Rama Afullo, co-founder and CEO of Satlyt, says his experience at Google and a brief stint at SpaceX’s Starlink in 2024 convinced him there is untapped potential for computing in orbit. After internal pitches at both companies failed to gain traction, he left and started Satlyt, with offices in Sunnyvale, California, and Nairobi, Kenya.
Satlyt develops software that enables satellites to run AI models and share large computing jobs across multiple spacecraft. The company has raised an $8 million seed round, Afullo told TechCrunch exclusively, and its software is scheduled to launch this Thursday on a SpaceX rocket alongside the first prototype from Google’s Project Suncatcher.
Approach and early demonstrations
Unlike SpaceX, Google or startups such as Starcloud and Cowboy Space Company, Satlyt does not plan to build its own spacecraft. Instead, it builds a platform intended to run complex workloads across satellites from different manufacturers — Afullo compares the approach to VMWare or Snowflake, offering users a way to run software without managing the underlying hardware. "If folks like SpaceX are the iPhone, we will build Android as a horizontally integrated, open ecosystem," he said.
Satlyt has already flown its software on two demonstration missions. Its immediate focus is improving operational efficiency: spacecraft commonly rely on ground flight controllers to resolve anomalies, and downlinking data to Earth is often costly and slow. Running AI onboard can reduce the need for expensive transmissions and speed up decision-making.
Earlier this year, Satlyt deployed Google DeepMind’s Gemma AI model on a spacecraft operated by Momentus. That model reduced the size of transmissions about onboard software errors by more than 60%, a reduction Afullo says can save hundreds of thousands of dollars per satellite each year.
This week’s mission and customers
This week’s launch will place Satlyt’s software on a satellite built by TakeMe2Space, an Indian startup that makes computing hardware for satellites. The mission will serve three customers:
- NASA, which is paying to test cloud-computing protocols in space;
- Stellerian, a space-surveillance startup that wants to test image-processing workloads;
- TakeMe2Space itself, to demonstrate that its satellite can host third-party software.
Currently, few high-powered GPUs are in orbit, so early opportunities align with the kinds of workloads Satlyt targets. The company plans to next attempt a shared compute system — a cloud spanning two different satellites — as soon as next year. If successful, Satlyt would function as a third-party orbital compute provider to spacecraft builders, turning satellites into revenue-generating managed services.
Investors and market outlook
Satlyt’s seed round was led by Houston-based Non Sibi Ventures. Bernard Harris, a partner at Non Sibi who served as a NASA astronaut for more than 20 years, helped inform the firm’s conviction in-space business models. Non Sibi partner Kent Lucas noted that Satlyt’s upside doesn’t depend solely on large orbital data-center economics or fast adoption: success can be driven by the sheer number of satellites launched.
Afullo’s goal is for Satlyt to be live on roughly 20% of satellites by the end of the decade, a moment he expects will coincide with most spacecraft builders including GPUs or similar advanced processors in their designs. "If you’re putting up a satellite without putting up a GPU on it, at the very least, you’re doing yourself a disservice," he said.
Visibility and next steps
Satlyt is one of dozens of startups vetted by TechCrunch and is appearing at TechCrunch Disrupt, taking place October 13–15 in downtown San Francisco as part of the Battlefield competition. The company’s near-term technical milestone is demonstrating a multi-satellite shared compute cloud; achieving that would mark a significant step toward a cross-vendor orbital compute layer.
