China AI industry forecast sees 35% intelligent compute by 2026

Even without full access to the report, the headline numbers are revealing: Beijing expects China’s AI industry not just to keep growing quickly in 2026, but to shift its compute mix decisively toward ‘intelligent’ workloads. Crossing a 35% threshold for AI‑oriented compute signals that the country is committing a large chunk of scarce power, chips and capital to model training and inference rather than traditional IT. That reinforces the idea that, for China, AI is now a core industrial policy pillar on par with semiconductors and green energy. ([finance.people.com.cn](https://finance.people.com.cn/n1/2025/1221/c1004-40628651.html?utm_source=openai))

For the global race to AGI, this matters because hardware, not algorithms, is increasingly the binding constraint. Whoever controls the densest, cheapest and most reliable AI‑optimized compute gets to run the biggest experiments and deploy the widest fleets of agents. China’s push to expand intelligent compute—alongside initiatives like the “Artificial Intelligence+” action plan and national AI application bases—suggests it aims to offset Western export controls with massive domestic capacity and vertically integrated ecosystems. That raises the floor of global competition even if frontier model weights remain mostly in U.S. hands.

At the same time, a state‑driven build‑out of AI infrastructure carries familiar risks: over‑capacity, misallocation and regional imbalances in who can actually access this compute. For AGI‑oriented labs and investors outside China, the takeaway is that the competitive landscape will be shaped as much by national compute policy and energy planning as by clever model architectures.