Workflow
寒武纪AI加速卡
icon
Search documents
2025人工智能计算中心发展白皮书2.0
Sou Hu Cai Jing· 2025-11-28 02:30
Core Insights - The report "2025 Artificial Intelligence Computing Center Development White Paper 2.0" emphasizes the importance of AI computing centers as a new type of infrastructure for digital economy development, highlighting their role in global technological competition and industrial upgrading [1][8][9] - The document outlines the transition from standalone AI computing centers to interconnected AI computing networks, which are necessary to meet the growing demands for computing power, data sharing, and ecological collaboration [1][9][10] Global AI Computing Center Development - Major countries and regions, including China, the US, Europe, and Japan, are actively promoting the construction of AI computing centers as part of their strategic initiatives [8][19] - The US plans to invest approximately $69 billion in AI and quantum computing research and development over the next five years, aiming to triple the number of AI research institutions by 2025 [14][15] - The EU has allocated nearly €100 billion for research and development, with €2.4 billion specifically for AI and high-performance computing infrastructure projects [15][19] Characteristics of China's AI Computing Centers - China's AI computing centers are characterized by government-led planning, adherence to independent technology routes, market-oriented operations by enterprises, and application-driven development [1][10][30] - The report highlights successful projects such as Shenzhen's Pengcheng Cloud Brain II and Wuhan's AI Computing Center, which utilize domestic AI chips like Huawei's Ascend and Cambricon's products [1][30][23] New Demands and Challenges - The development of AI computing centers faces new challenges, including the increasing demand for computing power driven by large models and big data, as well as the need to align with national carbon reduction goals [9][10][30] - The report suggests that the traditional model of standalone computing centers is insufficient, advocating for a shift towards interconnected AI computing networks to enhance resource utilization and reduce energy consumption [1][9][10] Recommendations for High-Quality Development - The report recommends coordinated high-quality construction of AI computing centers, orderly deployment of computing networks, exploration of market-oriented operational mechanisms, and strengthening of core technology research and standardization [1][10][30] - It emphasizes the importance of breaking down barriers to the flow of computing power, data, and algorithms to support the deep integration of AI with the real economy and promote high-quality digital economic development [1][9][10]