杭州将加大AI基础设施建设 构建新型算力体系

Core Viewpoint - Zhejiang Province has issued the "Implementation Opinions on Promoting the Innovative Development of Intelligent Computing Cloud (2025-2027)" to enhance the high-quality development of artificial intelligence and establish a new computing power system, aiming to create a leading computing power industry [1] Group 1: Development Goals - By 2027, the scale of intelligent computing cloud services provided through public cloud is expected to reach 60 EFlops, with breakthroughs in technologies such as super-large-scale computing clusters [1] - The initiative aims to cultivate 100 benchmark enterprises in intelligent computing cloud and establish over 300 typical application scenarios, positioning Zhejiang as a global hub for intelligent computing cloud technology, services, applications, and ecosystems [1] Group 2: Key Technology Development - Zhejiang will accelerate the research and development of key technologies for intelligent computing cloud, focusing on major projects related to AI chips, high-performance networks, and AI acceleration platforms [2] - The province aims to achieve international leadership in technologies such as domestic GPU integration and scheduling, as well as cloud database technologies [2] Group 3: Industry Applications - The initiative encourages the use of intelligent computing cloud in key industries such as manufacturing, healthcare, culture, transportation, finance, energy, and education, promoting the integration of AI and cloud computing [3] - It supports the development of over 50 general and vertical AI models through a "public cloud + API" model [3] Group 4: Ecosystem and Talent Development - Zhejiang will support the establishment of a competitive ecosystem for intelligent computing cloud, encouraging local enterprises to adopt AI solutions and collaborate with cloud computing companies [3] - The plan includes strengthening talent supply by incorporating intelligent computing cloud talent into the list of scarce talents and promoting educational initiatives in this field [3]