SpaceX could capture orbital compute
Starcloud
This is a race to define the buying bundle, not just the cost per GPU hour. If SpaceX reaches customers first with launch, satellite hardware, networking, and compute sold together, orbital compute could look like a SpaceX product category before Starcloud has enough missions to prove that its own economics work at commercial scale. That matters because early government and enterprise buyers often standardize on the first full stack that is usable, financeable, and already in orbit.
-
Starcloud is still pre commercial, with no disclosed revenue as of May 2026, Starcloud-2 framed as the first commercial mission, and broader cost competitiveness tied to launch costs falling toward about $500 per kilogram around 2028 to 2029. That leaves a window where the market can form before Starcloud proves repeatable unit economics.
-
SpaceX has the strongest bundle because it already controls the rocket, the satellite manufacturing base, and a live communications network through Starlink. The FCC accepted for filing SpaceX's application for an orbital data center system of up to one million satellites, explicitly tied to optical inter satellite links, which turns compute into an extension of infrastructure it already operates.
-
Blue Origin shows how category capture can happen even without the best compute design. Blue Ring already packages communications, power, data storage, and edge and AI computing for hosted payloads, and TeraWave is aimed at enterprise, data center, and government connectivity with deployment beginning in Q4 2027. For some buyers, good enough hosting plus trusted procurement is enough.
The category is heading toward a platform market where bundled procurement beats standalone hardware. Starcloud's path is to prove that orbital compute is not just technically possible, but commercially repeatable before larger incumbents make in space compute feel like a feature of their broader network and launch stack.