Core Viewpoint - The event focused on the construction of data infrastructure and its role in supporting the digital economy and technological innovation, emphasizing the need for a new infrastructure system that prioritizes data value release, network support, and computing power as a foundation [1][14]. Group 1: Event Overview - The "Data Infrastructure Theme Exchange Activity" was held in Guiyang, gathering representatives from government, industry, academia, and research to discuss the construction path of data infrastructure and the national integrated computing power network [1]. - Key officials, including Liu Liehong, the head of the National Data Bureau, highlighted the goal of creating a new infrastructure system aimed at high-quality standards, large-scale facility deployment, and market-oriented ecological operations [1]. Group 2: Regional Economic Performance - In the first half of 2025, Guizhou's GDP grew by 5.3%, with 49 key data centers under construction or in operation, achieving over 90% in intelligent computing and domestic production rates [2]. Group 3: Technological Innovations - Chen Jianfeng from the Chinese Academy of Engineering discussed innovations in green data centers, projecting that China's total computing power electricity consumption will exceed 300 billion kWh by 2024, with a significant reduction in energy consumption through new cooling technologies [3]. - The "cloud-edge computing" concept proposed by Wang Huaimin aims to connect dispersed computing resources through a collaborative mechanism, enhancing operational efficiency across over 900 computing centers [4]. Group 4: Industry Insights - Yu Xiaohui from the China Academy of Information and Communications Technology emphasized the need to address connectivity and circulation issues in data infrastructure, with the National Data Bureau's "three unifications" laying the groundwork for data connection [6]. - Xu Qiang from the National Information Center proposed a "three fives" strategy for computing power network construction, focusing on inclusive computing, green development, and addressing supply and service efficiency challenges [7]. Group 5: Monitoring and Evaluation - Hu Jianbo from the National Data Development Research Institute introduced a monitoring system with "four dimensions, three levels, and 60 indicators" to ensure effective management and utilization of data infrastructure [9]. Group 6: Local Practices and Achievements - Jiang Yang, Deputy Director of Guizhou Data Bureau, shared five initiatives that led to the establishment of 49 data centers and significant improvements in computing power and latency [11]. - Wang Wanjun from Jiangsu Data Bureau reported that the province's computing power reached 61.54 Eflops, with the digital economy's core industry contributing 11.8% to GDP [12]. Group 7: Key Outcomes - The event announced three significant outcomes: the establishment of a national integrated computing power network monitoring and scheduling platform, the second batch of pilot projects for national data infrastructure, and successful case studies demonstrating cross-domain integration [13][14].

Core Viewpoint - The National Data Bureau of China has launched the "National Integrated Computing Power Network Monitoring and Scheduling Experimental Verification Platform" at the 2025 China International Big Data Industry Expo, aimed at integrating dispersed computing power across the country to provide more accessible computing services [1] Group 1: Platform Overview - The platform has achieved initial monitoring and scheduling of eight major hubs and ten clusters under the "East Data West Computing" initiative, covering over 900 computing facilities nationwide [1] - The platform facilitates the integration of domestic computing power, idle computing power, and green computing power, significantly reducing rendering tasks for projects like the movie "Nezha 2" from 600 years to a few months [1] Group 2: Cost Implications - According to representative Lu Hongjun, computing power costs are crucial for small and medium-sized enterprises transitioning to "data + artificial intelligence" [1] - The platform is expected to reduce training costs by 25% and inference costs by 10% [1]

