The performance gap between United States and Chinese artificial intelligence (AI) models has “effectively closed,” even as the United States maintains a strong lead in data center infrastructure and investment, according to a new report released April 13 by the Stanford Institute for Human-Centered Artificial Intelligence.
The report finds that AI top models from the two countries have traded the lead multiple times since early 2025, with the top U.S. model ahead by just 2.7% as of March 2026.
At the same time, the United States hosts 5,427 data centers – more than 10 times the number of any other country – and leads in private AI investment, which reached $285.9 billion in 2025, although its ability to attract global AI talent has declined.
The findings are part of the institute’s latest AI Index report, which tracks global trends in artificial intelligence and provides a broad view of how technology is evolving across countries, industries, and policy environments.
According to the report, the narrowing gap in model performance reflects intensifying global competition, with China leading in publication volume, citations, patent output, and industrial robot installations, while the United States continues to produce more top-tier models and higher-impact patents.
The report also underscores the scale of U.S. infrastructure advantages, noting that U.S. dominance in data centers and energy consumption plays a critical role in supporting advanced AI development, even as global supply chains remain dependent on chip manufacturing concentrated in Taiwan.
Despite these advantages, the report highlights challenges facing the United States, particularly a steep decline in its ability to attract global AI talent, with the number of researchers and developers moving to the country dropping significantly in recent years.
The report’s co-directors said the data reflects a field that is expanding rapidly but unevenly.
“This is a technology that has reached mass adoption faster than the personal computer or the internet,” said Yolanda Gil, one of the co-directors. “The data does not point in a single direction,” Gil said, adding, “It reveals a field that is scaling faster than the systems around it can adapt.”
Co-director Raymond Perrault said that while leading AI models are becoming increasingly similar, the tools used to measure them are struggling to keep pace.
“As models converge, the tools used to evaluate them are struggling to stay relevant,” he said.
Another one of the trends identified in the report is the rapid spread of AI adoption among consumers. Generative AI tech reached 53% population adoption within three years, with usage varying widely across countries and often correlating with economic factors such as gross domestic product per capita, the report says.
In the United States, consumers are already deriving substantial value from AI tools, with the report estimating annual benefits of $172 billion, the report says.
At the same time, AI-driven productivity gains are emerging in several sectors, particularly in customer support and software development, where improvements of 14% to 26% have been observed. However, these gains are not uniform across all types of work.
The report notes that in fields where productivity gains are strongest, entry-level employment is beginning to decline. In software development, employment among younger workers has fallen sharply even as demand for more experienced workers continues to grow.
Education systems are also struggling to keep pace with rapid adoption, the report says. While four out of five U.S. high school and college students report using AI for school-related tasks, only about half of schools have established policies governing its use.
Beyond education, the report identifies AI sovereignty as an increasingly important focus for governments, with countries investing in domestic capabilities and infrastructure to maintain control over AI development and deployment.
At the same time, the ability to produce advanced AI models remains concentrated in the United States and China, even as open-source development is expanding participation from other regions and enabling more diverse contributions.
Finally, the report highlights a growing divide between expert and public opinion on AI’s impact. While a majority of experts express optimism about the technology’s effects on jobs and the economy, public sentiment remains more cautious, and trust in governments to regulate AI varies significantly across countries.