🔥36氪•Freshcollected in 1m
China's AI Compute Market to Reach 1.44 Trillion by 2029
💡Understand the massive scale of China's AI infrastructure investment and its impact on the future of AI hardware.
⚡ 30-Second TL;DR
What Changed
Domestic AI compute procurement projected to reach 1.44 trillion yuan by 2029.
Why It Matters
This projection signals a long-term, sustained investment trend in domestic AI hardware, likely accelerating the development of local GPU and NPU alternatives. It suggests a shift toward massive infrastructure scaling for Chinese AI enterprises.
What To Do Next
Evaluate the roadmap of domestic AI hardware providers to align your infrastructure strategy with the projected supply chain growth.
Who should care:Founders & Product Leaders
Key Points
- •Domestic AI compute procurement projected to reach 1.44 trillion yuan by 2029.
- •High demand for computing power is driving massive capital investment in the sector.
- •Analysts expect the emergence of multiple trillion-dollar market cap companies in the AI supply chain.
🧠 Deep Insight
AI-generated analysis for this event.
🔑 Enhanced Key Takeaways
- •The 1.44 trillion yuan projection is largely driven by the 'East Data, West Computing' (东数西算) national strategy, which aims to optimize the distribution of computing resources across China.
- •Domestic procurement is heavily incentivized by US export controls on high-end GPUs, forcing Chinese cloud providers and enterprises to pivot toward domestic alternatives like Huawei Ascend and Cambricon chips.
- •The market growth is increasingly focused on 'heterogeneous computing' architectures, integrating NPUs, GPUs, and FPGAs to handle diverse AI workloads beyond standard LLM training.
- •Local governments in Tier-1 cities are establishing state-backed 'Computing Power Centers' (算力中心) that act as primary procurement vehicles for domestic AI hardware, bypassing traditional commercial procurement cycles.
- •Energy efficiency and liquid cooling technologies have become critical competitive differentiators for domestic AI hardware providers due to the high power density requirements of large-scale clusters.
📊 Competitor Analysis▸ Show
| Feature | Huawei Ascend (910B/C) | Cambricon (MLU Series) | NVIDIA (H20/China-Specific) |
|---|---|---|---|
| Architecture | Da Vinci | MLUv02/v03 | Hopper (Modified) |
| Ecosystem | CANN / MindSpore | Cambricon NeuWare | CUDA / TensorRT |
| Interconnect | Ascend Fabric | MLU-Link | NVLink (Restricted) |
| Market Position | Domestic Leader | Specialized AI Inference | Legacy/Compatibility |
| Pricing | Premium (High Demand) | Mid-Range | High (Scarcity Premium) |
🛠️ Technical Deep Dive
- Huawei Ascend 910 series utilizes a 3D-stacked chiplet architecture to improve memory bandwidth and power efficiency.
- Domestic AI cards are increasingly adopting HBM3 or HBM3e memory standards, though supply remains a bottleneck due to international trade restrictions.
- Software stacks like MindSpore and PaddlePaddle are being optimized to provide 'CUDA-to-CANN' migration tools to lower the barrier for developers switching from NVIDIA hardware.
- Implementation of RDMA (Remote Direct Memory Access) over Converged Ethernet (RoCE) is becoming the standard for scaling domestic clusters to thousands of nodes.
🔮 Future ImplicationsAI analysis grounded in cited sources
Domestic market share of Chinese AI chips will exceed 50% by 2027.
Aggressive state subsidies and the forced migration of state-owned enterprises to domestic hardware are creating a protected market environment.
Consolidation will reduce the number of major domestic AI chip vendors to fewer than five.
The high capital expenditure required for R&D and the need for massive software ecosystem support will drive smaller players out of the market.
⏳ Timeline
2022-10
US Bureau of Industry and Security implements strict export controls on advanced AI chips to China.
2023-02
China releases the 'Overall Layout Plan for the Construction of Digital China,' prioritizing computing infrastructure.
2024-08
Huawei begins large-scale shipments of the Ascend 910B, signaling a shift toward domestic high-performance compute.
2025-05
Major Chinese cloud providers announce that over 40% of their new AI cluster deployments utilize domestic silicon.
2026-03
The 14th Five-Year Plan mid-term review emphasizes 'Self-Reliance' in AI hardware as a national security priority.
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 36氪 ↗