How Geopolitical Fragmentation Threatens Global AI Gains

Artificial intelligence promises substantial productivity gains, but those benefits depend on technologies, data, and expertise flowing across borders. As geopolitical tensions fragment the global economy, a new study warns that the architecture of international relationships may matter nearly as much as the pace of innovation itself.

Researchers at the IMF and UC Berkeley examined how geoeconomic fragmentation affects AI's economic impact using a multi-country model of trade and production networks. Their findings reveal that while AI innovation concentrates in the US, China, and EU, the economic gains can spread much wider through three distinct layers: frontier innovators, supply-chain providers, and adopters.

Two channels of disruption

Fragmentation damages AI diffusion through two mechanisms, according to the research first reported by CEPR's VoxEU. The first involves direct barriers—tariffs, export controls, investment screening, and incompatible regulations that make AI-related inputs costlier or less accessible.

The second channel proves more destructive: fragmentation can slow innovation at the frontier itself by reducing collaboration, limiting knowledge flows, shrinking market scale, and restricting access to complementary inputs.

In the researchers' benchmark simulation, AI raises modeled global GDP by 1.64% when productivity improvements at the frontier combine with more efficient use of AI-related inputs elsewhere. Trade fragmentation alone reduces this gain to 1.49%. But when fragmentation also hampers frontier innovation, global gains fall further to 1.29%.

The connector advantage

For economies outside the AI frontier, the study identifies a strategic opportunity. Countries maintaining diversified trade and investment links across geopolitical blocs can act as connectors, potentially offsetting fragmentation losses.

The researchers examined Middle East and North Africa (MENAP) economies as a case study, comparing forced alignment with one bloc against three active strategies: serving as a connector, improving AI preparedness, and strengthening local AI capacity.

A connector strategy—reducing legitimate trade and regulatory friction with partners across multiple blocs—more than offset fragmentation losses in their simulations. Improving digital infrastructure, skills, and institutions also enhanced absorption capacity, while developing local software, data services, and deployment capabilities deepened domestic gains.

Why it matters

This research reframes the fragmentation debate for technology leaders and policymakers. The primary global risk isn't simply rerouted AI trade—it's that geopolitical rivalry damages the innovation process creating technologies everyone needs. For countries unlikely to dominate frontier AI research, preserving diversified technology relationships and investing in absorptive capacity represents valuable economic optionality in an uncertain geopolitical environment.

The findings underscore that AI's characteristics as a general-purpose technology won't guarantee automatic diffusion of its benefits. In a fragmenting world, deliberate policy choices about openness, interoperability, and capability-building will shape which economies capture AI's productivity gains.

The research was conducted by Barry Eichengreen, Jingyuan Cui, Asmaa El-Ganainy, Yevgeniya Korniyenko, Elyad Shojaei, Li Zeng, and Shihangyin Zhang, with details first published by the Centre for Economic Policy Research.