Rama Afullo, a former product manager from Google and SpaceX, founded Satlyt to bring artificial intelligence to orbit. With $8 million in funding, Satlyt plans to deploy AI models on satellites, making space computing more efficient and cost-effective.

The timing couldn't be better. SpaceX is now fully committed to orbital data centers, and Satlyt's software will launch on a SpaceX rocket alongside Google's Project Suncatcher. Unlike building its own spacecraft, Satlyt focuses on software to operate AI models, much like VMware and Snowflake.

Early tests have shown significant benefits. In a recent mission, Satlyt's software reduced transmission sizes by more than 60% using Google's DeepMind AI model. This efficiency could save hundreds of thousands of dollars per satellite annually. Satlyt's next goal is to demonstrate a shared computing system across multiple satellites, potentially creating a true compute cloud in orbit.

The company hopes to be integrated into 20% of satellites by the end of the decade, with every spacecraft builder expected to install GPUs in their satellites. Satlyt's success hinges on the increasing number of satellites going up, not the economics of orbital data centers.