Satlyt raises $8 million to deploy AI in space

Satlyt, a startup specializing in space-based data processing and AI integration, has raised $8 million in a seed funding round. This achievement will allow the company to develop software that unifies AI models in space into a single network. This is reported by Techcrunch.com reports.

According to ixbt.com and TechCrunch, startup founder Rama Afullo previously worked in Google's cloud computing business and SpaceX's Starlink network. Drawing on his experience, he realized that the potential for using computing power in space was underestimated and decided to bring his idea to life.

Afullo once pitched this idea to both Google and SpaceX leadership, but both companies rejected it internally. Following this, he founded Satlyt, which is headquartered in Sunnyvale, California, and Nairobi, Kenya.

Space technology market and a new approach

Interest in space data centers has surged recently. While SpaceX is fully committed to orbital data center projects, Google is also pursuing its Project Suncatcher space initiative. However, unlike other companies, Satlyt has stated it does not plan to build its own spacecraft.

The startup team is creating specialized software that allows AI models to run on satellites from various companies. Afullo compared this approach to platforms like VMWare or Snowflake, emphasizing that it enables the execution of complex programs regardless of the underlying infrastructure.

According to a company representative, if SpaceX acts as the iPhone for the orbital data center market, Satlyt aims to position itself as a universal provider like Android in an open ecosystem. The software has already been successfully tested in two pilot missions.

Practical efficiency and future plans

Satellites typically rely on ground control centers to solve problems. Sending data to Earth is an expensive and slow process. Autonomous decision-making in space using AI provides significant cost savings.

Earlier this year, Satlyt installed the Gemma AI model, developed by Google DeepMind, on a spacecraft managed by Momentus. This model reduced data transmission volume for software errors by more than 60%, saving hundreds of thousands of dollars per satellite annually.

In an upcoming mission, Satlyt's software will begin operations on a spacecraft prepared by the Indian startup TakeMe2Space. Partners including NASA, Stellerian, and TakeMe2Space will participate in this project to test cloud computing protocols and space observation tasks.