Satlyt, co-founded by Rama Afullo, has raised $8 million to build software that runs AI workloads on satellites. Afullo previously worked in Google Cloud and at SpaceX's Starlink unit, and says he pitched the idea of orbital computing internally at both companies without success. Now he is pursuing it independently, with a focus on software rather than building his own spacecraft.
The company's approach is akin to VMware or Snowflake: a platform that works across many satellite operators' hardware. Its initial focus is operational efficiency, such as using AI to resolve anomalies or process sensor data in orbit, avoiding expensive and slow downlinks. Earlier this year, Satlyt deployed Google DeepMind's Gemma model on a Momentus spacecraft, reducing the size of a transmission about software errors by more than 60%.
This week's launch will put Satlyt's software on a TakeMe2Space satellite with three customers: NASA, which is testing cloud computing protocols; Stellerian, which wants to test image processing for space surveillance; and TakeMe2Space itself, to demonstrate hosting third-party software. Next year, Satlyt plans to show a shared computing cloud spanning two satellites. Investor Kent Lucas said the company does not need full orbital data centers to succeed, just the growing number of satellites. Afullo hopes to be live on 20% of satellites by the end of the decade.