China's MIIT is targeting 9,800 exaflops of AI compute by 2030 behind ¥3.8 trillion in spend.

The plan is more than a fourfold increase from the 2,185 eflops recorded at end of June 2026, itself up 177% year over year, with $532B in cumulative information infrastructure investment across 2026-2030. It explicitly funds inference facilities adapted for domestic chips alongside clusters of 10,000 GPUs or 100,000+ accelerator cards; 52 facilities with 10,000+ cards already exist. Related: Alibaba Cloud and Cambricon took Platinum seats on the PyTorch Foundation board, Ant Group went Gold. Platinum buys a Governing Board seat and a Technical Advisory Council seat, so Chinese accelerator vendors now have direct governance influence over PyTorch's device-agnostic backend work.