China's AI computing power jumps 2.8x on 4 trillion yuan plan

Key Takeaways: China's national intelligent computing capacity tripled in a year as the state plans roughly 4 trillion yuan in new computing-network investment.

China's national intelligent computing power reached 2.8 times last year's level by end-June, with the National Development and Reform Commission planning about 4 trillion yuan in new computing-network investment during the 15th Five-Year Plan period.

"China's pace of independent AI innovation has accelerated, achieving rapid breakthroughs across the entire industry chain," an NDRC spokesperson said at the July 31 press conference.

The first fully homegrown 100,000-card AI supercluster is now in operation, and open-source models from DeepSeek and Moonshot AI have surpassed 10 billion global downloads. The NDRC said the investment will be enterprise-led, expanding room for private capital, with a batch of integrated-circuit projects spanning design, equipment, manufacturing and packaging set to break ground at up to 100-billion-yuan scale.

The plan dwarfs China's current AI infrastructure spending of $7.8 billion in the first quarter, per IDC, and reflects a state-directed push to cut reliance on foreign chips as the US dominates the global market at $67.9 billion.

A State-Led Push to Close the Chip Gap

China's computing build-out is accelerating as the government funnels capital into domestic silicon. The 4 trillion yuan figure, roughly $560 billion, compares with IDC's forecast of $1.21 trillion in global AI infrastructure spending by 2030, a scale that shows how much of the world's build-out Beijing intends to command.

The 100,000-card supercluster, built entirely with domestic components, marks a milestone in China's effort to reduce dependence on Nvidia's accelerators, which remain restricted by US export controls. Domestic models are being rapidly adapted to homegrown computing chips, the NDRC said, a step that could shift the competitive balance in AI hardware. Over 120,000 high-quality datasets have been built, and AI-related sectors are growing at more than 30 percent, the commission said.

Cross-Sector Spillovers and the Fiber Bottleneck

The build-out is rippling through adjacent industries. AI demand from intelligent computing centers has driven fiber optic prices sharply higher in 2026, with upstream optical fiber preform expansion cycles of 18-24 months leaving supply tight and top players locking in orders through 2028, according to industry reports. Hollow-core and ultra-low-loss G.654E fiber are now treated as rigid demand for computing backbones.

China's complete industrial system is generating new crossover sectors, the NDRC said. Artificial intelligence combined with physical entities is forming embodied intelligence, while AI entering life sciences is changing drug development. Each intersection, the commission argued, can grow into a new industry — a dividend unique to China's full supply chain.

For investors, the direction is clear: Chinese semiconductor, AI infrastructure and domestic chip names stand to benefit from the state-directed spending. The 4 trillion yuan plan, spread across the 15th Five-Year Plan, implies sustained demand for domestic accelerators, networking gear and optical components. Yet China's AI infrastructure spending grew just 9.3 percent in the first quarter versus the US's 30.3 percent, per IDC, suggesting the ramp has room to accelerate — and that the biggest beneficiaries may still be ahead.

This article is for informational purposes only and does not constitute investment advice.