Satlyt, founded by a former Google and SpaceX product manager, raises $8M to run AI on satellites
TechCrunch Tim Fernholz ● Covered by 2 sources
A former Google and SpaceX PM just raised $8M to run AI on satellites. Satlyt wants to turn spare space hardware into a shared cloud, not build its own spacecraft.
Based on reporting by TechCrunch, Tim Fernholz — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Rama Afullo says he saw orbital data centers coming before the rest of the industry did. He had tried to sell the idea inside Google and SpaceX, got no both times, and then left SpaceX to build Satlyt, a company split between Sunnyvale and Nairobi. The pitch is simple enough: don’t launch a whole new satellite just to do compute. Make the ones already up there share the work.
Satlyt has now raised $8 million in seed funding, with Houston-based Non Sibi Ventures leading the round. The company is not trying to follow SpaceX, Google, Starcloud, or Cowboy Space Company into building spacecraft. It wants to sell software for satellites instead, something Afullo compares to VMware and Snowflake: infrastructure the customer doesn’t have to think about. His more pointed version is that if SpaceX is the iPhone of orbital data centers, Satlyt wants to be Android.
The company has already flown its software on two demonstration missions. Its first target is operational efficiency, which is the kind of unglamorous problem that tends to matter most. Satellites still lean on controllers on the ground when something goes wrong, and sending data back down is slow and expensive. Move more of that work onboard, and the savings can be real. Earlier this year, Satlyt put Google DeepMind’s Gemma model on a spacecraft operated by Momentus and cut one transmission about onboard software errors by more than 60%.
This week, Satlyt’s software is set to launch on a SpaceX rocket aboard a spacecraft built by TakeMe2Space, an Indian company that makes computing hardware for satellites. Three customers are on that mission: NASA, which is testing cloud-computing protocols in space; Stellerian, which wants to test image-processing workloads for space surveillance; and TakeMe2Space itself, which wants to prove its satellite can host other companies’ software. Satlyt’s next step is bigger: linking two different satellites into a shared computing system next year.
That’s the real bet here. Not a fantasy of giant data centers in orbit, but a platform that becomes useful as more satellites go up and more of them carry GPUs or similar processors. Afullo says he wants Satlyt on 20% of satellites by the end of the decade. Ambitious, yes. But at least it’s a plan that doesn’t require waiting for science fiction to become a launch manifest.
My take — AI-written commentary, not fact-checked reporting
This is the sane version of the space-compute hype cycle: software first, hardware later, if ever. The industry loves a shiny orbital cathedral, but the boring middle layer usually wins the money. Satlyt sounds less like a moonshot and more like the sort of plumbing space companies will grudgingly pay for when the bills start arriving.
Read more about this at: TechCrunch