Satlyt Raises $8M to Bring AI Cloud Computing to Satellites

Overview

Satlyt, a startup co-founded by former Google and SpaceX engineer Rama Afullo, has secured an $8 million seed funding round to develop software that enables artificial intelligence operations directly on satellites. The company aims to transform passive space assets into active computing nodes by allowing multiple satellites to share heavy computational workloads in orbit.

Headquartered in Sunnyvale, California, with operations in Nairobi, Satlyt is positioning itself as an infrastructure layer for the growing space economy. Rather than building hardware, the company focuses on creating a versatile software platform that can run AI models across spacecraft from various manufacturers.

The timing of this investment aligns with a broader industry shift toward orbital data centers. Major players like SpaceX are now aggressively pursuing in-space computing capabilities, creating a favorable environment for Satlyt’s ecosystem-based approach.

The Vision for Orbital Cloud Computing

Rama Afullo’s journey to space computing began during his tenure at Google’s cloud division and a brief stint at SpaceX’s Starlink network in 2024. He observed significant untapped potential for distributed computing in orbit but found that internal pitches for such initiatives were rejected by both companies.

“When I was at SpaceX, I tried to pitch this internally. They said no. When I was at Google, I tried to pitch this internally. They said no,” Afullo explained regarding his initial proposals.

Undeterred, Afullo left his roles to co-found Satlyt. The company’s core mission is to build software that allows satellites to handle large computing jobs without relying entirely on ground-based resources. This model draws parallels to virtualization pioneers like VMware and data platforms like Snowflake, which abstract complex infrastructure away from the end user.

Afullo compares Satlyt’s role in the space sector to the relationship between Android and the iPhone. While companies like SpaceX may build their own proprietary orbital data centers, Satlyt intends to provide an open, horizontally integrated ecosystem. This strategy allows their software to operate across satellites built by different manufacturers, fostering a more collaborative and interoperable space environment.

Demonstrated Efficiency and Strategic Partnerships

Satlyt has already validated its technology through two successful demonstration missions. The primary use case focuses on operational efficiencies for spacecraft, which traditionally depend on ground controllers to resolve anomalies. By processing data onboard, satellites can reduce the need for costly and slow data downlinks to Earth.

Earlier this year, the company deployed Google DeepMind’s Gemma AI model aboard a spacecraft operated by Momentus. The results were significant: the AI model reduced the size of transmissions regarding onboard software errors by more than 60%. According to Afullo, these efficiency gains can save satellite operators hundreds of thousands of dollars annually.

The upcoming launch, scheduled for Thursday, will place Satlyt’s software on a spacecraft built by TakeMe2Space, an Indian startup specializing in satellite computing hardware. This mission serves three distinct customers:

NASA: Testing protocols for cloud computing in space.
Stellerian: A space surveillance startup evaluating image processing workloads.
TakeMe2Space: Demonstrating the capability of hosting third-party software on their hardware.

This multi-client approach highlights the versatility of Satlyt’s platform and its ability to cater to diverse needs, from government research to commercial surveillance.

Investment Thesis and Future Roadmap

The seed round was led by Non Sibi Ventures, a Houston-based firm where partner Bernard Harris, a former NASA astronaut, provides critical industry insight. Another partner, Kent Lucas, emphasized that the investment thesis does not rely on the speculative vision of massive space data centers.

“We don’t need data centers in space for Rama to be wildly successful,” Lucas noted. “It can just be driven by the number of satellites going up.”

This pragmatic view distinguishes Satlyt from competitors chasing high-risk, high-reward orbital infrastructure. Instead, the company leverages the increasing volume of small satellites being launched, offering value through software that enhances their utility without requiring massive capital expenditure on new hardware.

Currently, high-powered GPUs in orbit are rare. Satlyt’s immediate focus remains on workloads that do not require extreme computational power, such as error detection and image compression. However, the company’s long-term goal is to create a shared computing cloud spanning multiple satellites.

Afullo expects to attempt this multi-satellite cloud demonstration next year. If successful, Satlyt would become a third-party compute provider, effectively turning any satellite equipped with their software into a revenue-generating service.

Market Outlook and Industry Context

Afullo projects that by the end of the decade, Satlyt could be operational on 20% of all satellites. This target assumes a future where virtually every spacecraft builder integrates GPUs or similar advanced processors into 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,” Afullo stated.

This perspective aligns with broader trends in the private space industry, where the focus is shifting from mere launch capacity to in-orbit functionality. As more companies enter the space domain, the demand for flexible, software-defined solutions like Satlyt’s is expected to grow.

Satlyt’s involvement in the Disrupt Battlefield competition further underscores its status among emerging tech innovators. The company’s ability to combine expertise from tech giants like Google and SpaceX with a lean, software-first approach positions it as a key player in the nascent field of space-based AI.

The success of Satlyt could catalyze a new business model in the space sector, one where computational power is a service available to all, regardless of whether a company owns its own fleet of satellites. As the economics of space become more viable, platforms that reduce operational costs and increase data value will likely become essential utilities for the next generation of space exploration.