Satlyt is building software that could let satellite operators run AI workloads in orbit and share computing across spacecraft. The company has raised an $8 million seed round, and its software is scheduled to launch on a SpaceX rocket alongside the first prototype for Google’s Project Suncatcher.
Rather than build satellites, Satlyt aims to provide a software layer that works across different spacecraft. Its first uses focus on processing data onboard and reducing the need to send information back to Earth.
Software designed to work across spacecraft
Founder Rama Afullo says his experience at Google’s cloud computing business and a brief stint at SpaceX’s Starlink communications network in 2024 led him to see potential for computers in space. He says he pitched the idea internally at both companies and was turned down, then left SpaceX to co-found Satlyt.
The company is headquartered in Sunnyvale, California, and Nairobi. Its approach is distinct from companies building orbital data centers or spacecraft: Satlyt wants to make software that can operate across satellites owned by different organizations.
Afullo compares the model to VMware and Snowflake, platforms that let customers run complex software without managing the underlying infrastructure. He describes Satlyt’s goal as an open ecosystem, with software that can work across many companies’ satellites.
Onboard processing could cut downlink costs
Satellites often rely on flight controllers on the ground to diagnose and solve problems. Sending data between a spacecraft and Earth, known as downlink, can be expensive and slow. Processing information or making decisions onboard could reduce the amount that needs to be transmitted.
Satlyt has flown its software on two demonstration missions. Earlier this year, it deployed Google DeepMind’s Gemma AI model on a spacecraft operated by Momentus. The model reduced the size of a transmission about onboard software errors by more than 60%, according to the source article.
That kind of reduction illustrates the company’s initial focus: applying AI to practical spacecraft operations, rather than depending on large, high-powered computing systems in orbit. Afullo says improved efficiency can save hundreds of thousands of dollars per satellite each year.
A multi-customer mission puts the software to work
The upcoming launch is set to carry Satlyt’s software on a spacecraft built by TakeMe2Space, an Indian startup that makes computing hardware for satellites. The mission includes three customers, each testing a different use for the spacecraft.
- NASA is paying to test protocols for cloud computing in space.
- Stellerian, a space surveillance startup, plans to test image-processing workloads.
- TakeMe2Space will demonstrate that its satellite can host software from other companies.
The mission connects Satlyt’s software to a broader question: whether spacecraft can host computing services for customers beyond their builders. The current opportunity, the article reports, is more likely to come from workloads like those Satlyt is pursuing, since few high-powered GPUs are in orbit.
The next step is linking two satellites
Satlyt’s next project is to show that a shared computing system, or cloud, can span two different satellites. The company expects to attempt this next year. If it succeeds, Satlyt could operate as a third-party provider of computing services in orbit.
The seed round was led by Houston-based Non Sibi Ventures. Partner Bernard Harris, who served as a NASA astronaut for more than 20 years, helped shape the firm’s conviction in the business. Non Sibi partner Kent Lucas said the company does not need space data centers to succeed; its opportunity could grow with the number of satellites being launched.
Afullo hopes Satlyt will be live on 20% of satellites by the end of the decade. He expects spacecraft builders to increasingly include GPUs and similar advanced processors, though that vision depends on the market’s development. For now, the company’s plan rests on making existing and future satellites more useful through software that can run onboard and, eventually, share computing across spacecraft.