算力产业

Search documents
中国算力平台全面贯通
Xin Hua She· 2025-08-23 10:01
2025中国算力大会开幕式现场。新华社记者 柴婷 摄 作为国家级综合性算力服务平台,中国算力平台集"供、需、服"于一体,可实现不同系统、平台和工具 之间的兼容性和互操作性,致力于为算力供需各方打造"中立、开放、公平、可信"的服务生态,进一步 促进算力资源的高效配置与价值释放,让算力服务真正实现普惠化、便捷化、易用化。 在产业各方的协同攻坚下,中国算力平台成功实现平台贯通、主体贯通、资源贯通、生态贯通、场景贯 通"五大贯通"。 截至7月底,中国算力平台运营层注册企业用户超1000家,入驻算力服务商逾100家,上架优质算力产品 110余项,接入主流基础大模型和垂类模型90余个,成功承载9000余张离散卡的供应汇聚、购买和即时 开通服务,为1000余名开发者提供了多源多元的在线调用服务,累计沉淀数十亿条算力监测大数据。 (记者柴婷) 8月23日,以"算网筑基 智引未来"为主题的2025中国算力大会在山西大同开幕。开幕式上,中国算力平 台正式完成山西、辽宁、上海、江苏、浙江、山东、河南、青海、宁夏、新疆10个省区市分平台接入工 作,实现"平台、主体、资源、生态、场景"全面贯通。 "十四五"期间,我国算力产业发展迅速, ...
工信部将开展算力赋能专项行动,涉及教育医疗能源等行业
Di Yi Cai Jing· 2025-08-23 08:50
深入开展算力强基"揭榜"行动,聚焦计算、存储、网络等重点方向。 作为数字经济时代的新质生产力,算力产业发展备受市场关注。 23日,以"算网筑基 智引未来"为主题的2025中国算力大会在山西省大同市开幕。 工业和信息化部副部长熊继军在致辞中表示,工信部将加强与相关部门协同,持续加强多元泛在、智能敏捷、安全可靠、绿色低碳的算力基础设施建设,切 实提升算力资源使用效率,推动算力网络"点、链、网、面"体系化高质量发展,持续巩固提升信息通信业竞争优势和领先地位。 他表示,工信部将深化算力赋能行业应用,推动构建算力产业良好生态。面向教育、医疗、能源等重点行业,开展算力赋能专项行动。推动"智算生态圈"建 设,推广一批标志性技术产品和解决方案,推动实施城域"毫秒用算"专项行动,加快建设中国算力平台。 工信部最新数据显示,截至今年6月底,我国在用算力中心标准机架达1085万架,智能算力规模达788 EFLOPS,干线400G端口数量大幅增加至14060个,存 力总规模超过1680EB,全国算力中心平均电能利用效率(PUE)降至1.42,算力基础设施规模和水平不断提升。 熊继军强调,将有序引导算力设施建设,切实提升算力资源供给质 ...
工信部将开展算力赋能专项行动 涉及教育医疗能源等行业
Di Yi Cai Jing· 2025-08-23 08:49
作为数字经济时代的新质生产力,算力产业发展备受市场关注。 23日,以"算网筑基智引未来"为主题的2025中国算力大会在山西省大同市开幕。 工业和信息化部副部长熊继军在致辞中表示,工信部将加强与相关部门协同,持续加强多元泛在、智能 敏捷、安全可靠、绿色低碳的算力基础设施建设,切实提升算力资源使用效率,推动算力网络"点、 链、网、面"体系化高质量发展,持续巩固提升信息通信业竞争优势和领先地位。 他表示,工信部将深化算力赋能行业应用,推动构建算力产业良好生态。面向教育、医疗、能源等重点 行业,开展算力赋能专项行动。推动"智算生态圈"建设,推广一批标志性技术产品和解决方案,推动实 施城域"毫秒用算"专项行动,加快建设中国算力平台。 熊继军强调,将有序引导算力设施建设,切实提升算力资源供给质量。推动完善算力布局政策体系,优 化布局算力基础设施,引导各地合理布局智能算力设施,持续开展国家绿色数据中心建设。加快突破 GPU芯片等关键核心技术,扩大基础共性技术供给。深入开展算力强基"揭榜"行动,聚焦计算、存储、 网络等重点方向,加快新技术新产品应用推广。 工信部最新数据显示,截至今年6月底,我国在用算力中心标准机架达1085 ...
