The US–China trade war has turned AI supply chains into geopolitical pinch points: tariff and export-control battles, divergent regulations, and security-driven decoupling concentrate risk in developing countries, creating short-term relocation gains for some but longer-term dependency and barriers to autonomous AI upgrading.

Main Finding

The U.S.–China trade war has transformed economic interdependence into a source of geopolitical exposure for developing countries: it creates short-term, asymmetric export opportunities via trade diversion while producing longer‑term lock‑ins, competing regulatory/technology spheres, and securitized supply chains that undermine developmental autonomy.

Key Points

• Three core mechanisms link the trade war to new vulnerabilities for developing countries:
• Trade diversion — tariff escalation and relocation of assembly create export winners (e.g., Vietnam, Mexico) but benefits concentrate among countries with existing production capacity; many low‑capacity developing economies are bypassed.
• Alignment pressures and regulatory/standards fragmentation — competing technology ecosystems (Chinese vs. Western standards) force countries into incompatible regimes, raising switching costs and narrowing strategic flexibility.
• Securitization of dependencies — export controls, sanctions, and infrastructure conditions convert participation in supply chains into geopolitical liabilities (especially in semiconductors, critical minerals, and digital infrastructure).
• Supply‑chain restructuring often produces layered rather than true diversification: final assembly may move to third countries while upstream inputs remain concentrated in the original hub (e.g., China), creating dual dependence (one power for inputs, another for markets).
• Financial and investment channels amplify effects: policy uncertainty reduces FDI and investment appetite; geopolitical alignment increasingly conditions investment decisions (friend‑shoring), privileging strategically aligned locations over purely efficient ones.
• Institutional erosion (e.g., WTO paralysis) shifts conflict resolution toward bilateral, power‑asymmetric bargaining, constraining non‑alignment options for developing countries.
• Net result: integration into globalization continues but is channeled through power‑political frameworks rather than market logic, limiting upgrading prospects and strategic autonomy.

Data & Methods

• Methodology: qualitative process tracing to identify causal mechanisms and sequences through which U.S.–China policy shocks propagate to developing countries (drawing on George & Bennett; Beach & Pedersen).
• Evidence base: synthesis of policy documents and secondary empirical literature and reports (WTO, IMF, World Bank, UN agencies, and academic studies cited in the article such as Farrell & Newman; Fajgelbaum et al.; Mattoo et al.; Cevik; Catalan).
• Analytical focus: mechanism identification and pattern consistency rather than formal statistical generalization; emphasis on supply‑chain restructuring, trade flow reallocation, regulatory bifurcation, and financial transmission channels.

Implications for AI Economics

• Technology bifurcation has direct consequences for AI ecosystems:
  • Critical hardware (GPUs, semiconductors) and fabrication capacity are strategic choke points; export controls and CHIPS‑type policies can sharply raise compute costs and constrain access to AI training capacity in developing countries.
  • Divergent standards and infrastructure (cloud platforms, data governance, surveillance/security requirements) risk creating separate AI ecosystems that are incompatible, reducing cross‑border model sharing, pre‑trained model access, and interoperability.
  • Data flow restrictions and competing privacy/security regimes will alter incentives for cross‑border data sharing, hurting model quality and transfer learning opportunities for lower‑income countries.
  • Investment in AI (cloud, talent, startups) is becoming geopolitically filtered: friend‑shoring and alignment criteria may divert AI capital away from neutral or non‑aligned developing economies.
• Likely economic effects on AI adoption and diffusion:
  • Increased compute and data costs → slower local model development and deployment.
  • Reduced technology diffusion → longer lag in AI productivity gains, fewer upgrading opportunities within GVCs.
  • Lock‑in to a single ecosystem → dependence on vendor/standards owners for models, platforms, and updates; increased switching costs.
• Policy and research priorities for AI economists and policymakers:
• Map compute and semiconductor exposure: quantify country‑level reliance on foreign chips, cloud providers, and fabrication inputs to identify choke points for national AI strategies.
• Assess cost impacts: measure how export controls / tariffs change AI training/operational costs and downstream adoption in firms and public services.
• Study model and data access: track how bifurcation affects access to pre‑trained models, datasets, and cloud APIs for developers in developing countries.
• Evaluate alternative strategies: cost‑benefit analyses of local compute investment, regional compute sharing, public cloud partnerships, and targeted subsidies for AI infrastructure.
• Standards diplomacy: promote regional and multilateral engagement on interoperable AI standards and data governance to reduce exclusive lock‑ins.
• Finance and industrial policy: design blended‑finance instruments and public investments to build domestic AI capabilities (compute, talent, data stewardship) while hedging geopolitical risk.
• Empirical designs: exploit natural experiments (e.g., sudden export controls, friend‑shoring announcements) to identify causal effects on investment, firm performance, and AI adoption.
• Practical short‑term recommendations for developing countries interested in resilient AI development:
  • Diversify hardware and cloud suppliers where possible; invest in regional compute cooperatives.
  • Strengthen data governance and privacy frameworks to enable cross‑border collaboration under shared rules.
  • Prioritize capacity building (education, compute grants) to reduce dependence on external platforms.
  • Engage in standards forums and regional alliances to preserve interoperability and bargaining power.

Summary takeaway: the dynamics identified for trade and semiconductors in the paper extend naturally to AI: geopolitical bifurcation and securitization of technology will shape which countries can access compute, models, and markets. AI economists should prioritize mapping exposures, measuring cost and adoption impacts, and evaluating policy levers that can mitigate lock‑in and enable more autonomous, resilient pathways for AI development in the Global South.