How the U.S. Can Win the AI Race

Artificial intelligence is reshaping the world. Who controls the reshaping—and under what conditions—remains far from settled. The past year has made clear that the race ahead is actually multiple overlapping contests unfolding at once.

The first is between the United States and China. American frontier labs continue to lead the push toward artificial general intelligence (AGI), investing extraordinary sums to build systems that could equal or surpass human cognition. The stakes are immense: AGI could shift the global balance of power and generate unprecedented economic growth.

China, however, is pursuing a different path and believes that widespread AI adoption will deliver the greatest gains. Its AI+ initiative targets 90% integration across key sectors by 2030, with a particular focus on AI-enabled manufacturing. The early results are striking: an Edelman poll found that 60% of Chinese employees use AI at least weekly, compared with 37% of American workers. While the U.S. positions its economy to build the most powerful AI systems, China is deploying AI to build the most powerful economy.

Read More: What Happens When AI Replaces Workers?

The U.S. and China are not the only forces shaping the global landscape. The European Union, seeking to lead through governance rather than innovation, adopted the world’s first comprehensive AI law in 2024. But uneven enforcement and a growing list of exceptions have created a regulatory morass that is stifling the continent’s transformation. Gulf states are charting a different course as Saudi Arabia and the UAE invest hundreds of billions of dollars in data-center infrastructure, positioning themselves as critical nodes in the AI boom.

Another competition is unfolding over the architecture of AI itself: open versus closed systems. The U.S. has historically championed technological openness, yet much of its AI leadership now centers on proprietary models. Meanwhile, open-source momentum is accelerating elsewhere. In January, Chinese company DeepSeek released its R1 model demonstrating that companies can build highly capable models without top-tier chips, a reminder that cheap and adaptable models may spread fastest globally.

The U.S. can remain the global AI leader in 2026 only if it competes on all fronts. Consolidating its advantage in frontier AI will require solving the energy bottlenecks that currently constrain compute capacity. At the same time, the private sector must accelerate AI adoption by upskilling workers and making powerful models accessible to small and midsize businesses. And the country should deepen its collaboration among government, industry, and research institutions to reclaim open-source leadership. Building the most powerful systems matters little if other nations deploy them more effectively.

The greatest risk America faces is winning the AI frontier and still losing the AI era.