Instead of building custom spacecraft, Satlyt operates as a platform provider, positioning itself as the VMWare or Snowflake of the final frontier. The company aims to unify disparate satellite hardware into a cohesive ecosystem, allowing operators to process sensor data and resolve anomalies in orbit. By reducing the reliance on slow, costly downlinks to Earth, Afullo’s software can save operators hundreds of thousands of dollars per satellite annually. This efficiency was recently proven when the team deployed Google DeepMind’s Gemma model on a Momentus spacecraft, successfully shrinking software error transmission sizes by over 60%.
In section Startups & Technology
Satlyt secures $8M to build the Android of space computing
After rejection from tech giants SpaceX and Google, engineer Rama Afullo is launching his own path to orbital computing. His startup, Satlyt, just closed an $8 million seed round to develop a horizontal software layer that enables AI models to run directly on satellites, bypassing the need for expensive ground-based data processing.
The startup’s technology faces a critical test this week as it launches aboard a SpaceX rocket on a satellite built by TakeMe2Space. The mission includes high-profile participants like NASA and space surveillance firm Stellerian, testing the feasibility of hosting third-party software in orbit. Investors led by Non Sibi Ventures remain confident, noting that Satlyt’s model is not beholden to the massive, capital-intensive infrastructure of dedicated orbital data centers. Rather, the company is betting on the rapid proliferation of GPUs in space, with Afullo targeting a presence on 20% of all orbiting spacecraft by the end of the decade.
Comments (0)
No comments yet. Be the first!