Developing countries don't need big AI models, World Bank says
The World Bank's 2026 development report advises developing economies against competing on large AI models and data centres, urging a focus on local adaptation instead.
Developing economies should not compete with advanced economies in building large artificial intelligence models or hyperscale data centres because they do not need them to harness AI's benefits, the World Bank has said in its flagship World Development Report 2026.
The World Development Report 2026: The Promise of Artificial Intelligence, released by the World Bank Group, advises policymakers to instead act urgently to adapt small, low-cost AI tools to local conditions, a message the Bank considers a lifeline for growth.
"AI has thrown developing economies a lifeline, and they should seize it," said Indermit Gill, Senior Vice President and Chief Economist of the World Bank Group, in a statement accompanying the report. "They do not need large models or big data centers to reap its benefits. By adapting small, low-cost AI tools to local conditions, they can bring better medical care, education, judicial services and agricultural extension within reach of millions."
Gill's remarks, reported by Nairametrics, form the backbone of a "foundations-first" approach the Bank has been building since 2025. Instead of chasing frontier-scale models, the Bank urges countries to strengthen the "Four Cs" of AI readiness: connectivity, compute access, context through local data, and competency via skills development.
The structural gap remains wide. Middle- and low-income countries hold only about 23 percent of global data center capacity, according to a World Bank factsheet published last year. While internet usage sits at 93 percent in high-income countries, it drops to 54 percent in lower-middle-income nations and just 27 percent in low-income ones, illustrating the connectivity gap that underpins AI readiness.
For immediate compute needs, the Bank's technical guidance notes that developing countries will largely rely on cloud services, as data centres for AI remain concentrated among a few large firms in high-income countries. However, in larger middle-income countries, private investment in local data centres can improve latency and enable regional AI ecosystems. The report cites African Union joint financing of regional data centres as one model for pooling costs.
Lower-income countries face a more fundamental balancing act: any support for data centres must be weighed against investments in basic physical and digital infrastructure, including reliable electricity and internet access, which remain prerequisites for AI adoption. The Bank's ongoing data infrastructure programs already help countries coordinate supply, demand, and regulatory rules across data centres, cloud services, and connectivity.
The World Bank has also called for cloud computing and data centres to be treated as critical infrastructure, arguing in a recent article that cloud-friendly regulation and cross-border data flows are essential for AI-ready economic growth. Its broader Digital and AI program continues to support governments in expanding computing capacity, data ecosystems, and AI governance frameworks alongside sustainable infrastructure.
The report directly challenges the narrative that countries in Africa and elsewhere must build frontier-scale large language models to participate in the AI economy. For startups and policymakers in markets like Nigeria, the strategy validates a cloud-first, local-adaptation approach over headline-grabbing national model projects, and signals potential World Bank financing for pragmatic AI foundations.
