Satlyt raises $8M to run AI software directly on satellites

In short: Satlyt raised $8 million to build software that helps satellites run AI onboard instead of relying on computers on Earth.

What happened

Satlyt, a new company co-founded by former Google and SpaceX product manager Rama Afullo, announced an $8 million seed funding round. The company is based in Sunnyvale, California, and Nairobi. The round was led by Non Sibi Ventures, a Houston-based firm.

Satlyt is not building its own satellites. Instead, it is building software that can run on many different companies’ spacecraft. Afullo describes the goal as being like Android, meaning more open software that can work across many devices, rather than a closed, all-in-one system.

The company says it has already tested its software on two demonstration missions. Earlier this year, Satlyt deployed Google DeepMind’s Gemma AI model (an AI program that can analyze and summarize information) on a spacecraft operated by Momentus. Satlyt said this cut the size of a transmission about onboard software errors by more than 60%.

Satlyt’s software is also set to launch this week on a SpaceX rocket. It will run on a spacecraft built by TakeMe2Space, an Indian startup that makes computing hardware for satellites. Satlyt said the mission includes work for NASA, which is testing “cloud computing in space” (sharing computing work across machines, like renting storage and processing online), and Stellerian, which wants to test image processing for tracking objects in orbit.

Why it matters

Sending data from space down to Earth is often slow and expensive. If satellites can do more thinking onboard, they can send back fewer, smaller messages, which can save money and help operators respond faster when something goes wrong.

Source: TechCrunch AI