Loading
Loading
compute
SpaceX is becoming an AI-infrastructure company: reusable launch (Starship) drives $/kg-to-orbit toward the point where space-based compute pencils out, and the announced AI1 satellite is the first spacecraft designed as a data center.
If launch cost keeps falling, the binding constraint on AI compute shifts from terrestrial grid queues and cooling to launch mass and heat radiation. SpaceX is the only firm that controls both the rocket and the satellite, so it sets the pace for whether orbital compute is real — and AI1 shows it is willing to run silicon hot and radiators edge-on to make the physics close.
AI1's 70 m solar span is about a Boeing 747's wingspan (~68 m) — and the radiator that cools its 150 kW payload is the thin cross in the middle, not the wings. Dimensions: SpaceX (AI1 & Starlink published specs); illustration by Peregrinations.
AI1 is built as an orbital data center — a ~150 kW compute payload whose hardest engineering problem is dumping heat in a vacuum. The radiator, not the rocket, is the binding constraint. Run the orbital-vs-ground economics → or read why we put compute in orbit →
Announced orbital AI-compute satellite: ~120 kW average / 150 kW peak payload cooled by a 110 m² edge-on radiator. The first satellite built as a data center — explore its economics in the orbital-compute tool.
Fully reusable heavy-lift vehicle; the lever that drives launch cost toward the $/kg range where orbital compute starts to compete with the ground.
LEO broadband constellation; the cash-flow engine and the satellite-manufacturing muscle (lasers, mass production) that AI1 is built on top of.
The workhorse reusable launcher and today's ~$2,700/kg-to-LEO benchmark the orbital-compute economics start from.
Sets SpaceX's launch-cost-first strategy and the AI1 orbital-compute bet, tied to xAI's compute demand and Tesla's silicon and energy work.