Satlyt Secures $8M to Build AI Software for Satellites After SpaceX and Google Rejected the Idea

Former SpaceX Product Manager Launches Satlyt After Tech Giants Said No

Rama Afullo, a Kenyan American engineer who previously worked at Google's cloud computing division and SpaceX's Starlink communications network, has raised $8 million in seed funding for Satlyt1

1

. The satellite software provider, headquartered in Sunnyvale, California, and Nairobi, aims to build software that enables multiple satellites to share the work of large computing jobs in space. The funding round was led by Houston-based Non Sibi Ventures2

2

. Afullo's vision for orbital data centers was rejected when he pitched it internally at both SpaceX and Google, prompting him to leave and co-found Satlyt in 20241

1

. The timing proved fortuitous—SpaceX has since embraced orbital data centers, and Satlyt's software launched Thursday on a SpaceX rocket alongside the first prototype for Google's space data center effort, Project Suncatcher1

1

.

How Satlyt's Platform Enables AI on Satellites to Cut Costs

Unlike competitors such as SpaceX, Google, Starcloud, and Cowboy Space Company, Satlyt isn't building its own spacecraft. Instead, the AI software provider for satellites focuses on creating platforms to operate AI models on satellites, comparing its approach to VMWare and Snowflake—companies whose platforms let users run complex software without worrying about underlying infrastructure1

1

. Most satellites include limited onboard computing equipment and typically send collected data to ground-based servers rather than process it locally2

2

. This creates challenges since the average satellite only flies within range of its operator's terrestrial antenna for a few minutes each day, limiting data transmission and causing accuracy loss during data downlink2

2

. Satlyt's platform addresses these issues by enabling satellites with GPUs to process data onboard using AI models, reducing the amount of information that must be sent to Earth2

2

.

Proven Results: 60% Reduction in Data Transmission Using Gemma AI Model

Satlyt has already flown its software on two demonstration missions, with initial focus on finding operational efficiencies for spacecraft that typically depend on ground-based flight controllers to solve problems1

1

. The first satellite to feature Satlyt's platform launched into low-Earth orbit in March aboard a spacecraft operated by Momentus2

2

. Earlier this year, Satlyt deployed Google DeepMind's Gemma AI model onboard the Momentus spacecraft, where it ran the open-source Gemma 3 1B language model2

2

. The companies evaluated the system by simulating common satellite software errors, with Satlyt's platform using the model to collect error logs, identify root causes, and suggest fixes2

2

. The deployment cut the size of transmissions about onboard software errors by more than 60%, reducing telemetry that had to be beamed down to Earth by over 64%1

1

2

2

. Afullo says this kind of efficiency can save hundreds of thousands of dollars per satellite each year1

1

.

New Launch Brings Orbital Computing Tests for NASA and Space Surveillance

This week's launch puts Satlyt's software to work onboard a spacecraft built by TakeMe2Space, an Indian startup that builds computing hardware for satellites1

1

. The mission involves three customers: NASA, which is paying the company to conduct orbital computing tests for protocols for cloud computing in space; Stellerian, a startup focused on space surveillance that wants to test image processing workloads; and TakeMe2Space itself, demonstrating the satellite can host other companies' software1

1

2

2

. As of April, Satlyt was working to upgrade its stack to a newer, more capable AI model called Gemma 4 E2B, with a goal of lowering the model's memory usage by more than 90%2

2

.

Building a Shared Computing Cloud Across Multiple Satellites

Satlyt's next project aims to create a shared computing system—a cloud—spanning two different satellites, which the company expects to attempt next year1

1

. If successful, this will allow Satlyt to build a true compute cloud in orbit as a third-party provider to spacecraft builders, operating AI workloads in space across multiple platforms1

1

. Afullo describes the approach: "The way to think about it is that we turn your satellite into a revenue-generating managed service"1

1

. The satellite software provider is reportedly working on new software tools that will make it possible to distribute workloads across multiple satellites, with the first system featuring this technology set to launch into orbit next year2

2

.

Why Investors Bet on Satlyt's Vision for 20% Satellite Market Share

Non Sibi Ventures partner Bernard Harris, who served as a NASA astronaut for more than 20 years, helped ground the firm's conviction in the novel, space-based business1

1

. Partner Kent Lucas explained one reason for the investment: "We don't need data centers in space for Rama to be wildly successful, right? It can just be driven by the number of satellites going up"1

1

. This positions Satlyt advantageously since it doesn't depend on the highest-octane vision of space data centers, which faces uncertain economics and a launch bottleneck1

1

. Afullo's hope is to be live on 20% of satellites by the end of the decade, when he expects virtually every spacecraft builder to be putting GPUs and similar advanced processors in their satellites1

1

. "If you're putting up a satellite without putting up a GPU on it, at the very least, you're doing yourself a disservice, right?" he says1

1

. Right now, few high-powered GPUs are in orbit, but the most likely opportunity lies in the kinds of workloads Satlyt is already demonstrating1

1

.