Satlyt Raises $8M to Run AI on Satellites, Cuts Data Transfer 60%

Executive Overview

Satlyt, a startup founded by former Google and SpaceX product manager Rama Afullo, has closed an $8 million seed round led by Non Sibi Ventures. The company provides software that enables artificial intelligence models to run directly on satellites, reducing the need to transmit data to Earth. In a recent test, Satlyt deployed Google DeepMind's Gemma model on a Momentus spacecraft, cutting software error data transfer by over 60%—a reduction that founder Rama Afullo says can save hundreds of thousands of dollars annually per satellite. The software is set to launch this Thursday on a SpaceX rocket alongside Google's Project Suncatcher prototype, marking a significant step toward orbital computing.

Technical Specification Table

Deep-Dive Features & Architecture

Satlyt was founded by Rama Afullo, a Kenyan-American engineer who previously worked in Google's cloud computing division and spent a brief period at SpaceX's Starlink network in 2024. The company emerged after internal attempts to adopt orbital data center concepts failed. Satlyt focuses on building software that allows multiple satellites to share the burden of large computing tasks in space, likening its business model to VMware and Snowflake, which enable complex software to run without worrying about underlying infrastructure.

Unlike SpaceX, Google, and startups such as Starcloud and Cowboy Space Company, Satlyt does not plan to build its own spacecraft. Instead, it offers a software platform that operates across satellites from multiple companies. Afullo describes his strategy: "If companies like SpaceX represent the iPhone, we will build Android as an open, horizontally integrated ecosystem."

Earlier this year, Satlyt deployed Google DeepMind's Gemma model on a spacecraft operated by Momentus, where the model reduced software error data transfer by over 60%—a feat Afullo says can save hundreds of thousands of dollars annually per satellite. This week, Satlyt's program launches on a vehicle built by Indian company TakeMe2Space, with three customers: NASA, which is paying to test space cloud computing protocols; Stellerian, specializing in space surveillance and orbital object tracking; and TakeMe2Space itself, to demonstrate the satellite's ability to host third-party software.

Benchmark & Competitive Performance

Satlyt distinguishes itself in the orbital data center market by not relying on building spacecraft or physical infrastructure, reducing capital risks and launch bottlenecks. While companies like SpaceX and Google face uncertain economic challenges and rocket launch constraints, Satlyt's model focuses on scalable software as the number of satellites increases. According to Kent Lucas, a partner at Non Sibi: "We don't need data centers in space for Rama to be hugely successful; it could be..."

The company's key performance metric—a 60%+ reduction in data transfer—directly translates to operational cost savings. By processing sensor data onboard rather than transmitting raw data to Earth, satellites can operate more autonomously and reduce dependency on ground control. This benchmark positions Satlyt as a leader in onboard AI efficiency for space applications.

Industry Impact & Enterprise Adoption

Satlyt's approach could accelerate the adoption of AI in space by making it accessible to satellite operators without the need for custom hardware. The company's software is already being tested by NASA for space cloud computing protocols, indicating strong institutional interest. Stellerian's use of Satlyt for image processing demonstrates commercial viability in Earth observation and space situational awareness. By hosting third-party software on TakeMe2Space's satellite, Satlyt proves its platform can support a multi-tenant ecosystem, akin to a cloud computing provider in orbit.

With a target of being on 20% of satellites by 2030, Satlyt aims to become the de facto operating system for orbital AI. The company's asset-light model allows it to scale rapidly as more satellites are launched with advanced GPUs and processors. This could democratize access to space-based AI, enabling smaller operators to leverage advanced analytics without massive infrastructure investments.

Conclusion

Satlyt's $8 million seed round and successful demonstration of Gemma on satellites mark a pivotal moment in space computing. By focusing on software that reduces data transfer and enables onboard AI, Satlyt addresses critical cost and latency challenges in satellite operations. With backing from Non Sibi Ventures and partnerships with NASA and others, the company is well-positioned to capture a significant share of the emerging orbital data center market. As satellite constellations grow, Satlyt's open ecosystem strategy could become the standard for running AI in space.

Media Source: TechCrunch AI | Official Company Statement: Original Source | Fact Verification & Analysis: AI Tools Oasis