Satlyt Seed Round: $8M for AI Software on Other People’s Satellites

The Satlyt seed round funds a software answer to a question that hardware companies have dominated in 2026: where should space data be processed? Several well-funded startups are building dedicated computing satellites. Satlyt owns no spacecraft. It installs its platform on satellites that are already flying or about to launch, so that operators can process data on board and send down only the results. This article, part of the Kurums Space Economy hub, looks at the round, the model, the customers and the risks.

What was announced in the Satlyt seed round?

Satlyt raised $8 million in seed funding, announced on 1 October 2026 and led by non sibi ventures. Reported co-investors are TLCOM, Antler, Slauson & Co., Launch Africa Ventures, Enza Capital, Askya Investment Partners, Demos Capital, BAG Collective, Gaingels and AXIAN Investment, plus follow-on money from existing backers.

The valuation was not disclosed. Payload reports that the round takes total funding “well past $10M”, without an exact figure. The company says the proceeds will expand its engineering and delivery teams, support more spacecraft and customers, increase research spending and build out its network of virtual data centres in orbit.

The timing was deliberate. The announcement coincided with a new deployment on SpaceX’s Transporter-18 rideshare mission, which Space in Africa reported was targeted for 1 October 2026. Announcing a round on launch day ties the financing to a visible product milestone.

What does Satlyt do?

Satlyt builds software that lets a satellite process its own data and run AI models on board. Operators can program it before launch or upload applications in flight. The satellite then downlinks only the outputs that matter, which saves scarce bandwidth and ground-station time.

The published results come from a deployment on a spacecraft built by Momentus. SiliconANGLE reports that the first satellite running the platform launched in March 2026 and runs Google’s open-source Gemma 3 1B language model. Payload dates the demonstration of AI inference with Gemma in space to May 2026. The model was used on the satellite’s own system logs: collecting error data, identifying root causes and suggesting fixes.

The outcome was a large cut in data sent to the ground. SiliconANGLE reports telemetry transmission fell by more than 64% in testing, and Condia reports a reduction in overall data of up to 97%. SiliconANGLE adds that the company is testing a newer model, Gemma 4 E2B, with the aim of cutting its memory use by more than 90% so it fits on constrained satellite computers.

Chief executive Rama Afullo described the practical case to Condia: “Running applications on existing spacecraft gives operators a way to do more with the computing hardware they already have in orbit.”

Who are the founders and team?

Satlyt was founded in 2024 by chief executive Rama Afullo. SiliconANGLE reports he was previously a product manager at SpaceX’s Starlink and, before that, a cloud architect at Google. Space in Africa describes him as Kenyan-American and names Nelson Psenjen as chief technology officer.

The same report names two senior engineers, Junn Wangari (space systems) and Leina Moli (space software), and describes a Kenyan-American leadership team with a substantial engineering presence in Nairobi through a Kenyan subsidiary. Total headcount was not published.

The dual base is a cost and talent decision as well as a story. A US headquarters in Sunnyvale gives access to NASA contracts and US customers. An engineering centre in Nairobi gives access to software talent at lower cost than Silicon Valley. The investor list reflects both sides, with African venture firms such as TLCOM, Launch Africa Ventures and Enza Capital alongside US funds.

How does the business model work and who are the customers?

Satlyt positions itself as a managed software layer that earns revenue from computing capacity on satellites it does not own. Afullo’s summary, reported via Dealroom: “The way to think about it is that we turn your satellite into a revenue-generating managed service.” Pricing has not been disclosed.

The reference points he uses are enterprise software companies, not space companies. He compares the positioning to VMware and Snowflake, and frames the contest with vertically integrated players this way: “If they are the iPhone, we’ll build Android as a horizontally integrated, open ecosystem.”

Named customers and partners so far are a mix of government and commercial. Payload reports a NASA STTR contract worth $4 million, awarded in November 2025, for delay-tolerant edge protocols. Space in Africa says the NASA research software, developed with the University of Houston, is one of two applications on the Transporter-18 spacecraft. The other is a commercial imagery-processing application; SiliconANGLE identifies Stellerian as the company planning image processing on the system. The spacecraft comes from Indian builder TakeMe2Space.

The partnership list is longer than the revenue list. Payload reports memoranda of understanding with xOrbita and Symphony Space, and Space in Africa reports MoUs with the Kenya Space Agency and Angola’s space programme office, GGPEN. An MoU is a statement of intent and carries no committed payments.

