17.1M
▲ 2.5× since Q4 2024
H100-equivalents of AI computing capacity, Q4 2025; Nvidia 66.7%
Q1 2022Q4 2025
What the chart shows
The computing capacity of AI chips sold by the major designers reached about 17.1 million H100-equivalents by the fourth quarter of 2025, up from 0.28 million at the end of 2022. The AI Index estimates growth of about 3.3 times a year since 2022.
Nvidia chips account for 11.4 million H100-equivalents, two-thirds of the total. Google’s TPUs add 3.1 million, Amazon’s chips 1.9 million and Huawei’s 0.7 million. AMD reached 1.0 million by the third quarter of 2025.
Key numbers
| Measure | Value |
|---|---|
| Total, Q4 2025 | 17.07M |
| Nvidia, Q4 2025 | 11.38M |
| Google, Q4 2025 | 3.10M |
| Amazon, Q4 2025 | 1.88M |
| Huawei, Q4 2025 | 0.71M |
| AMD, Q3 2025 | 1.02M |
How to read it
- An H100-equivalent expresses computing power as the number of Nvidia H100 chips that would deliver it.
- Estimates are inferred from revenue, financial disclosures and analyst reports, so they show trends rather than exact counts.
- Coverage varies: Nvidia and Google from 2022, others from 2024. AMD has no data for Q4 2025, so that quarter’s total leaves AMD out.
Source: Epoch AI, 2026, via the AI Index Report 2026 (Stanford HAI), figure 1.2.3: Global computing capacity from AI chips across major designers, 2022–25. Chart and text by AI Adoption.
Numbers checked against the published data on 19 October 2026.
Download the data (CSV) · Cite this chart
Numbers checked against the published data on 19 October 2026.
Download the data (CSV) · Cite this chart
Discussion