Stanford index: talent and performance shift

Stanford’s 2026 Artificial Intelligence Index says the United States still leads on top models, but China has cut the gap to 2.7% on the benchmarks Stanford tracks. (hai.stanford.edu) Stanford said United States and Chinese models have traded places at the top multiple times since early 2025. In February 2025, DeepSeek-R1 briefly matched the top United States model, and by March 2026 Anthropic’s top model led by 2.7%. (hai.stanford.edu) The report separates model performance from the rest of the artificial intelligence race. Stanford said the United States still produces more top-tier models and higher-impact patents, while China leads in publication volume, citations, patent output, and industrial robot installations. (hai.stanford.edu) The talent map is shifting too. Stanford said the number of artificial intelligence researchers and developers moving to the United States has fallen 89% since 2017, including an 80% drop in the last year alone. (hai.stanford.edu) That change tracks a broader geographic spread in who can train, hire, and keep artificial intelligence workers. Stanford’s research and development chapter says the field now spans publications, patents, models, data centers, and talent across more countries, even as gender gaps in artificial intelligence talent have shown “no meaningful progress” since 2010. (hai.stanford.edu) The shift is happening as use moves from experiment to routine. Stanford said generative artificial intelligence reached nearly 53% population-level adoption within three years, faster than the personal computer or the internet. (hai.stanford.edu) Companies are moving at the same pace. Stanford’s economy chapter says 88% of organizations used artificial intelligence in 2025, up from 55% a year earlier, and says the gains and disruptions are showing up unevenly across sectors and workers. (hai.stanford.edu) Governments are reacting by trying to build more domestic capacity. Stanford’s policy chapter says “artificial intelligence sovereignty,” meaning more control over local chips, data, and compute, has become a central national policy goal, but the infrastructure remains unevenly distributed. (hai.stanford.edu) Stanford’s broader conclusion is not that one country has won. It is that the gap between what artificial intelligence systems can do and what institutions can measure, govern, and absorb is widening as the competition spreads. (hai.stanford.edu)