中国智能算力规模达788 EFLOPS
Zhong Guo Xin Wen Wang· 2025-08-23 06:53
2025中国算力大会举行1场开幕式、2场主论坛,聚焦算力产业发展政策,展示算力发展最新成果,研讨 算力未来发展趋势。 开幕式上,中国算力平台(山西)启动仪式和中国算力平台贯通仪式举行。《2025综合算力指数》《2025 算力发展报告》《2025存力发展报告》《2025运力发展报告》《2025智算服务发展报告》等研究报告发 布。(完) (文章来源:中国新闻网) 中新社大同8月23日电 (记者刘育英)8月23日,以"算网筑基智引未来"为主题的2025中国算力大会在山西 省大同市开幕。记者从会上获悉,中国智能算力规模达788 EFLOPS(每秒百亿亿次浮点运算)。 中国工业和信息化部副部长熊继军在开幕式上表示,中国算力产业迈向高质量、规模化发展的新阶段。 截至今年6月底,中国在用算力中心标准机架达1085万架,智能算力规模达788 EFLOPS,存力总规模超 过1680 EB(2^60字节),全国算力中心平均电能利用效率(PUE)降至1.42,算力基础设施规模和水平不断 提升。 熊继军表示,工信部有序引导算力设施建设,推动完善算力布局政策体系,持续开展国家绿色数据中心 建设;强化企业创新主体地位,加快突破关键核心技 ...
2025中国算力大会:截至今年6月底智能算力规模达788EFLOPS
Huan Qiu Wang· 2025-08-23 04:34
【环球网科技综合报道】8月23日,以"算网筑基智引未来"为主题的2025中国算力大会在山西省大同市 开幕。工业和信息化部副部长熊继军、山西省副省长汤志平出席开幕式并致辞。大同市委副书记、市长 刘俊义主持开幕式。 熊继军在致辞中指出,工业和信息化部扎实推进算力基础设施建设,不断强化产业科技创新能力,持续 推动算力应用落地,我国算力产业迈向高质量、规模化发展的新阶段。截至今年6月底,我国在用算力 中心标准机架达1085万架,智能算力规模达788 EFLOPS,干线400G端口数量大幅增加至14060个,存 力总规模超过1680EB,全国算力中心平均电能利用效率(PUE)降至1.42,算力基础设施规模和水平不 断提升。 熊继军强调,工业和信息化部将加强与相关部门协同,持续加强多元泛在、智能敏捷、安全可靠、绿色 低碳的算力基础设施建设,切实提升算力资源使用效率,推动算力网络"点、链、网、面"体系化高质量 发展,持续巩固提升信息通信业竞争优势和领先地位。有序引导算力设施建设,切实提升算力资源供给 质量。推动完善算力布局政策体系,优化布局算力基础设施,引导各地合理布局智能算力设施,持续开 展国家绿色数据中心建设。强化企业 ...
工信部:有序引导智能算力基础设施 适度超前动态平衡
Zhong Guo Zheng Quan Bao· 2025-08-22 20:14
Group 1 - The 2025 China Computing Power Conference was held in Datong, Shanxi from August 22 to 24, focusing on optimizing national computing power layout and promoting the development of intelligent computing infrastructure [1] - As of March this year, China's total computing power ranks second globally, with 10.43 million standard racks in use and an intelligent computing power scale of 748 EFLOPS, supporting AI development and high-quality economic growth [1] - A report released on August 22 highlighted that the rapid growth of data production is not matched by the expansion of storage resources, with a projected national data annual output of 41.06 ZB in 2024 but only 2.09 ZB of total storage capacity [1] Group 2 - Current challenges in computing power include inefficient resource scheduling and underutilization of GPU performance, with new business forms requiring ultra-low latency and high throughput [2] - Shanxi province is leveraging the "East Data West Computing" strategy to transform its economic growth model from coal and electricity export to computing power and services [2] - Shanxi's computing center currently has 514,000 standard racks and an intelligent computing power scale of 32 EFLOPS, with an average PUE of 1.2, ranking among the best in the country [2]
工信部:有序引导智能算力基础设施适度超前动态平衡
Zhong Guo Zheng Quan Bao· 2025-08-22 20:09
我国算力总规模已经位居全球第二,智算比例不断提升。记者从工信部了解到,截至今年3月,我国在 用算力标准机架达1043万架,智能算力规模达748EFLOPS,为人工智能发展提供了坚实支撑,有力促 进了经济社会高质量发展。 中国算力大会首次在山西举办。作为资源大省,山西当前正抢抓"东数西算"战略机遇、以算力产业为核 心引擎推动经济增长方式由"输煤炭、输电力"向"输算力、输服务"转变。山西省工信厅副厅长刘勇表 示,山西正努力打造京津冀外溢算力需求和算力产业的最佳承载地,构建"通算+智算+超算"多元协同 的算力供给体系。目前,全省在用算力中心机架规模达到51.4万标准机架,智算规模达到32EFLOPS, 算力中心平均PUE仅1.2,多项指标稳居全国前列。 ● 本报记者 杨洁 存力方面,8月22日发布的《先进存力中心研究报告(2025年)》称,数据质量和数据规模是AI发展的 关键,但当前数据生产量的快速增长与存储资源扩充速度并不匹配,数据"应存未存"的现象依然存在。 2024年全国数据年产量达41.06ZB,但存储总量仅为2.09ZB,建议构建存算协同新模式,助力数据要素 价值深度释放,推动AI强国建设。 8月22日至 ...
