The Pulse
Satlyt Raises $8 Million to Run AI on Existing Satellites
Satlyt raised an $8 million seed round led by Non Sibi Ventures to expand software that runs AI and processes data aboard satellites. The Sunnyvale and Nairobi startup is testing whether its software can coordinate computing across spacecra
AI.info Team · Written with AI ·
$8 million for software, not spacecraft
Satlyt raised $8 million in seed funding to run AI and data-processing software on satellites operated by other companies. The round, announced October 1, 2026, was led by Houston-based Non Sibi Ventures. Founder and CEO Rama Afullo said the financing will help the company grow its team and bring its software to more spacecraft and customers.
The Sunnyvale, California, and Nairobi-based startup is not building satellites of its own. It is developing software intended to run on computers already aboard spacecraft, letting operators process data in orbit rather than send every raw image or system log back to Earth. Satlyt’s longer-term goal is to coordinate computing across satellites made by different companies.
“The folks like SpaceX who are doing orbital data centers — if they are the iPhone, we’ll build Android as a horizontally integrated, open ecosystem,”
Rama Afullo, as quoted by TechCrunch
Afullo previously worked in Google’s cloud computing business and at SpaceX’s Starlink network. He told TechCrunch that he had pitched the idea of computing in space at both companies and was turned down. Satlyt’s software is designed to give operators a way to add applications without buying into one spacecraft maker’s complete hardware system.
Onboard processing can trim transmissions
Sending information from orbit to Earth, known as downlink, can be costly and slow. Satlyt’s software aims to let satellites analyze sensor readings or respond to certain operational problems onboard, transmitting selected results instead of all the underlying data. The approach could reduce communications demands, but its value depends on what a spacecraft’s onboard computers can handle.
Satlyt has flown its software on two demonstration missions, TechCrunch reported. Earlier in 2026, the company ran Google DeepMind’s Gemma AI model aboard a spacecraft operated by Momentus. The software cut the size of a transmission about onboard software errors by more than 60%, according to the report; Afullo said that kind of reduction could save hundreds of thousands of dollars per satellite each year.
NASA, Stellerian and TakeMe2Space join a planned mission
A planned launch this week is set to carry Satlyt software on a spacecraft built by Indian company TakeMe2Space. The mission includes NASA, which is paying to test protocols for cloud computing in space; Stellerian, which plans to test image-processing workloads; and TakeMe2Space, which will demonstrate that its spacecraft can host software from other companies.
The work is an early test of Satlyt’s cross-company pitch: software that can run across different spacecraft, rather than a computing system tied to one satellite fleet. The company’s next stated milestone is to try to create a shared computing system spanning two satellites in 2027. TechCrunch reported that few high-powered GPUs are currently in orbit, a constraint on the more demanding AI workloads Satlyt could eventually support.
A two-satellite cloud is the next test
Non Sibi partner Kent Lucas told TechCrunch the company does not need a large orbital data center market to succeed. “It can just be driven by the number of satellites going up,” he said. That thesis puts Satlyt’s near-term prospects on a more modest measure than building large computing stations in space: whether satellite operators will adopt its software on existing missions.
Afullo told TechCrunch he hopes Satlyt will be running on 20% of satellites by the end of the decade. The nearer test is whether it can make software work across two separate spacecraft, with different operators and onboard systems. Until that demonstration, Satlyt has flown software on individual missions, not a shared orbital cloud.