top, taking decisions, orienting change, and adapting to local context and priorities. And it does not remove the risk. Security and privacy demand even more care as automation deepens.
The new frontier is sovereign and local infrastructure
Cheaper capability is not the same as sovereign capability. If the compute, the data, and the models all sit offshore, the dependency has simply moved up the stack. Our working answer is a hybrid AI stack: edge for the field and low bandwidth, local for sovereign data and control, cloud for scale, and even space for connectivity where ground infrastructure is thin. Not one layer, a deliberate combination of all.
The reframe: sovereignty is no longer about owning a model. It is about owning the architecture underneath it.
Acknowledgments
The handbook is the work of Lana Graf, Anastasia Nedayvoda, and Eveline Smeets at IFC and the World Bank Group. Thanks to Paul Nguyen for highlighting it to the community.
Want to comment, ask a question, or join the discussion? Continue the conversation on LinkedIn.