大同:“算”出一片新天地
Sou Hu Cai Jing· 2025-08-22 03:39
Core Viewpoint - Datong City is leveraging its geographical, climatic, and energy advantages to develop its computing power industry, aiming to transform from a resource-based economy to a data-driven economy, with significant investments and strategic planning in place to support this transition [6][8][11]. Group 1: Industry Development - Datong has been recognized as one of the seven national-level data labeling bases, ranking third in the country for urban computing power index, highlighting its growing significance in the AI industry [6]. - The city has established a comprehensive computing power ecosystem, covering hardware manufacturing, data center construction, cloud services, and data applications, creating a closed-loop system [11]. - Since the establishment of the Qinhuai Energy Information Technology Industrial Base in 2018, Datong has attracted major projects, with total investments in computing power centers exceeding 28.2 billion yuan, and over 736,000 servers operational [11][13]. Group 2: Infrastructure and Technological Innovation - Datong's computing power industry is supported by advanced infrastructure, including two high-end server production lines with an annual capacity exceeding 100,000 units, and a national-level data labeling base leading the way in data services [13]. - The city has been recognized for its green computing initiatives, with innovations such as "waterless cooling" technology that reduces energy consumption significantly [11][14]. - Datong's computing power centers are designed to utilize renewable energy, with over 56% of its total power capacity coming from renewable sources, aiming for over 60% by 2027 [18]. Group 3: Strategic Planning and Future Outlook - The local government has implemented a three-year action plan for data industry development, focusing on integrating computing power with data and industry applications to enhance economic growth [21][22]. - Datong aims to create a multi-layered computing power supply system and explore cross-regional computing power scheduling, contributing to the national integrated computing power network [22]. - The city is committed to fostering a collaborative ecosystem involving government, academia, and industry to cultivate data talent, with plans to train at least 3,000 data professionals annually [15].
算力产业上市公司上半年业绩频报喜
Zheng Quan Ri Bao· 2025-08-21 16:37
Core Insights - The demand for AI computing power is surging, leading to significant growth in the performance of companies within the computing power industry [1][4] - By 2025, China's intelligent computing power is projected to reach 1037.3 EFLOPS, with expectations to grow to 2781.9 EFLOPS by 2028 [1] Company Performance - Hongjing Technology Co., Ltd. reported a revenue of 1.188 billion yuan for the first half of 2025, a year-on-year increase of 567.14%, with net profit reaching 60.28 million yuan, up 725.73% [2] - Haiguang Information Technology Co., Ltd. achieved a revenue of 5.464 billion yuan, a 45.21% increase year-on-year, and a net profit of 1.201 billion yuan, growing by 40.78% [2] - Zhongji Xuchuang Co., Ltd. expects a net profit between 3.6 billion to 4.4 billion yuan for the first half of 2025, reflecting a year-on-year growth of 52.64% to 86.57% [3] - Chengdu Xinyi Communication Technology Co., Ltd. anticipates a net profit between 3.7 billion to 4.2 billion yuan, with a year-on-year increase of 327.68% to 385.47% [3] - Shenzhen Yingweike Technology Co., Ltd. reported a revenue of 2.573 billion yuan, a 50.25% increase, and a net profit of 216 million yuan, up 17.54% [3] Industry Trends - The AI application proliferation is driving explosive growth in computing power demand, benefiting companies across the computing power supply chain [1][4] - The demand for key infrastructure such as optical modules is surging, particularly for CPO technology, which reduces data center power costs and enhances transmission efficiency [2] - The rapid development of AI technology is increasing the demand for data center computing power, with liquid cooling technology becoming a mainstream choice for heat dissipation [3]
“数聚江宁 数领未来”论坛在宁举办
Nan Jing Ri Bao· 2025-08-21 00:00
Group 1 - The forum "Data Gathering Jiangning, Leading the Future" was held in Jiangning, where the "Three-Year Action Plan for Optimizing Data Infrastructure Construction to Promote High-Quality Development of the Digital Economy (2025-2027)" was released [1] - The Action Plan aims to establish a complete digital economy industrial chain by 2027, focusing on three digital industry clusters: smart grid, information communication, and future networks, along with two data-featured industrial parks [1][2] - The plan includes 15 tasks across four areas: optimizing computing power supply systems, promoting efficient data circulation, applying AI large models, and creating demonstration application scenarios [1][2] Group 2 - The Action Plan emphasizes the development of demonstration scenarios in smart cities, smart governance, and smart living, aiming to enhance data resource aggregation and sharing [2] - Jiangning has 443 enterprises in the core digital economy sector, achieving nearly 90 billion yuan in output value in the first half of the year, indicating a growing digital industry cluster [2] - The "Data Computing Network Intelligent Integration Ecosystem" was launched to unify the scheduling and supply of computing resources, significantly lowering the barriers and costs of data computing [3]