Sou Hu Cai Jing·2025-08-24 09:46Core Insights - The 2025 China Computing Power Conference in Datong marks a significant milestone in the country's computing power sector, themed "Building Foundations with Computing Networks, Guiding the Future with Intelligence" [1] - The conference highlighted the rapid development of China's computing power industry, achieving a scale of 788 EFLOPS in intelligent computing power and a total storage capacity exceeding 1680 EB during the 14th Five-Year Plan period [1] - The China Computing Power Platform has successfully integrated ten provincial sub-platforms, enhancing the overall ecosystem and resource allocation [1] Group 1 - The China Computing Power Platform aims to create a neutral, open, fair, and trustworthy service ecosystem, enhancing compatibility and interoperability among different systems [4] - As of July, the platform has registered over 1,000 enterprise users and onboarded more than 100 computing service providers, offering over 110 high-quality computing power products [4] - The platform has achieved "five major integrations" involving platform, entities, resources, ecosystem, and scenarios [4] Group 2 - Datong ranks third nationally in computing power index, positioning itself as a key player in transforming resource-based economies [5] - The city has invested 28.2 billion yuan in its computing power center, with a construction scale of 25,000 P, equivalent to over 326,000 standard racks, achieving an overall shelving rate of 90% [5] Group 3 - China Telecom announced its AI computing power ecosystem at the conference, revealing plans to establish Datong as a "new heart of computing power" [7] - The company has built a full liquid-cooled domestic intelligent computing cluster in the Beijing-Tianjin-Hebei and Yangtze River Delta regions, with a total computing power exceeding 77 EFLOPS [7] - China Telecom's new initiatives include the "Xirang" computing power scheduling system and the launch of the AI computing optimization AICO system, aimed at facilitating easier and more economical access to AI computing power for enterprises [7]