The United States and China are building rival AI ecosystems that leave other countries dependent on their capital, chips and technical standards, former Google CEO Eric Schmidt and two coauthors argue in Foreign Affairs. Their thesis is that financing and infrastructure may decide the contest as much as model breakthroughs.
The authors compare OpenAI and Anthropic to colonial-era trading companies: private organizations that negotiate with governments while remaining tied to national power. They characterize the US approach as exporting closed systems alongside political relationships, while China spreads open-weight models and its own standards.
That analogy is an interpretation, not a measured finding. The concrete constraints behind it are easier to observe. Advanced AI requires large amounts of electricity, minerals, semiconductor capacity and investment. US corporations plan to spend more than $600 billion on AI in 2026, according to the essay, while both blocs face debt costs and domestic resistance to data-center construction.
Taiwan is the most immediate shared vulnerability because a conflict involving TSMC could disrupt the supply of leading chips to both sides. Countries choosing an AI stack therefore face more than a software decision: they may inherit cloud dependencies, governance rules and geopolitical exposure. The authors’ framing may overstate the unity of each bloc, but it highlights why access to compute and supply resilience increasingly shape national AI policy.