Satlyt Tests Orbital Computing Software on Real Hardware
Original: Satlyt takes another step toward turning satellites into space data centers
Why This Matters
Edge computing in orbit could cut satellite downlink bottlenecks and enable real-time AI processing.
Satlyt, a startup building software to turn satellites into distributed data centers, completed a successful orbital computing test in October 2026, partnering with the University of Houston on a NASA-funded project after 13 months of development.
Satlyt, founded in 2024 by ex-Google and SpaceX engineer Rama Afullo, ran its most significant test yet earlier this month: software that can distribute computing tasks across multiple satellites based on available compute and communication bandwidth. The test used a simulated network of four computers, four users, and eight services — but crucially, one physical computer aboard an actual spacecraft in orbit ran the software, with the other three nodes simulated. The system autonomously generated task plans on orbital hardware and returned results to Earth.
The vision: users set permissions and operating limits, and the software handles task allocation across satellites on its own. Satlyt's software is already deployed on three orbiting satellites. The test identified a trade-off — the best-performing scheduling approach also demanded the most compute. More testing is needed to validate how multiple physical computers exchange data in practice.
The work was conducted under a NASA-funded research collaboration with the University of Houston. NASA has not selected Satlyt for any mission or formal deployment. Still, Afullo pointed out that NASA's own missions face the same constraints the startup targets: limited onboard compute, patchy communications, and slow data returns to Earth. Satlyt placed in the top five at TechCrunch Disrupt Battlefield 2026.