China’s AI Compute Growth Relies on US Chips and Offshore Cloud Access

China's rapid expansion in AI computing power is heavily dependent on U.S.-designed chips and access to offshore cloud services, according to a recent report.

The Institute for Progress (IFP) projected that total compute available to Chinese entities will increase from approximately 4.6 million to 8.8 million H100-equivalents by the end of 2027. This metric is based on Nvidia's data-centre graphics processing units.

China's remote access to AI compute is anticipated to rise to 2.8 million H100-equivalents by the end of 2027, up from about 1 million this year. The IFP report included data from domestic chips, legal imports, smuggled processors, and remote cloud access.