The Great AI Divide: Navigating U.S. and Chinese dominance
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Sam Winter-Levy of the Carnegie Endowment for International Peace stated the U.S. and China control 90% of global computing power, attract between 70% and 80% of global AI investment, and concentrate the world's top AI researchers, putting most of the Global South in a dependent position.
- Anthropic's Mythos model is being rolled out only to small, select groups of companies initially, per Winter-Levy, exemplifying a broader shift among U.S. frontier developers toward managed-access approaches that let companies choose who gets technology.
- Cornell University's Aditya Vashistha said AI systems are designed for WEIRD societies (Western, educated, industrialized, rich, and democratic) representing only 14–15% of the world's population, embedding biases against the 85% majority, including 1 billion people with disabilities whose safety benchmarks aren't even accounted for.
- Peter Micek of Access Now noted the first two U.N. General Assembly AI resolutions were each led by a single country — the U.S. and China respectively — and unlike most resolutions, neither collaborated with the other, with the Biden-era U.S. resolution applying human rights to the entire AI lifecycle.
- Middle powers face three options — forming coalitions, building sovereign models (as the UAE and India discuss), or bandwagoning with a great power — but sovereign efforts still rely on Nvidia-designed U.S. chips and U.S.-serviced data centers, per Winter-Levy.
- Severe compute constraints are forcing companies to ration access to AI systems, and both the U.S. and Chinese governments are playing increasingly assertive roles in determining who can use these technologies, compounding the leverage gap.
Why it matters: When 90% of AI compute and 70–80% of investment sit in two countries, every other nation's AI strategy — from sovereign models to regulation — runs through Washington or Beijing. The convergence of managed-access rollouts, compute rationing, and government export controls means middle powers and the Global South face AI systems built without their languages, values, or disabilities, with diminishing room to build independent alternatives.
Ask SkimNews
