SpaceX filed an FCC application on January 30 for up to 1 million satellites to function as solar-powered AI data centers in low Earth orbit. The proposal comes three days after the FCC approved 7,500 additional Starlink satellites and amid reports of merger discussions between SpaceX and Elon Musk's xAI.
The Real Numbers
There are currently 15,000 satellites in orbit. SpaceX has launched 11,000 Starlink satellites to date. The company previously requested approval for 42,000 satellites but deployed far fewer - a pattern regulators know well. The Verge describes the 1 million figure as "negotiation bait," not a serious deployment target.
FCC Chairman Brendan Carr has adopted a phased approval approach for mega-constellations, granting licenses in tranches while monitoring collision risks and debris accumulation. SpaceX will need to demonstrate viability at smaller scale first.
What This Means for Enterprise
The technical proposition: LEO satellites with continuous solar power could bypass terrestrial data center constraints around electricity and cooling. The pitch targets AI workloads requiring massive compute without ground infrastructure limits.
The practical challenges: Latency remains the fundamental trade-off. LEO networks add 25-35ms minimum versus terrestrial fiber. For financial trading systems or real-time applications requiring sub-10ms response times, physics dictates ground infrastructure. Multi-constellation routing and handover delays during satellite transitions compound latency issues.
Cooling in space presents another problem. Without atmospheric convection, heat dissipation relies on radiation - orders of magnitude less efficient than terrestrial HVAC. Blue Origin is developing radiation-hardened compute for government clients, suggesting the current focus is defense applications rather than general enterprise workloads.
What to Watch
Starship deployment capability matters more than satellite count. SpaceX expects the first Starship orbital payloads this year after 11 test launches since 2023. Until Starship proves reliable at scale, the data center proposal remains theoretical.
Amazon's Project Kuiper deployment delays (seeking extensions for 1,600+ satellites) demonstrate launch capacity constraints across the industry. The bottleneck is getting hardware to orbit, not FCC paperwork.
For CTOs evaluating space-based compute: the physics works better for batch processing and machine learning training than latency-sensitive production workloads. Keep watching the pilots. History suggests announcements run years ahead of implementations.