Share with your CTO
The AI infrastructure crunch is pushing serious engineering minds toward an unlikely solution: orbital data centres that trade Earth’s resource constraints for space’s radiation, debris, and latency problems. SpaceX has already unveiled its AI1 Compute Satellite as a proof-of-concept, while the IEA warns that AI-driven electricity demand from terrestrial data centres is set to surge dramatically this decade. The pitch is real solar abundance and no zoning fights, but the engineering realities, football-field-sized radiators, no hardware refresh cycles, multi-second latency, push the viable near-term use case toward edge processing for satellites, not replacing hyperscale cloud.
What this means for your business
If your infrastructure roadmap depends on expanding hyperscale compute capacity in the next three to five years, space data centres belong in the “monitor, don’t build toward” category for now. The constraint driving this conversation is genuine: terrestrial data centres face compounding pressure from energy costs, community resistance, and cooling limits. But the companies most exposed to that constraint are the hyperscalers building at gigawatt scale, not enterprises consuming cloud capacity. The real signal here is that the supply problem is severe enough to generate genuinely exotic engineering proposals.
The thermal physics are worth understanding because they expose the core trap. On Earth, cooling relies on convection and evaporation, moving heat into air or water. In vacuum, the only escape route is radiating heat as infrared energy, which requires enormous surface area. The 10-megawatt figure in the engineering literature, roughly a modest data centre’s heat output, demands radiator structures spanning two football fields. That’s not a detail to engineer around; it’s a structural cap on orbital compute density that no materials breakthrough will eliminate quickly. Space computing will be sparse and specialized for a long time.
The realistic near-term winner is edge processing for space-native workloads, analyzing Earth observation imagery on-orbit, supporting intelligence and scientific missions, running compute for satellite constellations. That’s a valuable and growing market, but it’s orthogonal to enterprise AI infrastructure planning. CTOs who track this story as a hedge against terrestrial capacity constraints are watching the wrong variable. The leading indicator to watch instead is whether terrestrial data centre energy costs and permitting delays actually force hyperscaler pricing up, because that pressure, not orbital ambition, is what would change your infrastructure budget equation in the next renewal cycle.
Based on reporting from Space data centres: Can orbiting AI infrastructure solve Earth’s computing crisis?, originally published 2026-07-17 17:17:00.
