The country's first "four-in-one" computing network scheduling platform built by China Mobile has been officially put into use recently.

"Four-in-one" refers to a computing system that integrates four computing capabilities: general computing power, intelligent computing power, super computing power, and quantum computing power.

It is understood that the computing power scheduling platform can support hundreds of millions of computing power calls per day, can schedule 8/0 of the country's computing power scale, and improve the efficiency of computing network integration by 0%. The chip localization rate of the platform's own intelligent computing center exceeds 0%, and it is compatible with 0 kinds of domestic AI chips, which is of great significance to ensure the security of the supply chain and promote the high-quality development of the national digital economy.

Lou Tao, Deputy General Manager of China Mobile Cloud Competence Center: We have built the world's largest cloud backbone private network, covering more than 5 prefectures and cities, and the latency of prefectures and cities is 0 milliseconds, and the latency in the province is 0 milliseconds.

In recent years, relevant state departments have issued a series of policy documents to accelerate the construction of a national integrated computing power network and promote the high-quality development of computing infrastructure, promote the integrated layout of general computing power, intelligent computing power and super computing power, and vigorously promote the systematic development of computing power networks.

Fang Xinping, Deputy Director of the Information Development Bureau of the Cyberspace Administration of the People's Republic of China: Strengthen the integrated development of multiple computing power such as general computing, intelligent computing, and supercomputing, further optimize the structure of computing resources, improve the utilization level of computing power, and reduce the utilization cost of computing power.

Artificial intelligence promotes the upgrade from "cloud computing" to "cloud intelligent computing".

Experts pointed out that the recent rapid development of artificial intelligence is triggering structural and deep-seated changes in computing network infrastructure.

Under this trend, the core carrier of the computing network has been upgraded from the traditional "cloud computing" to the "cloud intelligent computing" with the deep integration of cloud and AI.

This is the first one-stop online shopping agent in China. Its interface is similar to many large models we use daily, if the enterprise has the need to order computing power, as long as the enterprise scale and specific application requirements are entered in the dialog box, you can intelligently formulate the computing power configuration and purchase plan, and quickly find the right computing power resources, and the resource matching efficiency is 100% higher than the traditional ordering model.

Dong Ruoyun, a technician: We can accurately match the reasonable and appropriate resource specifications and resource types that customers really need, and can avoid waste of resources.

This robot dog that is interacting with reporters has applied quantum computing technology, and it can perform 22 intelligent interactive actions such as dancing and push-ups, and can also talk directly to people.

Experts pointed out that the vigorous development of artificial intelligence large models has promoted the optimization and reconstruction of computing power layout. In the next 40 years, the scale of China's smart computing power will grow by more than 0.0 times, with an average annual compound growth rate of nearly 0%. The computing power structure will also change significantly, and the inference computing power demand will exceed the training computing power demand.

Zheng Weimin, Academician of the Chinese Academy of Engineering: In the past, it was mainly for large-scale model training, but from now on, the inference computing machine will gradually surpass the training computing machine. By year 2028, the computing power of inference will exceed the computing power of training.

AI + computing power drives the digital transformation of thousands of industries

At present, "AI + computing power" has been widely used in various industries such as urban governance, medical and health, and people's livelihood services, and has become a new engine to drive the digital transformation of society.

In the centralized control hall of the second power plant in Shanghai, the reporter saw that technicians were developing and testing a "large-scale model platform for all scenarios of thermal power". Technicians told reporters that traditional thermal power is facing transformation pressures such as carbon emissions, rising costs, and new technology iterations. Artificial intelligence technology can improve the production and operation efficiency of power plants through data analysis, intelligent prediction, optimization control and other means, and help power plants achieve cost reduction, efficiency increase, energy conservation and emission reduction.

In this hospital in Hangzhou, doctors have begun to use AI doctor assistants to record and analyze cases, which will automatically accurately record and classify relevant information, greatly improving the efficiency of treatment and diagnosis. The AI doctor assistant will upload all the patient's medical records and examination image data to the Zhejiang Health Cloud Platform for real-time viewing, and it can also realize data exchange between medical institutions at all levels. Even if the patient is seeking medical treatment in a different place, the doctor can easily access the patient's previous medical information, and the efficiency of mutual recognition of inspection and examination has jumped from minutes to seconds.

支撐健康雲平臺的是華東地區規模最大的數據中心——杭鋼集團的雲計算數據中心，規劃超2萬個機櫃，配備30多萬台伺服器。數據中心的算力除了支撐健康雲以外還同時為氣象雲等公共服務雲提供服務。專家指出，隨著高水平演算法、高性能算力、高品質數據的持續投入，AI整體能力將實現指數級增長，與此同時，AI在更大範圍、更廣領域、更深層次的應用也將大幅降低使用成本。AI任務將成為算力網路基礎設施承載的主要內容，到2030年在全網流量中的佔比將達到64%。

(Reporter Sun Jiwei, Fang Liang, Jin Jian, Nie Qixing, Yang Dunhuang)

Editor/Wang Haozhou