With the official release of the National Development and Reform Commission, China's first national 100000 card artificial intelligence super cluster was put into operation at the Zhengzhou core node of the national supercomputing Internet. This milestone event marks the official entry of China's computing infrastructure construction into a new stage of 100000 card level deployment. From scientific computing to intelligent computing, from large model training to cutting-edge scientific research exploration, this super computing power network with peak computing power per second equivalent to 200 years of continuous computing for all mankind is injecting strong momentum into China's digital economy.

The first super cluster is officially put into use

At the core node of the national supercomputing Internet in Zhengzhou, the first national 100000 card artificial intelligence super cluster was put into operation. The uniqueness of this cluster lies in the fusion of scientific computing and intelligent computing, which provides a forward-looking layout for potential new computing needs in the future. Bu Jingde, Chief Engineer of High end Computing at the National High Performance Computer Engineering Technology Research Center, stated that the peak computing power of the new computing cluster per second is equivalent to the continuous computing of the entire human race for 200 years. Currently, it is supporting over 300 computing tasks in 26 fields such as new materials, innovative drugs, and artificial intelligence. At the national hub nodes of the Guangdong Hong Kong Macao Greater Bay Area, a new domestically produced super node Wanka cluster has also been launched simultaneously, directly targeting the intelligent computing power needs of the new generation of large model research and development. Liu Zhijun, the person in charge of the planning and construction of China Telecom's Guangdong Hong Kong Macao Greater Bay Area integrated data center, introduced that domestic large-scale models are moving from billions of parameters to trillions of parameters. The newly launched intelligent computing cluster can compress the training time of trillions of parameter large-scale models from one year to about six months, while providing a full chain inspection and testing environment for large-scale models.

The computing power channel accelerates the implementation

The expansion of computing power has driven the continuous improvement of the overall scale of China's computing power network. As of the end of June 2026, the scale of intelligent computing power in China has increased by 177% year-on-year, reaching 218.5 trillion floating point operations per second. Faced with the uneven distribution of computing power resources, the construction of a national integrated computing power network is accelerating to achieve unified monitoring and scheduling of computing power resources. The research team of Pengcheng National Laboratory is developing a new generation of optical fiber, which is as thick as a human hair and has four internal channels visible under a 60x microscope. The transmission capacity can be increased by more than four times, effectively breaking through the capacity bottleneck of ordinary optical fibers. Next, the research team will take the lead in creating a national level computing power dispatch channel with ultra large bandwidth and a length of thousands of kilometers between Shenzhen and Guizhou through low latency and optical direct connection, named the Shenzhen Guizhou Line, to accelerate the flow of computing power resources between computing hubs and artificial intelligence application gathering places. Keywords: high-tech news and information, computing power, artificial intelligence

Trillion level investment space opens up

Fragmented computing resources will be accelerated for integration, improving the overall utilization rate of the country's computing facilities. The national integrated computing power network is accelerating its formation, centered around the eight major national hub nodes of East Digital West Computing, forming a millisecond level computing power service circle in the main regions of the east and west of the country. The new national integrated computing power network monitoring and scheduling experimental verification platform has entered trial operation. Over 60% of the country's computing power has been included in unified monitoring, and the ability to access and accurately match computing power continues to improve, serving hundreds of application scenarios such as artificial intelligence, industrial manufacturing, and scientific research. Guo Mingjun, Director of the Computing Economy Department of the National Information Center of the National Development and Reform Commission, pointed out that during the 15th Five Year Plan period, the construction of the computing power network will increase direct investment by 4 trillion yuan, mainly through social investment. Focusing on key areas such as computing power monitoring and scheduling, computing power collaboration, and large-scale intelligent computing clusters, it will create broad investment space for various business entities, continuously empower the digital and intelligent transformation of thousands of industries, and help promote the high-quality development of China's digital economy.Editor/Gao Xue