Editorial illustration for Satlyt raises USD 8M seed to run AI on satellites, launching Thursday on SpaceX
Satlyt Raises $8M to Run AI on Satellites
Satlyt, a startup splitting its operations between Sunnyvale and Nairobi, has raised an $8 million seed round to build software that runs AI workloads on satellites. The company's software launches Thursday on a SpaceX rocket, sharing the ride with the first hardware prototype from Google's own orbital computing project, Project Suncatcher.
The founder, Rama Afullo, spent time at Google's cloud division and briefly at SpaceX's Starlink team in 2024 before striking out on his own. He says he pitched the idea of space-based data centers at both companies and got turned down twice. Now those same companies are racing into the field he says he saw coming.
Satlyt's bet is narrower than its rivals'. While SpaceX, Google, and startups like Starcloud and Cowboy Space Company are building their own spacecraft, Afullo's team is staying out of the hardware business entirely. Instead, Satlyt wants to be the layer that lets AI models run across many satellites at once, regardless of whose metal they're sitting on.
Earlier this year, Satlyt deployed Google DeepMind’s Gemma AI model onboard a spacecraft operated by Momentus. The model cut the size of a transmission about onboard software errors by more than 60%, the kind of efficiency that Afullo says can save hundreds of thousands of dollars per satellite each year.
Why this matters
Satlyt's $8M seed is small money next to what Google and SpaceX are now pouring into the same idea, but the timing is the real story. Afullo pitched orbital data centers inside both companies and got turned down, only to watch SpaceX and Google's Project Suncatcher chase the same concept months later, on the same rocket, on the same day. That's a useful reminder for founders: rejection from a big lab isn't always a verdict on the idea, sometimes it's a verdict on timing and internal politics.
For AI builders, the bigger question is whether space-based compute is actually about AI workloads or about solar power and thermal efficiency, since vacuum and sunlight solve cooling and energy problems that land-based data centers can't. If Satlyt's software ends up running training or inference jobs rather than just telemetry, that's a genuine new compute substrate worth tracking. If it's mostly satellite ops dressed up in AI language, the Thursday launch is more symbolic than structural.
Either way, watch who else starts building hardware for this before deciding it's real.
Related Reading
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
From Archives SpaceX Bundles X Premium, Grok AI, and Cursor for Developers in One Subscription
From Archives NVIDIA Broadcast 2.3 Brings AI Effects to RTX Spark-Optimized Builds
Common Questions Answered
What is Satlyt's primary business model and how much funding did they raise?
Satlyt is a startup that builds software to run AI workloads on satellites, and they raised an $8 million seed round to develop this technology. The company operates between Sunnyvale and Nairobi, with founder Rama Afullo leading the effort to bring AI computing capabilities to orbital platforms.
How did Satlyt demonstrate the efficiency gains of running AI models on satellites?
Satlyt deployed Google DeepMind's Gemma AI model onboard a spacecraft operated by Momentus earlier this year, which reduced the size of transmissions about onboard software errors by more than 60%. This efficiency improvement can save hundreds of thousands of dollars per satellite annually, according to founder Rama Afullo.
What is the significance of Satlyt's software launching on SpaceX on Thursday?
Satlyt's software launch on a SpaceX rocket is historically significant because it shares the ride with Google's Project Suncatcher hardware prototype, marking a convergence of multiple companies pursuing orbital computing on the same day. This timing is notable because founder Rama Afullo had previously pitched the same orbital data center concept to both Google and SpaceX before being rejected.
What was Rama Afullo's background before founding Satlyt?
Rama Afullo spent time at Google's cloud division and briefly worked on SpaceX's Starlink team in 2024 before leaving to start Satlyt. His experience at both companies gave him insight into the potential for running AI workloads on satellites and the opportunity to pursue this vision independently.
Why does the article suggest that Satlyt's rejection from Google and SpaceX wasn't necessarily a negative outcome?
The article notes that Afullo pitched orbital data centers to both companies and was rejected, only to watch SpaceX and Google pursue the same concept months later on the same rocket and day. This demonstrates that rejection from major tech labs can sometimes reflect timing and internal priorities rather than the viability of the idea itself.
Further Reading
- Papers with Code - Latest NLP Research - Papers with Code
- Hugging Face Daily Papers - Hugging Face
- ArXiv CS.CL (Computation and Language) - ArXiv
