We present key data on over 180 AI accelerators, such as graphics processing units (GPUs) and tensor processing units (TPUs), used to develop and deploy machine learning models in the deep learning era.

Our open database of AI chip imports, using global customs and domestic production data to estimate the geographic distribution of AI chips.

The computing power of AI chips sold each quarter has grown about 2.5× per year since 2024, down from a surge in 2023.

Our open database of AI chip sales, using financial reports, company disclosures, and more to estimate compute, power usage, and spending over time for a wide variety of AI chips.

Our explorer uses customs records to estimate national shares. Since 2022, 72% of AI compute has ended up in the US and about 10% in China, including chips China makes itself.

Our estimates of how much computing power the leading AI developers rent or use, measured in Nvidia H100-equivalents (H100e).

China is 2.7× as exposed to semiconductor supply shocks as the US, Epoch AI estimates. Epoch traces chips through input-output tables to model the impact of a Taiwan shock and decoupling.

Memory shipped through 2027 could run 33–171 million concurrent frontier-model agents, or billions using efficient open models. Epoch AI estimates inference capacity from HBM supply, serving benchmarks, and agent-hour costs.

Huawei plans major gains in AI chip performance by 2030, but US export controls cap its most important scaling levers. Epoch AI estimates Huawei will produce less than 4% as much AI compute as Nvidia in 2026, a share that could be around 1% by 2028 without access to foreign memory.

Epoch AI's analysis of trade data finds a pattern consistent with $3B worth of chips smuggled to China via Malaysia. China recorded $3.8B of server imports from Malaysia in 2024–25; Malaysia recorded only $0.6B of exports.