What is the funding history?

Satlyt has disclosed one priced round in detail, the $8 million seed. Payload puts total funding at well past $10 million, which implies earlier capital of more than $2 million. The earlier rounds were not itemised in the coverage; existing backers made follow-on investments in the seed.

Who are the competitors?

Satlyt competes with companies building dedicated orbital computing hardware and with large technology groups running their own experiments. It also competes with the default option, which is for each satellite operator to write its own on-board software.

The hardware camp has attracted larger cheques. Payload notes that Satlyt positions its software as complementary to hardware-focused orbital data centres, and SiliconANGLE points out that Google launched an AI-powered satellite of its own the week before the round. TechCrunch reported in February 2026 that Sophia Space raised a $10 million seed to demonstrate space computers.

Vertically integrated operators are the other benchmark. SpaceX’s Starlink controls its own satellites, software and network, and Afullo has effectively conceded that market. He told Payload the goal is to “proliferate across basically every non-SpaceX Starlink satellite,” with a target of 30% of non-Starlink satellites in orbit by the end of the decade.

Earth-observation operators are the natural first customers because imagery is the heaviest data to downlink. Companies such as Planet Labs have long invested in their own data pipelines, which shows both the size of the need and the risk that the biggest operators will keep building in-house.

Why did investors back it?

The lead investor pointed to capital efficiency. A partner at non sibi ventures, identified in Space in Africa’s report as Lucas, said: “Satlyt is addressing that opportunity through deployments on third-party platforms.” Using other companies’ satellites removes the largest cost in the sector.

Compare the cost structures. A company building computing satellites must fund design, manufacturing, launch and replacement. Satlyt’s costs are mostly engineers. That makes an $8 million seed last longer and gives software-like gross margins if customers pay. It also makes the company dependent on hosts for flight opportunities.

Early evidence helped. By the time of the round the company had software running in orbit, a published result with a recognised AI model, a NASA contract and a second mission on the pad. Afullo summarised the job to Space in Africa: “Our job is to make those applications practical to deploy and operate.”

There is also a geographic thesis. Several of the investors focus on African technology companies, and a space software firm with engineering in Nairobi and MoUs with two African space agencies fits a portfolio that has few deep-tech options.

What does the Satlyt seed round signal for space seed investing?

It signals investor appetite for asset-light space companies that borrow enterprise software models. In a year when Crunchbase News data shows $20.3 billion flowing into space startups through August, much of it into capital-heavy hardware, a seed round for pure software is a deliberate contrast.

It also shows a broader investor base. None of the named participants is a traditional aerospace fund. Generalist and regional investors are underwriting space risk when it is packaged as software with a named customer.

Two other rounds in this series make useful comparisons. The Charter Space seed round is another asset-light bet, on insurance broking. The Project-S seed round takes the opposite approach and builds its own radar payloads. More deals are tracked in the space startup news archive.

For founders, the point to copy is the launch-day announcement backed by a working demo. The point to treat with caution is the ambition gap. A 30% share of non-Starlink satellites is a goal, and the evidence so far covers a handful of spacecraft.

What should customers and investors watch next?

Three things: the results of the Transporter-18 deployment, the first demonstration of computing shared across two satellites in 2027, and the conversion of memoranda of understanding into paid contracts. Each one tests a different part of the pitch.

The Transporter-18 mission tests whether two unrelated applications, one from NASA-funded research and one commercial, can run side by side on a single spacecraft. The 2027 goal tests the network idea that gives the company its “virtual data centre” label. The contracts test whether operators see enough value to pay. A seed round of $8 million gives the company time to answer all three, but probably not time to miss more than one of them before it needs to raise again.

Frequently Asked Questions

Sources

• Payload: Satlyt seed round coverage
• Space in Africa: Satlyt Raises USD 8 Million Seed to Turn Satellites into Virtual AI Data Centers
• SiliconANGLE: Satlyt raises $8M for satellite AI software
• Condia: Satlyt raises $8m to turn satellites in orbit into data centres
• Dealroom: Satlyt raises $8M seed to run AI on satellites without building its own
• TechCrunch: Sophia Space raises $10M seed to demo novel space computers
• Value Add VC: Space tech startup funding hits record $20.3B in 2026 (Crunchbase News data)