地面算力能源瓶颈
Search documents
全球竞逐太空数据中心,张善从院长即将深度拆解算力如何“上天”
Xin Lang Cai Jing· 2025-12-03 13:37
Core Insights - The global consensus is forming around relocating data centers to space, with Beijing planning to build a centralized large-scale data center system in the 700-800 km dawn-dusk orbit, featuring over 1 GW of power capacity [1][4][12] - Major tech companies, including Google and SpaceX, are also pursuing space-based data centers to overcome terrestrial power and resource limitations [2][9][12] Group 1: Space Data Center Developments - Beijing's data center will consist of three subsystems: space computing, relay transmission, and ground control, with each space data center expected to host around 1 GW of power and accommodate millions of servers [1][4] - Google's "Project Suncatcher" aims to deploy solar-powered AI data centers in space, with plans to launch experimental satellites by 2027 [2][8] - SpaceX is expanding its next-generation Starlink satellites to support space data centers, while Axiom Space is developing a data center unit for the International Space Station [9][12] Group 2: Energy Consumption and Challenges - According to Gartner, global data center electricity consumption is projected to surge from 448 TWh in 2025 to 980 TWh by 2030, with AI-optimized servers' consumption increasing nearly fivefold [3][10] - The International Energy Agency (IEA) predicts that data center and AI energy consumption will double by 2026, raising concerns about local power supply and environmental impacts [11] - The unique characteristics of the dawn-dusk orbit allow satellites to receive nearly continuous sunlight, significantly enhancing solar energy generation compared to ground-based systems [4][11] Group 3: Implementation Timeline and Technical Challenges - The construction of Beijing's space data center is planned in three phases: from 2025 to 2027, focusing on energy and cooling technology breakthroughs; from 2028 to 2030, on in-orbit assembly and cost reduction; and from 2031 to 2035, on large-scale satellite production and deployment [5][12][13] - Significant engineering challenges include addressing heat dissipation in a vacuum environment and ensuring the longevity of sensitive components like GPUs through radiation protection [13]