China Now Produces 57% of Elite AI Researchers, U.S. Share Falls

China Widens Lead in Training AI Researchers

China has established itself as the world's largest source of elite artificial intelligence talent, producing 57 percent of top-tier AI researchers in 2025—an 11 percentage point jump from 2022, according to new research from Carnegie China. Meanwhile, the United States' share of researchers who earned undergraduate degrees domestically fell to just 13 percent.

The findings come from an analysis of more than 10,000 authors whose papers were accepted at the Neural Information Processing Systems conference (NeurIPS), considered one of the most prestigious venues in AI research. Carnegie researchers Damien Ma and Binyi Yang tracked where these elite researchers earned their degrees and where they currently work, creating a longitudinal picture of global AI talent flows.

U.S. Remains Top Destination Despite Tensions

While China trains more AI researchers, the United States continues to attract them at scale. Among AI researchers working in the U.S., the share holding undergraduate degrees from China rose by 4 percentage points in 2025. The U.S. recorded a net gain of 2,145 researchers, while China saw a net loss of 1,729.

However, the traditional brain drain from China to America is slowing. The ratio of Chinese-origin researchers in the U.S. versus American-origin researchers in China dropped from 46:1 in 2022 to 30:1 in 2025. More significantly, 69 percent of Chinese-origin researchers now remain in China to work, up from 57 percent three years earlier.

This shift likely reflects two forces: expanded opportunities in China's growing AI industry and tighter U.S. visa restrictions for Chinese graduate students in STEM fields, according to the researchers.

Asia Rises, Europe Falls Behind

Beyond the U.S.-China dynamic, the study reveals a broader reshuffling of global AI talent. South Korea and Singapore made notable gains, with South Korea achieving a 77 percent talent retention rate—second only to the U.S. at 89 percent. South Korea's KAIST jumped twelve places to become the third-ranked institution for producing top AI talent, while Singapore's National University of Singapore and Nanyang Technological University both broke into the top twenty.

Europe, by contrast, continued to lose ground. The continent's retention rate stands at just 60 percent, and leading institutions like ETH Zurich and the University of Oxford dropped out of the top institutional rankings entirely. India also saw its share of top AI talent decline by 2 percentage points, with no Indian institution making the top rankings.

Why it matters

AI capabilities increasingly depend on access to elite research talent—the people who push models to new frontiers. China's growing ability to both train and retain these researchers challenges assumptions that the U.S. will maintain its technological edge through talent attraction alone. For policymakers, the data suggests that visa restrictions may be accelerating China's domestic AI ecosystem rather than constraining it. For companies, the geographic distribution of top researchers signals where breakthrough innovations are likely to emerge and where competition for talent will be fiercest.

Methodology and Scope

The Carnegie study analyzed authors from the 2025 NeurIPS conference, which accepted 5,823 papers with 25,677 total authors. The researchers used AI tools to process career data for 40 percent of unique authors—those with complete educational and employment histories. The methodology tracks three career stages: undergraduate origin, graduate training, and current workplace as of 2026.

Peking University ranked as the number one institution for AI talent among the top thirty, displacing Google from its 2022 position.

These findings were first reported by Carnegie China in a feature published by the Carnegie Endowment for International Peace.