Rama Afullo pitched orbital data centers at Google and SpaceX and was turned down at both. Now his startup, Satlyt, has raised $8 million to pursue the idea anyway — but with a twist: it won't build satellites or space data centers. Instead, Satlyt makes software that lets AI models run on existing satellites, allowing spacecraft to process data and diagnose problems without sending everything to Earth. The company is headquartered in Sunnyvale, California, and Nairobi.
Satlyt's software is scheduled to launch Thursday on a SpaceX rocket, alongside Google's first Project Suncatcher prototype. It will run on a spacecraft built by TakeMe2Space, with paying customers including NASA, which wants to test cloud-computing protocols in orbit, and Stellerian, a space-surveillance startup testing image processing. Satlyt has already flown two demonstrations; earlier this year it deployed Google's Gemma model on a Momentus spacecraft and cut the size of an error-related transmission by more than 60%, which the company says can save hundreds of thousands of dollars per satellite annually.
Afullo compares Satlyt's role to VMware or Snowflake, and positions it as the Android to what he expects will be SpaceX's iPhone-like orbital data centers. The seed round was led by Non Sibi Ventures; partner Kent Lucas said Satlyt does not need full space data centers to succeed, only more satellites. Next year, the startup plans to demonstrate a shared computing cloud spanning two satellites, and it hopes to run on 20% of satellites by the end of the decade.
Both TechCrunch articles report the same facts; no substantive differences emerged, though one notes the story was originally published on October 1.