AI 資料中心移至外太空？

• Power Generation: Space-based solar arrays require approximately 1 square mile per gigawatt at 30% cell efficiency; sun-synchronous orbits positioned at dawn/dusk transition optimize continuous solar exposure[3] • Thermal Management: Elimination of water cooling requirements; autonomous robotics and in-space servicing systems required for maintenance without costly human missions[4] • Data Transmission: Laser-based communication replacing fiber optic cables to reduce processing latency and improve data relay speeds[4] • Launch Economics: Cost-effectiveness threshold estimated at $200/kg to low Earth orbit; dependent on SpaceX Starship scaling to 180 launches/year by 2035[3] • In-Space Assembly: ISAM (In-Space Servicing, Manufacturing, and Assembly) technologies reduce pre-launch preparation complexity and enable orbital infrastructure maintenance[4] • Radiation Hardening: Edge computing satellites designed with radiation-hardened components for space environment resilience[2]

The emergence of orbital AI infrastructure represents a potential multi-trillion dollar hardware supercycle driven by terrestrial power constraints[2]. However, industry consensus indicates meaningful commercial viability remains 10+ years away despite near-term stock market enthusiasm. The convergence of reusable launch economics, AI power demands, and government space policy (White House Space Executive Order) is creating enabling conditions for orbital compute as a complementary rather than replacement infrastructure layer. Companies are positioning enabling technologies (launch services, autonomous systems, thermal management) as near-term investment opportunities before full orbital data center viability. Environmental sustainability claims require validation—space-based solutions may prove more sustainable long-term than terrestrial alternatives if manufacturing and launch impacts are optimized, but this remains speculative pending operational deployment.