AI Race: How the U.S. Can Combat China
AI Race: How the U.S. Can Combat China
Publish Date: 2026-06-22 07:02:00
Source Domain: foreignpolicy.com
Policy analysts often frame the artificial intelligence race as a sprint between Washington and Beijing, with each running on its own track. The United States is driving toward artificial general intelligence, an AI system that surpasses human brainpower and entirely transforms society. China, however, is building efficient, scalable AI capabilities and embedding them in economic and public life worldwide. Silicon Valley’s focus is on discovery, Beijing’s on diffusion.
Ultimately, these racetracks will converge. AI dominance will derive from having best-in-class technologies and entrenching them into industries, governments, and education systems across the globe. If Washington treats the contest solely as a race to hit new capability benchmarks, it could lead in invention but fall behind in influencing where and how AI is used worldwide.
To prevail in the AI race, the United States should augment its clear advantage in frontier innovation with a concerted strategy to foster global adoption of its AI stack, the set of technical layers needed to build, operate, and use AI systems, from the computing infrastructure underneath to the applications with which people interact.
This would not only give the United States a leg up on China in the defining technology of the future. If it prioritizes values, safeguards, and accountability, a U.S. plan to promote AI openness might also help stanch the backlash against U.S. technology and rebuild the country’s historic reputation as a force for rights and government responsibility. Done well, it could even secure U.S. global economic leadership and begin to restore faith that U.S. values may yet have something to offer the world.
Currently, the United States still supplies much of the world’s default AI stack: Nvidia dominates data center processing units; Amazon Web Services, Microsoft Azure, and Google Cloud command nearly two-thirds of global cloud infrastructure; and U.S. labs produced 40 notable AI models…
