China targets 9,800 EFLOPS of AI computing capacity by 2030, a fourfold boost
pstAsiatech · x · 2026-09-09
China's MIIT released a five-year plan targeting 9,800 EFLOPS of intelligent computing capacity by 2030 — more than quadruple the 2,185 EFLOPS recorded at end-June (up 177% YoY) — plus 3.8 trillion yuan ($532B) in cumulative information infrastructure investment over 2026–2030, SCMP reports.
The plan calls for 'orderly deployment' of clusters with 10,000 and even 100,000+ accelerator cards, application-specific inference facilities, and stronger adaptation of infrastructure to home-grown chips. China has already built 52 clusters with over 10,000 cards each; capacity hit 2,450 EFLOPS by end-July.
More from Infra
- Public cloud for AI can cost 10-20x more than on-prem, warns cloud expert — DavidLinthicum · 2026-09-09
- Scry: open-source self-hosted game streaming applies DLSS 5 to any game — kingroka · 2026-09-09
- GoatRemote Brings Siri Remote Voice + Trackpad Control to Linux, Running AI Locally in Under 0.6s — mayfer · 2026-09-09
- ASML wins commitments from top chipmakers for its latest production gear — pstAsiatech · 2026-09-09
- AI computing demand may never be sated, says Nvidia partner Iren's CEO — pstAsiatech · 2026-09-09
- Nvidia reportedly adopting copper-diamond composite materials for cooling — zephyr_z9 · 2026-09-09