Satlyt Demonstrates Software for Distributed Satellite Computing in NASA-Funded Test

Satlyt announced that its orbital computing platform passed a key validation test earlier this month, showing that software can allocate processing jobs among satellite computers while in flight. The achievement moves the company closer to its goal of turning existing spacecraft into a networked data center, a capability that could speed delivery of insights from remote sensing and other space-based applications.

The experiment was conducted in partnership with the University of Houston under a NASA-funded research project. Over a 13-month development cycle, Satlyt built a simulated environment comprising four computing nodes, four distinct users, and eight services. The software evaluated the available compute capacity and communication links, then generated a plan for assigning and executing tasks across the nodes. A single physical computer mounted on a spacecraft executed the simulation, returned the results to Earth, and confirmed that the planning component operated correctly in orbit. The test demonstrated that the system can produce execution plans on hardware in space, although further trials are needed to verify data exchange between multiple physical computers.

According to CEO Rama Afullo, the platform is designed to let operators define only high-level constraints such as permissions, priorities, and resource limits. Within those bounds, the software decides how to distribute workloads, effectively creating a virtual data center that spans several satellites. Afullo highlighted that current NASA missions face limited onboard compute, intermittent downlink windows, and the need to return actionable information quickly, issues his software aims to mitigate. He cited a wildfire-detection scenario where onboard AI could process imagery and transmit only the relevant alert, reducing latency and bandwidth consumption.

Satlyt, founded in 2024 by Afullo after stints at Google and SpaceX, already has its code deployed on three orbiting satellites. The company placed in the top five of this year’s TechCrunch Disrupt Battlefield, signaling strong interest from the venture community. Its broader vision includes supporting AI workloads, 4K video streaming from lunar assets, and enabling developers to build space applications without recreating infrastructure for each mission.

The recent test also revealed technical gaps that the team must address. An issue was identified in how the software timestamps communication delays, mistakenly recording arrival time instead of elapsed delay. Additionally, the experiment underscored the need for reliable inter-satellite links, secure permission frameworks, and robust recovery mechanisms when a node becomes unavailable. Afullo acknowledged that while the planning module succeeded, comprehensive validation across varied workloads, hardware configurations, and connectivity conditions remains pending.

Looking ahead, Satlyt plans to broaden validation efforts, integrate its software with additional spacecraft, and demonstrate true workload sharing across separate platforms. The company is also exploring applications beyond low Earth orbit, leveraging its NASA-funded research to support high-bandwidth lunar missions. Success in these areas could reshape how satellite operators handle data, shifting more processing from ground stations to the edge of space. For founders and investors, Satlyt’s progress illustrates the emerging market for space-based compute services and raises the question of how quickly the industry will adopt distributed orbital processing as a standard capability.