Stanford HAI's 2026 AI Index Report tracks a steep rise in the computing capacity behind artificial intelligence. In the Research and Development chapter, a stacked bar chart shows AI chip computing capacity growing from 0.05 million H100-equivalents in the first quarter of 2022 to 17.07 million in the fourth quarter of 2025, an increase of more than 340 times in under four years.

Nvidia accounts for the largest share of that capacity throughout the period and supplies more than 60 percent of the total. Google and Amazon provide much of the remainder, and Huawei holds a small but growing share. The report attributes the buildout to hyperscaler data center expansion and sustained demand for frontier model training and inference.

The United States hosts 5,427 data centers, more than ten times the count of any other country, according to the report. AI data center power capacity rose to 29.6 gigawatts, a level comparable to New York state at peak demand. The report also notes that a single company, TSMC, fabricates almost every leading AI chip, and that a TSMC facility in the United States began to operate in 2025.

On model development, US institutions produced 59 notable AI models in 2025, compared with 35 from China. Open-source activity continues to grow, with GitHub AI projects reaching 5.58 million in 2025. US-based projects hold 30 million cumulative GitHub stars across projects that have passed the 10-star threshold. The AI Index also finds that the United States hosts more AI talent than any other country while attracting new talent at the lowest rate in over a decade.

Source: Stanford HAI AI Index - https://hai.stanford.edu/ai-index/2026-ai-index-report/research-and-development