China’s Computing Power Network Spurs New Industries
China’s national integrated computing power network is moving from blueprint to build-out, and the effort is opening fresh industrial opportunities across the artificial intelligence supply chain — from domestic AI chips and cooling equipment to a fast-emerging “token economy.” The push follows the Central Committee of the Communist Party and the State Council, which on October 9 issued the Opinions on Developing New Quality Productive Forces, explicitly calling for the construction of a national integrated computing power network and a national blockchain network.
A report published October 14 by the Economic Information Daily and carried by Xinhua described an emerging industrial loop — hardware at the foundation, platform enablement in the middle, and real-world scenario deployment downstream — as the network takes shape under a “one network, one chessboard, one integration” framework.
Context: From East Data to Integrated Network
The integrated network builds on the East Data, West Computing initiative launched in February 2022, which established eight national computing hubs to channel data processing toward the western regions where land and green energy are more abundant. That project is now anchored by ten major computing clusters that are collectively pushing the industry toward networked, standardized operation.
According to the Ministry of Industry and Information Technology, China’s intelligent computing capacity reached 2,185 EFLOPS (FP16) as of June 2026, up 177 percent year-on-year, with an overall rack utilization rate of 71.4 percent. More than 70 computing transmission channels connecting hubs and key regions have been built, and 17 national computing interconnection regional nodes have been approved. The China Computing Platform has achieved nationwide integrated computing monitoring, aggregating more than 10,000 registered enterprise users, over 2,000 listed computing products, and 300-plus connected large models.
A Cluster Built for Speed
Langfang, a Hebei city bordering Beijing, illustrates how tightly the network is being knit together. It has built 36 data centers and connected them to core Beijing nodes through an ultra-low-latency fiber link with one-way latency as low as 0.33 milliseconds and overall response times under 1 millisecond — allowing computing resources to serve the capital’s real-time scientific and internet demands.
“The national integrated computing power network is not a simple stacking of computing power, but connects computing power, electricity, data, models and scenarios into an intelligent economy infrastructure network,” said Li Mingyan, director of the Intelligent Economy Research Institute at the Saizhi Industry Research Institute, in remarks carried by Xinhua.
According to Li, the network can serve demand from AI model training, autonomous driving, smart finance, and the low-altitude economy, while cross-regional scheduling better matches western green power with eastern computing demand — lowering costs and promoting renewable energy consumption.
The Rise of the Token Factory
Perhaps the most novel industrial layer is the “token factory.” Tokens are the smallest units of information that AI models process, and as AI shifts from model competition toward agent-scale deployment, demand for them has exploded.
At the Beijing Yi Hao Token Factory in the Beijing Economic-Technological Development Area, banks of computing racks in “lights-out” machine rooms run continuously. The facility currently produces 300 to 500 billion tokens per day and has been adapted to connect with mainstream domestic models.
“Previously, the production, supply and use of tokens were mostly handled in closed loops within major model enterprises,” said Liu Li, a technical expert at iSoftStone Intelligent Computing, which operates the facility. “With the rise of the agent economy and accelerated computing network construction, market demand for tokens has risen exponentially, and independent third-party token factories have emerged.”
A report by 21jingji cited Liu as saying the factory can currently process 500 billion tokens a day, with a target of 1.4 trillion daily — equivalent to roughly 200 million ordinary AI question-and-answer sessions.
The China Academy of Information and Communications Technology recently released its Token Economy Development Research Report, showing that the ten East Data, West Computing clusters are advancing token factory construction. Zhangjiakou, Shaoguan and Qingyang already possess large-scale token production capacity, while five clusters including the Yangtze River Delta and Wuhu have completed platform service launch.
“New value carriers represented by tokens make computing measurable, divisible, priceable and tradeable,” said Li Wei, deputy director of CAICT’s Cloud Computing and Digitalization Research Institute. That new circulation system, she said, requires unified full-domain scheduling and cross-regional sharing — forcing computing construction to upgrade from building individual centers to networking into a system.
Analysis: Investment and Guardrails
Scale is driving major investment. China’s National Development and Reform Commission has specified that during the 15th Five-Year Plan period, computing network construction will add 4 trillion yuan (about $560 billion) in direct investment, with enterprise capital expected to dominate — creating significant room for private investment.
Industry analysts project the demand dynamics will be dramatic. China Telecom Research Institute forecasts annual token consumption could reach 10 billion billion in 2026 and exceed 350 billion billion by 2030, a compound annual growth rate near twelvefold. Computing demand is expected to grow roughly tenfold per year over the next two to three years.
Li Mingyan expects that large-scale construction of computing clusters and token factories will directly drive demand in domestic AI chips, liquid cooling equipment, high-speed optical modules, and green energy facilities, while intermediate services such as scheduling, model inference optimization, and token metering and settlement will spawn new platform enterprises.
Yet Beijing is signaling caution against overheating. The October policy document warned against “rushing headlong” and blind investment, and the NDRC has stressed that computing network layout must not proceed in a disorderly rush but should emphasize supply-demand alignment and coordinated computing-power and electricity planning.
What’s Next
With the policy framework now in place and investment targets set, the coming years will test whether China can convert raw computing scale into a genuinely tradeable, efficient market. The key question is whether the network can standardize pricing, scheduling and settlement fast enough to keep pace with token demand — while avoiding the overcapacity and bubbles that regulators have explicitly warned against.
For now, the industrial loop from computing clusters to token factories is widening, and the sectors that feed it — chips, optics, power equipment and platform services — are positioning for the next phase of the build-out.