Core Insights - The 2025 China Computing Power Conference highlighted significant advancements in China's computing power industry, indicating a shift towards high-quality and large-scale development [1][2] - The construction of an integrated computing power network is essential for promoting computing power as a societal service, akin to water and electricity [1][4] Group 1: Computing Power Development - As of June 2025, China has 10.85 million standard racks in use and an intelligent computing power scale of 788 EFLOPS, with total storage exceeding 1680 EB [1][2] - The average Power Usage Effectiveness (PUE) of computing centers has decreased to 1.42, reflecting improved energy efficiency [1] - The report indicates that the number of large models released in China has reached 1509, ranking first globally [2] Group 2: Integrated Computing Power Network - The conference announced the completion of access for 10 provincial platforms to the national computing power platform, marking a significant step towards a unified computing power network [3] - The national supercomputing internet platform has connected over 30 major computing centers, enhancing resource sharing and operational efficiency [3][4] - The platform aims to improve the utilization efficiency of computing resources, with an average increase of 40% since its launch [4] Group 3: Future Directions and Challenges - The Ministry of Industry and Information Technology plans to establish a comprehensive standard and rule system for computing power interconnectivity by 2026 [5] - Key challenges include technical issues related to heterogeneous connections and operational discrepancies among different computing centers [5][6] - The focus will be on developing efficient computing facilities, intelligent scheduling, and ensuring security and sustainability in the computing power network [6]

2025中国算力大会在山西大同盛大启幕，以"算网筑基，智引未来"为主题，标志着我国在算力领域的又一里程碑事件。会上，中国算力平台成功接入了包 括山西、辽宁、上海等在内的十个省区市分平台，实现了从平台到生态、从资源到场景的全面贯通。 近年来，我国算力产业发展迅猛，"十四五"期间更是取得了显著成就。智能算力规模达到了788EFLOPS，干线400G端口数量大幅增加至14060个，存力总 规模也突破了1680EB。同时，"华彩杯"算力应用创新项目征集活动反响热烈，累计征集项目超过2.3万个。 为应对算力产业发展中的挑战，我国以"点、链、网、面"为主轴，积极推动全国一体化算力网的建设。在中国信息通信研究院的引领下，自2017年起，产 业各界携手并进，稳步推进中国算力平台的建设。 作为国家级综合性算力服务平台，中国算力平台集算力供应、需求和服务于一体，致力于打造一个中立、开放、公平、可信的服务生态。通过提升不同系 统间的兼容性和互操作性，平台进一步促进了算力资源的高效配置与价值释放。 大会现场，栾晓维还发布了人工智能算力优化AICO体系，这一产学研共创的"算力加速引擎"旨在解决AI算力落地的痛点，让企业能够更便捷、经济 ...

Core Insights - The recent 2025 China Computing Power Conference highlighted significant advancements in China's computing power industry, emphasizing a transition to high-quality and large-scale development [1][2] - The Ministry of Industry and Information Technology announced plans to enhance the quality of computing resources and promote the construction of green data centers [2][3] Industry Developments - As of June 2023, China has 10.85 million standard racks in use for computing power centers, with an intelligent computing power scale of 788 EFLOPS and a total storage capacity exceeding 1680 EB [1] - The average Power Usage Effectiveness (PUE) of national computing power centers has decreased to 1.42, indicating improved energy efficiency [1] Infrastructure and Policy - The Ministry will guide the construction of computing facilities and optimize the layout of computing infrastructure across regions [2] - The "National Supercomputing Internet" initiative aims to create an integrated computing network, connecting various computing resources and applications [3] Technological Innovations - The National Supercomputing Internet has already connected over 30 backbone computing centers and attracted more than 640,000 registered users, offering over 1,600 applications and 600 popular open-source models [3] - China Mobile introduced the Computing Network Brain 3.0, which enhances system capabilities from automation to intelligence and supports various AI services [4]

Group 1 - The 2025 China Computing Power Conference was held in Datong, Shanxi Province, focusing on new trends in computing power networks and the future of the computing power industry [1] - As of June 2023, China has 10.85 million standard racks in use for computing power centers, with an intelligent computing power scale of 788 EFLOPS and a total storage capacity exceeding 1680 EB [1][2] - The average energy utilization efficiency (PUE) of national computing power centers has decreased to 1.42, indicating improvements in energy efficiency [1] Group 2 - The Ministry of Industry and Information Technology aims to enhance the quality of computing power resource supply and optimize the layout of computing power infrastructure [2] - The national supercomputing internet has connected over 30 backbone node computing centers, with more than 200 members in the consortium [2][3] - The supercomputing internet is designed to integrate computing power supply, application development, operation services, and user capabilities into a unified network and service platform [3] Group 3 - The national supercomputing internet has provided over 1,600 application software, more than 600 popular open-source models, and over 70 SaaS platforms, attracting over 640,000 registered users [3] - China Mobile showcased its Computing Network Brain 3.0, which supports various intelligent scheduling layers and offers simplified interaction experiences [4] - The Computing Network Brain has developed over 20 task-oriented services, widely applied in key areas such as energy, government affairs, and transportation [4]

