The country has nearly closed its gap to the U.S. in AI bot performance, while continuing to best global competition in number of patents, publications, and rollout of robots, according to the Stanford University Institute for Human-Centered Artificial Intelligence (HAI) 2026 AI Index report released this week.

The report found a shrinking gap in Arena scores—a metric indicating relative performances of large language models—between the top AI bots in the U.S. and China. In May 2023, the U.S.’s top model, OpenAI’s GPT-4, led with more than 1,300 Arena points compared with China’s fewer than 1,000. By March 2026, that gulf shrank to just 39 Arena points, with the top U.S. model, Anthropic’s Claude Opus 4.6, leading China’s Dola-Seed 2.0 by just 2.7%.

While the U.S. still beats China in the number of top AI models—50 compared with 30—China has more publication citations than the U.S., accounting for 20.6% of AI citations in 2024 compared with the U.S.’s 12.6%. China also has nearly nine times the volume of industrial robot installations, leading the world with more than 295,000, compared with the U.S.’s 34,200.

“For years, the U.S. outpaced all other global regions on AI—in model size, performance, artificial intelligence research, citations, and more,” said Stanford’s summary of the report. “But China emerged as an AI counterweight to the U.S., gradually gaining ground, and this year it appears to have nearly erased any U.S. lead.”

China’s AI surge

Despite fewer investment dollars and wider regulatory constraints, China has changed the narrative of its ability to compete against the U.S. in a broader tech war. Spurred by its 2025 “DeepSeek moment,” China has poured funding into AI startups, with IPOs in Hong Kong last quarter reaching a five-year high of $110 billion across 40 new listings.

China has also quietly invested in its electricity infrastructure, adding more electricity demand than the entire consumption of Germany every year, David Fishman, a China energy analyst with the Lantau Group, previously said in an interview with Fortune. The country’s reserve margin has never dipped below 80%, Fishman said, essentially giving it twice the necessary capacity to grow AI compute.

China’s compute capacity is a far cry from the U.S.’s own ability to sustain and grow AI infrastructure. The American power grid system is crumbling as a result of decades of underinvestment, making it vulnerable to extreme weather and natural disasters, and ultimately creating a bottleneck Goldman Sachs suggests would stymie AI growth in the U.S. 

“We’ve actually reduced our exposure to U.S. tech,” Mohit Kumar, Jefferies global macro strategist, told Fortune at the bank’s Asia Forum in Hong Kong last month. “We believe that China is the big winner in this tech war for a number of reasons: valuation, wider adoption of AI, an advantage in power generation.”

American private investment in AI still far exceeds China’s, reaching $285.9 billion in 2025, more than 23 times China’s $12.4 billion. The U.S. funded 1,953 new AI companies last year, more than 10 times any other country, the Stanford report noted.

America’s slowing AI brain gain

AI’s momentum swing in China’s favor may be contributing to a slowdown in tech talent entering the U.S. The Stanford report found the number of AI scholars moving to the U.S. dropped 89% since 2017, and that decline is happening precipitously, accelerating 80% in the past year alone. At this juncture, more researchers are still entering the U.S. than leaving it.

“The U.S. is home to the most AI researchers and developers of any country by far,” the report summary said. “But the flow of these experts into the country is dramatically slowing.” 

Economists have warned a continued loss of expertise would further erode the edge the U.S. has over China in its talent pool. An April 2025 Hoover Institution report conducted in partnership with Stanford HAI found China has built a massive cohort of homegrown talent, with nearly all researchers behind DeepSeek’s five foundational papers educated or trained in China. Though about a quarter of DeekSeek researchers were educated in U.S. institutions, most returned to China, creating a “one-way knowledge transfer” in China’s favor, according to the report.

“These talent patterns represent a fundamental challenge to U.S. technological leadership that export controls and computing investments alone cannot address,” the authors wrote.