Rama Afullo, a former engineer at Google’s cloud computing division and a brief member of SpaceX’s Starlink team in 2024, has founded Satlyt. This new company, based in Sunnyvale, California, and Nairobi, aims to develop software for satellites capable of handling extensive computing tasks in space. Recently, Satlyt raised an $8 million seed round, marking a significant step forward for its mission.
Afullo’s vision emerged from his experiences at Google and SpaceX, where he recognized there was untapped potential for computers in space. Despite pitching his ideas internally at both companies, he faced rejection. Now, he’s set on addressing that gap with Satlyt, which plans to enable multiple satellites to collaborate on computing tasks. Importantly, Satlyt does not plan to build its own spacecraft; instead, it focuses on creating software that runs AI models on various satellites, similar to how VMWare and Snowflake provide infrastructure independence for complex applications.
This venture comes at a time when SpaceX is increasingly interested in orbital data centers. Satlyt’s software is expected to launch aboard a SpaceX rocket alongside the first prototype of Google’s Project Suncatcher, an initiative aimed at establishing a space data center.
Satlyt’s software has already been tested on two demonstration missions, concentrating on enhancing operational efficiencies for spacecraft. Traditional spacecraft operations often rely on ground-based controllers to resolve issues, which can be slow and costly. By integrating AI for on-board decision-making, Satlyt aims to reduce the costs associated with downlinking data to Earth.
Earlier this year, Satlyt successfully deployed Google DeepMind’s Gemma AI model on a spacecraft operated by Momentus. This deployment reduced the size of transmissions regarding 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.
This week’s mission will utilize Satlyt’s software on a spacecraft developed by TakeMe2Space, an Indian startup specializing in satellite computing hardware. The mission includes three customers: NASA, which is funding tests for cloud computing protocols in space; Stellerian, focused on space surveillance to evaluate image processing workloads; and TakeMe2Space, aiming to demonstrate the satellite’s capacity to host third-party software.
Currently, there are few high-powered GPUs in orbit. However, Satlyt is looking to establish a shared computing system across multiple satellites, with plans to attempt this next year. If successful, it will allow the company to build a true compute cloud in orbit as a third-party provider to spacecraft builders.
The recent funding round was led by Non Sibi Ventures, with partner Bernard Harris, a former NASA astronaut, expressing confidence in the company’s approach. One reason the firm invested was that Satlyt doesn’t depend on the highest-octane vision of space data centers, which faces uncertain economics and a launch bottleneck. Instead, Harris emphasized that the company’s success could be driven by the increasing number of satellites being launched.
Afullo aims to have Satlyt’s software operational on 20% of satellites by the end of the decade, predicting that virtually every spacecraft builder will incorporate GPUs and similar processors into their designs. He argues, “If you’re putting up a satellite without putting up a GPU on it, at the very least, you’re doing yourself a disservice.”
Satlyt is among dozens of other startups vetted by TechCrunch and appearing at Disrupt as part of our broader Battlefield competition happening October 13-15 in downtown San Francisco.
