China moves to build national computing network for AI era
BEIJING — China is accelerating work on a national computing power network, a sprawling digital infrastructure project intended to connect data centers, artificial intelligence processors and energy resources as the country seeks to strengthen the foundations of its AI economy.
The National Development and Reform Commission, China’s top economic planning agency, said Thursday that computing power has become a key productive force in the digital economy and that an integrated national network is central to high-quality growth. Li Chao, a commission spokesperson, said market forces would play a decisive role, while government agencies would improve coordination and provide policy guidance.
By the end of March, China’s intelligent computing capacity had reached 1.882 million PFlops, 2.5 times the level recorded a year earlier, according to the commission. The rapid expansion reflects surging demand from artificial intelligence companies, cloud platforms, manufacturers and local governments seeking access to high-performance computing for model training, industrial automation and public services.
The effort builds on China’s “east data, west computing” strategy, which seeks to shift heavy data processing from the crowded and power-hungry eastern seaboard to western and inland regions with cheaper land, more renewable energy and room for large server farms. Under that plan, officials have called for closer coordination among eastern, central and western regions, as well as better integration of computing power, data and algorithms.
Officials have described the network as a kind of public utility for the AI age: computing capacity that can be scheduled, traded and delivered across regions much as electricity is dispatched across a grid. State-linked reports say real-time monitoring systems are intended to track distribution, workloads and utilization, allowing resources to be allocated more efficiently and helping avoid wasteful duplication of data centers.
Chip limits and power demand pose challenges
The program also has a strategic dimension. China faces continuing U.S. restrictions on access to advanced AI chips and is trying to reduce reliance on foreign semiconductor suppliers. Recent reports have said Beijing is preparing a multiyear plan worth about 2 trillion yuan, or roughly $295 billion, to finance AI data centers and strengthen the national computing network.
State-backed telecommunications operators, including China Mobile and China Telecom, are expected to play a major role in running the infrastructure. The broader goal is to connect scattered computing facilities into a more cohesive grid, making it easier for companies and public agencies to draw on capacity across regions rather than relying only on local data centers.
That linkage between computing and energy planning is central to Beijing’s pitch. Western regions such as Inner Mongolia, Gansu and Ningxia have abundant wind and solar resources, while coastal technology hubs have intense demand for cloud and AI services. Moving workloads westward could ease pressure on megacities and make fuller use of renewable generation, though latency-sensitive applications may still need computing close to users.
The obstacles are substantial. Advanced AI data centers require not only electricity and fiber connections but also powerful processors, high-bandwidth memory, cooling systems and software capable of scheduling jobs across different facilities. Domestic chipmakers are under pressure to meet rising demand, especially for the most demanding model-training workloads, as Beijing encourages greater use of Chinese hardware.
China is not alone in confronting the energy implications of artificial intelligence. Around the world, rapid growth in data-center demand is forcing governments and utilities to rethink power grids, transmission planning and clean-energy deployment. For Beijing, the national computing power network is meant to address those pressures while strengthening technological self-sufficiency. Whether it can deliver affordable, efficient and secure computing at national scale will help shape China’s position in the global AI race.
