AI use spans nearly 150 countries, but the infrastructure that runs it—data centers, chips and cloud systems—is concentrated in the US and China. Nine companies control about 70% of AI computing hubs: American firms run 87, Chinese firms 39, and European firms only six. Governments are pursuing "AI sovereignty," and the EU estimates achieving technological autonomy could cost close to $600 billion over the next decade.
AI Is Global — But Its Infrastructure Is Concentrated in the US and China
Artificial intelligence is now used in nearly 150 countries, yet the physical and software infrastructure that powers it is heavily concentrated in a few places. The United States and China dominate key components—data centers, cloud services, specialized processors, storage software and networking hardware—while European firms lag far behind.
According to a recent Oxford University report, nine companies control roughly 70% of the market for AI computing hubs. Of the hubs counted, 87 are run by American companies (about two‑thirds of the global total), 39 by Chinese firms, and only six by European companies. Taken together, the US and China account for roughly 90% of global AI computing power.
Why That Concentration Matters
Countries without substantial domestic AI infrastructure face a stark choice: rely on US or Chinese technology stacks or invest heavily to build their own. That dependence raises strategic and economic questions about security, data sovereignty, supply chains and political leverage.
An emerging response is a push for "AI sovereignty"—the effort by governments outside the US and China to host and control critical parts of the technology stack on their own soil. The Washington think tank Center for New American Security (CNAS) reported that state‑backed AI projects outside the United States and China increased fivefold between 2024 and 2025. India, Japan and Singapore are among the countries supporting domestic infrastructure and model development, and Canada has announced plans to reduce reliance on American providers.
The European Union has unveiled a technology‑sovereignty package to expand domestic production of advanced semiconductors, AI models and cloud services. The Center for European Policy Analysis estimates achieving sovereignty in cloud computing and AI could cost Europe nearly $600 billion over the next decade—an investment roughly equivalent to 30% of the EU's annual budget over that period, illustrating the scale of the commitment required.
French President Emmanuel Macron has proposed a "third way": a coalition of middle powers—India, Brazil, Canada, Gulf states and others—pooling computing capacity and investment to reduce global dependence on the US and China.
Faced with these dynamics, governments confront a clear tradeoff: spend vast sums to recreate technologies already developed abroad or accept continued dependence on two dominant ecosystems. The choices made now will shape the global AI landscape—economically, strategically and politically—for years to come.
Help us improve.
