Carnegie China: China passed the US as the top destination for elite AI talent in 2025, rising from 2022's 27.1% share to 40.6%; the US fell from 46.4% to 34.2%
First reported by Theinformation ·
If you are an AI researcher, your peers are increasingly choosing China, influencing where cutting-edge work is done.
In 2025, China surpassed the United States as the primary destination for elite Artificial Intelligence (AI) talent. China's share of these researchers grew from 27.1% in 2022 to 40.6% in 2025. Conversely, the United States' share declined from 46.4% in 2022 to 34.2% over the same period. This shift indicates a growing trend of AI talent choosing to remain and work within China. The data suggests a significant rebalancing in the global distribution of top-tier AI expertise.
This migration of elite AI talent signifies a critical inflection point in the global AI landscape. China's rise suggests a maturing domestic AI ecosystem capable of attracting and retaining top-tier researchers, potentially accelerating its innovation and development in AI applications and foundational research. The United States' decline, while still retaining a significant portion of talent, indicates a need to re-evaluate its strategies for fostering AI innovation and attracting global expertise. This competition for talent is likely to intensify, impacting research breakthroughs and the commercialization of AI technologies worldwide.
The trend highlights a potential decentralization of AI innovation away from traditional hubs. Companies and institutions globally will need to adapt to a more distributed talent pool, potentially leading to diversified approaches in AI development and application. This also raises questions about intellectual property, collaboration, and the geopolitical implications of where advanced AI research is concentrated. Investors and policymakers will be closely watching how this talent shift influences national AI strategies and the future competitive dynamics of the AI industry.
AI-written summary. May contain errors.