SpaceX and Google explore orbital data centers

Google interest would turn orbital compute from a science project into a real enterprise procurement category. SpaceX already controls the rocket, the satellite factory, and the in orbit network through Starlink, and the xAI combination gives it a built in first customer for GPU capacity. A Google test deployment matters because it suggests a second path to revenue, selling compute capacity to outside cloud buyers instead of only feeding Musk’s own AI stack.1 2 3 4

• The immediate value is demand validation, not scale. Reporting in May 2026 said Google was discussing launching test products for orbital data centers with SpaceX. That points to prototype workloads first, which is how a new infrastructure category usually gets de risked before large contracts follow.1
• SpaceX is trying to bundle the whole stack. Internal research describes orbital compute as a third business line next to launch and Starlink, with xAI as captive demand. That means every customer dollar could touch launch fees, satellite manufacturing, network transport, and rented compute time.2 3 4
• The closest pure play comparison is Starcloud. It is focused only on orbital GPU hosting and had $200M of estimated funding as of March 2026. SpaceX can attack the same market with a far larger wedge because it already has launch capacity, in house satellites, and an installed internet constellation.4 5

The next step is a small number of pilot satellites and narrow workloads, then a push to win hyperscaler and sovereign AI contracts if launch costs keep falling. If Google becomes an early tenant, SpaceX moves closer to being valued not just as a launch company or connectivity provider, but as a new kind of cloud infrastructure owner.1 2 3