China's intelligent computing hits ~2,185 EFLOPS, targeting 9,800 by 2030
ingliguori · x · 2026-09-24
China is scaling AI compute infrastructure aggressively: intelligent computing capacity reached roughly 2,185 EFLOPS in mid-2026, with a 2030 target of 9,800 EFLOPS—about 4.5× growth in a few years.
The expansion supports more training, inference, robotics workloads, industrial AI, and agentic systems. The author argues the AI race is becoming an infrastructure race, and questions whether compute capacity will prove as strategically important as model quality.
More from Infra
- NVIDIA's DGX Spark Appears Unavailable, May Never Return to Sale — GabGarrett · 2026-09-24
- Not every AI task needs an LLM: 'decide' may become a standard model call — bigdata · 2026-09-24
- Prime Intellect launches Prime Sandboxes: MicroVM sandboxes purpose-built for RL training — xeophon · 2026-09-24
- M5 Ultra 96GB local inference: 3,200 tok/s aggregate prefill over 112M tokens with custom MLX server — Every-Fortune-3151 · 2026-09-24
- 3D DRAM slated for 2029: bandwidth of 3x HBM5E at sub-1pJ/bit — zephyr_z9 · 2026-09-24
- Brain vs H100 physical ledger: 80B transistors against 86B neurons — yaroslavvb · 2026-09-24