Satlyt raises $8M seed to build open software platform for AI workloads on satellites

Satlyt, founded by Rama Afullo, a former Google and SpaceX product manager, raised an $8 million seed round led by Non Sibi Ventures to build software that enables AI models to run on satellites. Afullo previously pitched the idea internally at both Google and SpaceX but was rebuffed, prompting him to launch Satlyt with headquarters in Sunnyvale, California, and Nairobi.

The company’s software is scheduled to launch this week aboard a SpaceX rocket alongside the first prototype for Google’s orbital data center effort, Project Suncatcher. Unlike competitors such as SpaceX, Google, Starcloud, and Cowboy Space Company, Satlyt does not plan to build its own spacecraft. Instead, it is developing a software platform designed to operate AI models across multiple satellite providers, drawing a comparison to VMWare and Snowflake for orbital computing.

Satlyt’s initial focus is on operational efficiencies, such as reducing the size of transmissions for onboard software errors by more than 60% using Google’s Gemma AI model. This capability was demonstrated earlier this year when Satlyt deployed Gemma on a spacecraft operated by Momentus, a change Afullo says can save hundreds of thousands of dollars per satellite annually.

The upcoming mission, in partnership with TakeMe2Space, will test Satlyt’s software on a spacecraft built by the Indian startup. The mission includes three customers: NASA, which is testing cloud computing protocols in space; Stellerian, which is evaluating image processing workloads; and TakeMe2Space, which aims to demonstrate the satellite’s ability to host other companies’ software.

Looking ahead, Satlyt plans to demonstrate a shared computing cloud spanning two satellites next year, with the goal of establishing itself as a third-party provider of in-orbit compute resources. Afullo stated, “The way to think about it is that we turn your satellite into a revenue-generating managed service.”

Afullo expressed confidence that by the end of the decade, 20% of satellites will carry advanced processors like GPUs, making Satlyt’s software platform increasingly relevant. He argued, “If you’re putting up a satellite without putting up a GPU on it, at the very least, you’re doing yourself a disservice.”