Satlyt Orbital Computing software has completed a successful test aboard a satellite, bringing the startup closer to its goal of connecting spacecraft into distributed computing networks. The experiment demonstrated how software can schedule processing tasks using available computing power and communication links.
The test followed 13 months of development with the University of Houston under a NASA-funded research project. Satlyt founder and CEO Rama Afullo said the results represent an early step toward enabling multiple satellites to share computing workloads.
The company’s software is already installed on three satellites. However, the latest experiment did not demonstrate computing tasks moving between separate spacecraft.
Satlyt Orbital Computing Test Uses Simulated Network
The experiment involved a simulated network of four computers, four users and eight services. One physical computer aboard an orbiting spacecraft evaluated how to assign tasks across the network. The remaining computing nodes were simulated.
The software assessed processing capacity and communication availability before generating execution plans. Results were successfully transmitted back to Earth, confirming that the scheduling system could operate on hardware in orbit.
Afullo said the researchers identified trade-offs between performance and computing requirements. One scheduling approach delivered better results but consumed more processing resources.
The experiment also revealed a timing measurement issue. The software recorded arrival timestamps instead of elapsed communication delays. Satlyt acknowledged the problem but said the core software execution and result transmission succeeded.
Turning Existing Satellites Into Virtual Data Centers
Founded in 2024, Satlyt develops software that allows satellites to process information closer to where it is collected. Afullo previously worked at Google and SpaceX before establishing the company.
Its longer-term goal is to coordinate computing resources across spacecraft operated by different organizations. Satellite operators would define permissions, priorities and operating limits, while the software would manage tasks within those constraints.
This approach could reduce dependence on transmitting large volumes of raw information to Earth. For example, a satellite detecting a wildfire could process observations onboard and prioritize relevant information for transmission.
The concept builds on advances in onboard satellite AI, which increasingly enables spacecraft to analyze observations without constant ground-based processing.
Unlike Google’s Project Suncatcher, which explores dedicated orbital AI computing infrastructure, Satlyt aims to use processing hardware already installed on existing spacecraft.
Further Testing Needed Before Satellites Share Workloads
Despite the successful experiment, Satlyt faces several technical challenges. Its software must operate reliably across different hardware, workloads and intermittent communication conditions.
Connecting multiple spacecraft will also require secure data transfers, operator permissions and dependable resource scheduling. The network must recover when communication links become unavailable.
NASA funded the research involving the University of Houston but has not selected Satlyt for a specific mission. The company is also investigating applications beyond low Earth orbit, including support for 4K video streaming from the Moon.
Satlyt plans broader validation before demonstrating computing tasks shared between separate spacecraft. That next phase will determine whether its scheduling software can move beyond simulated networks toward operational space-based computing systems.
