Satlyt, a startup headquartered in Sunnyvale, California, and Nairobi, has secured an $8 million seed round led by Houston-based Non Sibi Ventures to build software that lets AI models run on satellites and share large computing workloads in space.
Co-founder Rama Afullo, a Kenyan American engineer who previously worked at Google Cloud and SpaceX’s Starlink, said both companies rejected his internal pitches for orbital computing. Rather than building spacecraft, Satlyt aims to be a cross-platform software layer, which Afullo likened to Android in a market where players such as SpaceX resemble the iPhone.
Earlier this year, Satlyt ran Google DeepMind’s Gemma model aboard a Momentus spacecraft, cutting the size of an onboard error transmission by more than 60%, which Afullo said could save hundreds of thousands of dollars per satellite annually. Its software will fly on a SpaceX launch Thursday aboard a TakeMe2Space satellite, serving NASA, space surveillance startup Stellerian and TakeMe2Space itself.
The company plans to demonstrate a shared cloud spanning two satellites next year and hopes to run on 20% of satellites by the end of the decade. Non Sibi partner Kent Lucas said Satlyt’s success depends on rising satellite numbers rather than full orbital data centers, while partner Bernard Harris, a former NASA astronaut, helped shape the firm’s conviction.
Featured image: Credit: Satlyt