扬子晚报/紫牛新闻记者万承源 校对陶善工 据介绍，在算网基础设施建设上，基于国家一体化算力网总体布局，将进一步发挥南京市"未来网 络""信息高铁"等技术优势，在确定性算力网络、高速数据网络、算力协同调度建设等方面开展相关合 作，构建南京-成都-拉萨高速直连网络，实现南京市与拉萨市算力协同调度。 在算力产业发展上，以西藏绿电算力服务长三角地区市场需求，共享算力、数据、算法、应用等大市 场，在国家数据跨境安全管理制度框架下，探索开展跨境业务合作，构建跨区域产业融合新体系。 在数字人才培养上，发挥南京市"宁数行"数字素养提升行动品牌效应，双方研究建立数字人才联合培养 机制，依托南京市高校科研资源，开展数字专业人才联合培养，服务全国一体化算力网建设。 近日，在第九届未来网络发展大会上，南京市数据局与拉萨市数据局签署协议《城市算力网"结对子"合 作协议》，在算网基础设施建设、算力产业发展、以及数字人才培养等方面将开启全面深入合作。 ...

Core Viewpoint - The IEEE standard for sharing data and models for artificial intelligence across multiple computing centers has been officially released, led by Pengcheng Laboratory, which aims to enhance interoperability, security, and efficiency in cloud collaboration among distributed computing centers [1] Group 1: Standard Development - The standard is titled "IEEE Standard Requirements and Framework for Sharing Data and Models for Artificial Intelligence across Multiple Computing Centers" with the designation IEEE Std 3404 -2025 [1] - The development involved collaboration with top research institutions such as Peking University and various leading companies including China Unicom, China Mobile, ZTE Corporation, and Huawei Technologies [1] Group 2: Impact and Scope - The standard addresses the challenges of interoperability, security, and efficiency in AI data and model sharing across multiple computing centers [1] - Pengcheng Laboratory has already developed over 15 standards related to computing networks, with initiatives recognized by various international and domestic standard organizations [1]

Group 1 - The core viewpoint of the article is the official release of the IEEE standard for sharing data and models for artificial intelligence across multiple computing centers, led by Pengcheng Laboratory [1] - The standard, identified as IEEE Std 3404-2025, aims to address interoperability, security, and efficiency issues in cloud collaboration among wide-area computing centers [1] - The development of this standard involved collaboration with top research institutions and leading companies, including Peking University, Huawei Technologies, and ZTE Corporation [1] Group 2 - Pengcheng Laboratory has already developed over 15 standards related to computing networks, which have been initiated and published in various international and domestic standard organizations [1] - The collaborative effort includes participation from organizations such as the National Data Standardization Technical Committee and the AITISA Alliance [1]

Core Viewpoint - The establishment of the national integrated computing network has progressed from planning to implementation, with the release of nine technical documents aimed at standardizing the network's operation and management [1] Group 1: Technical Document Release - The National Data Standardization Technical Committee has solicited public opinions on two technical documents related to the national integrated computing network, completing the release of nine technical documents [1] - The nine technical documents cover various aspects such as computing resource management, multi-dimensional billing, and operational services, marking a significant step in the network's development [1] Group 2: Implementation and Future Steps - The release of these documents lays the technical foundation for a unified national computing resource system, facilitating efficient monitoring and scheduling [1] - The National Data Bureau plans to accelerate the implementation of these technical documents to create a user-friendly, green, and secure integrated computing network, supporting the construction of a digital China and a unified national market [1]