China targets fourfold boost in AI computing capacity by 2030 in major tech push
China's MIIT announced a five-year plan targeting 9,800 eflops of intelligent computing capacity by 2030, requiring more than a fourfold increase from the current ~2,450 eflops The plan calls for orderly deployment of massive AI computing clusters, including facilities with 10,000+ GPU cards and clusters with 100,000+ accelerator cards 3.8 trillion yuan (US$532 billion) in cumulative infrastructure investment is planned for 2026–2030, building on the "East Data, West Computing" initiative launch
Analysis
TL;DR
- China's MIIT announced a five-year plan targeting 9,800 eflops of intelligent computing capacity by 2030, requiring more than a fourfold increase from the current ~2,450 eflops
- The plan calls for orderly deployment of massive AI computing clusters, including facilities with 10,000+ GPU cards and clusters with 100,000+ accelerator cards
- 3.8 trillion yuan (US$532 billion) in cumulative infrastructure investment is planned for 2026–2030, building on the "East Data, West Computing" initiative launched in 2022
- Chinese-designed AI chips captured 41% of domestic shipments in 2025, though fragmented software ecosystems and compatibility issues remain significant adoption barriers
- The plan also advances next-generation communications, including 6G trials in the 6GHz band, 5G-Advanced rollout, and research into space-based computing
Why It Matters
China's aggressive AI infrastructure expansion signals a strategic commitment to competing globally in AI compute capacity, directly impacting the availability and cost of training and inference resources worldwide. The scale of investment—nearly half a trillion dollars over five years—will reshape the global AI hardware supply chain and accelerate demand for both domestic and international chip manufacturers. For AI practitioners, this means China could become a dominant force in AI development while simultaneously creating new competitive pressures and potential supply constraints.
Technical Details
- Computing capacity targets: China aims to reach 9,800 eflops by 2030 from ~2,450 eflops as of July 2025, representing a ~4x increase. Daily AI token usage surged from ~100 billion in early 2024 to 140 trillion by March 2025.
- Cluster specifications: The plan mandates "orderly deployment" of intelligent computing clusters with 10,000+ graphics processing cards and clusters with 100,000+ cards, alongside application-specific inference computing facilities.
- Infrastructure distribution: Over 80% of intelligent computing capacity is concentrated in eight national hubs under the "East Data, West Computing" framework, which relocates power-intensive workloads to western regions with cheaper land and abundant energy.
- Resource aggregation: A government-backed platform has consolidated ~316 eflops from 155 companies and 578 resource pools, encompassing roughly 720,000 GPUs, reducing fragmentation across previously isolated computing resources.
- Domestic chip adoption: Chinese-designed AI chips accounted for 41% of domestic shipments in 2025, with the plan explicitly calling for greater infrastructure adaptation to home-grown computing chips despite ongoing software ecosystem and compatibility challenges.
- Communications infrastructure: China operates ~5.1 million 5G base stations (69% of global total), with 5G-Advanced (5G-A) already deployed in 330+ cities. The 6GHz band was approved for 6G trials, and 3GPP Release 21 (first global 6G standard) is targeted for completion in 2029.
Industry Insight
- The sheer scale of China's compute expansion will intensify global demand for AI accelerators, potentially constraining supply for international competitors and driving up hardware costs—companies should secure chip procurement strategies well in advance.
- China's push for domestic chip adoption despite software fragmentation signals a long-term bet on self-sufficiency; AI practitioners working with Chinese partners should prioritize compatibility testing and invest in ecosystem adaptation tools early.
- The convergence of massive compute clusters with next-generation communications (5G-A, 6G, space-based computing) creates opportunities for edge AI, real-time inference at scale, and new distributed computing architectures that blend terrestrial and satellite networks.
Disclaimer: The above content is generated by AI and is for reference only.