Data Insight
Nov. 18, 2026

Agentic inference capacity grew 4.6× in a year, versus 2.6× for compute capacity

Between Q2 2025 and Q2 2026, the estimated agentic inference capacity of AI chips sold to date grew 4.6×, compared with 2.6× growth in computing power. We measure agentic capacity using our Inference Performance Index (IPI), which compares hardware on a fixed set of workloads with long, cached contexts.

The growth is largely driven by the Hopper-to-Blackwell transition: Nvidia’s Blackwell and Blackwell Ultra families account for 74% of the added capacity, with GB300 systems delivering roughly 8× the inference capacity of H200s per GPU. We expect this pace to slow as quarterly additions add to an increasingly large base. Chips sold in Q2 2026 added only 3.3× the capacity of those sold in Q2 2025. Continued hardware–model co-design should nevertheless keep inference capacity growing faster than peak computing power.

Epoch's work is free to use, distribute, and reproduce provided the source and authors are credited under the Creative Commons BY license.

Learn more about this graph

The chart shows the estimated agentic inference capacity of AI chips sold to date in each quarter from Q2 2025 to Q2 2026, as a multiple of the Q2 2025 total, with bars broken down by chip designer and chip family. Lines show total inference capacity and peak 8-bit computing power, each relative to its own Q2 2025 level.

Download this data

Explore this data