Satlyt raises $8M to run AI workloads on satellites

Rama Afullo pitched the idea of computing in orbit at both Google and SpaceX while working there, and both turned him down, he told TechCrunch. Afullo, a Kenyan American engineer, had worked in Google’s cloud computing business before a brief stint at SpaceX’s Starlink communications network in 2024. He left to co-found Satlyt, headquartered in Sunnyvale, California, and Nairobi, to build software that lets multiple satellites share the work of large computing jobs. The company has raised an $8 million seed round, Afullo exclusively told TechCrunch.

Unlike SpaceX, Google, or startups such as Starcloud and Cowboy Space Company, Satlyt isn’t building its own spacecraft. Instead, it’s building software to run AI models on other companies’ satellites, comparing its approach to VMware and Snowflake — platforms that let users run complex software without managing the underlying infrastructure. “The folks like SpaceX who are doing orbital data centers — if they are the iPhone, we’ll build Android as a horizontally integrated, open ecosystem,” Afullo said.

Satlyt has already flown its software on two demonstration missions, aimed at helping spacecraft resolve problems onboard instead of relying on ground-based flight controllers, since sending data back to Earth is expensive and slow. Earlier this year, Satlyt deployed Google DeepMind’s Gemma AI model onboard a spacecraft operated by Momentus; the model cut the size of a transmission about onboard software errors by more than 60%, which Afullo says can save hundreds of thousands of dollars per satellite each year. On Thursday, Satlyt’s software is set to launch on a spacecraft built by Indian startup TakeMe2Space, alongside the first prototype for Google’s orbital data-center effort, Project Suncatcher. The mission involves three customers: NASA, which is testing cloud-computing protocols in space; space-surveillance startup Stellerian, testing image-processing workloads; and TakeMe2Space itself.

Next year, Satlyt plans to attempt linking two separate satellites into a shared computing system — a step toward what Afullo calls a true compute cloud in orbit. He hopes to be live on 20% of satellites by the end of the decade. The seed round was led by Houston-based Non Sibi Ventures, whose partner Bernard Harris, a NASA astronaut for more than 20 years, backed the bet.

Why it matters

Few satellites currently carry high-powered GPUs, but as more spacecraft builders add them, a shared software layer like Satlyt’s could become essential infrastructure for pooling compute in orbit well before anyone builds dedicated data centers in space.