Enterprise & Industry

Four things we’d need to put data centers in space

SpaceX filed to launch 1M orbital data centers, aiming to solve AI's energy and water crisis.

Deep Dive

A new space race is emerging, not for colonies, but for compute. MIT Technology Review details how tech giants like SpaceX, Amazon, and Google are seriously exploring orbital data centers to sustainably fuel the AI revolution. SpaceX's FCC application to launch up to one million orbital data centers highlights the scale of the ambition. The core promise is environmental: space offers constant solar power in sun-synchronous orbits and a cold vacuum to radiate excess heat, potentially solving AI's massive energy and water-cooling demands on Earth.

However, significant engineering hurdles remain. The first is thermal management. In the vacuum of space, heat can only be expelled via radiation, not convection, requiring massive radiator surfaces. Electronics in constantly illuminated orbits would also face baseline temperatures of 80°C, necessitating advanced cooling systems like the refrigerant-fluid pumps developed by Thales Alenia Space. The second major challenge is radiation. Earth's orbit is bombarded with cosmic particles that can corrupt standard computer chips, requiring new radiation-hardened hardware, as tested by startups like Starcloud with its Nvidia H100-equipped satellite.

Despite these obstacles, a 2024 feasibility study led by former Thales director Yves Durand concluded that Europe could deploy gigawatt-scale orbital data centers—rivaling the largest terrestrial facilities—before 2050. These would feature solar arrays larger than the International Space Station. With launch costs falling due to rockets like SpaceX's Starship, the economic and environmental calculus for moving compute heavenward is becoming a serious part of the industry's long-term roadmap.

Key Points
  • SpaceX filed an FCC application to launch up to 1 million orbital data centers, joining Amazon and Google in the concept.
  • Core benefits include unlimited solar power and natural radiative cooling, solving AI's energy/water crisis on Earth.
  • Major hurdles are heat radiation in a vacuum and developing radiation-hardened chips to survive constant cosmic particle bombardment.

Why It Matters

This represents a radical long-term solution to the unsustainable energy and environmental footprint of the global AI compute infrastructure.