Satlyt takes another step toward turning satellites into space data centers

Onboard satellite data processing costs less, which means more AI models can be run on satellites.

Satlyt, a startup aiming to transform satellites into orbital data centers, has successfully completed a crucial test of its software. The company's technology enables multiple satellites to share computing tasks in orbit, optimizing resource allocation based on communication availability and processing power. This test, conducted in collaboration with the University of Houston as part of a NASA-funded research project, demonstrated the software's ability to manage and execute tasks across a simulated network of four computers. While only one physical computer was used, the results indicated the software can generate plans for onboard hardware, a significant step towards distributed satellite computing. Satlyt's software is already deployed on three active satellites, and this test validates its core planning component for space-based data processing, reducing reliance on ground stations and enhancing the speed of data analysis. Despite challenges like time-referencing delays, the core functionality of onboard execution and output return was successful, paving the way for future demonstrations of inter-satellite computing.

Satlyt's successful test signals a move towards more autonomous and efficient satellite operations, where data is processed closer to its source. This capability is particularly vital for time-sensitive applications like wildfire detection, where immediate analysis can significantly improve response times and data utility. By reducing the dependency on slow and intermittent downlinks to Earth, Satlyt's software aims to unlock new possibilities for real-time decision-making in space.

The implications extend to developers, who may soon be able to build space applications without the need for extensive custom infrastructure for each mission. This shift towards standardized, on-orbit computing could democratize access to space-based AI and data processing, fostering innovation across various sectors, from Earth observation to scientific research and beyond.