Sunday, November 1, 2026

Culture. Creativity. Influence

Kenyan-founded Satlyt raises $8 million to build AI-powered data centres in space

Founded by Kenyan-American engineer Rama Afullo, Satlyt has raised $8 million to expand software that runs artificial intelligence directly on satellites. With operations in Nairobi and Silicon Valley, the startup is developing the foundations of a shared computing network in orbit.

4 min read

For decades, satellites have been designed to collect information in space and send it back to Earth. Images, measurements and technical data travel through communications networks before computers on the ground turn them into something useful.

A Kenyan-founded technology company wants to change that relationship.

Satlyt, a space software startup operating from Nairobi and Sunnyvale, California, has raised $8 million to develop technology that allows satellites to process information and run artificial intelligence models while still in orbit.

Founded by Kenyan-American engineer Rama Afullo, a former Google and SpaceX employee, the company is working towards what it calls virtual AI data centres in space.

Its approach is different from that of companies proposing to launch enormous new computing facilities beyond Earth. Satlyt wants to make greater use of the processors already installed on satellites, allowing operators to run software on existing spacecraft and eventually share computing capacity between them.

The company announced its funding on October 1, 2026, after demonstrating elements of its technology in orbit.

An $8 million bet on computing beyond Earth

Satlyt's latest financing round was led by Houston-based investment firm Non Sibi Ventures.

Other investors included TLcom Capital, Antler, Slauson & Co., Launch Africa Ventures, Enza Capital, Askya Investment Partners, AXIAN Investment, Demos, BAG Collective and Gaingels, alongside existing backers.

The funding is intended to expand the company's engineering and customer-delivery teams and support additional software deployments on third-party spacecraft.

Unlike satellite manufacturers, Satlyt is not primarily trying to design and launch its own fleet.

Its business is software.

The company develops a computing layer through which satellite operators can install, manage and execute applications on the hardware already aboard their spacecraft.

This approach could allow more operators to use artificial intelligence in orbit without building entirely new satellite systems.

The distinction is also important financially. Developing software for existing spacecraft presents different investment requirements from constructing orbital data centres or manufacturing a large satellite constellation.

The $8 million round supports the next stage of Satlyt's development; it is not a budget for launching a fully operational network of space-based data centres.

The engineer who took his idea beyond Google and SpaceX

Satlyt's story begins with Rama Afullo, an engineer whose career has crossed some of the world's most influential technology companies.

Afullo worked in Google's cloud computing business before joining SpaceX's Starlink operation in 2024.

Those experiences exposed him to two rapidly developing technological fields: cloud infrastructure and satellite communications.

He became convinced that the computing systems aboard satellites could be used more extensively.

In an October 2026 interview with TechCrunch, Afullo explained that he had raised the idea internally at both Google and SpaceX but had not persuaded either company to pursue his proposed approach.

He subsequently founded Satlyt in 2024.

The startup maintains a presence in both Silicon Valley and Nairobi, connecting the American space industry with engineering talent and technology networks in Kenya.

That geographic structure is central to its identity.

Satlyt is not a Kenyan government space programme, nor is it developing a national satellite constellation. It is a private technology company founded by a Kenyan-American entrepreneur with operations on two continents.

Its international ambitions are reflected in both its investor base and its technical partnerships.

Why satellites need artificial intelligence

Modern satellites collect enormous quantities of information.

Earth-observation spacecraft produce imagery used for agriculture, environmental monitoring, mapping and disaster response. Other satellites collect scientific measurements or monitor the health of their own systems.

Much of that information must be transmitted to ground stations before it can be processed.

But communication opportunities can be limited, particularly when a satellite passes over a ground station for only a short period.

Transmitting large amounts of raw data also consumes bandwidth, electricity and operational time.

Onboard processing offers an alternative.

A satellite equipped with suitable computing hardware can analyse information before transmitting it, selecting important observations or summarising technical problems.

An Earth-observation satellite, for example, could process an image and identify information relevant to a specific application rather than sending every pixel to the ground immediately.

A spacecraft could also analyse technical logs to help operators understand a malfunction.

In both cases, some of the work traditionally performed on Earth moves closer to where the data originates.

This is a form of edge computing: processing information near its source.

Putting Google's Gemma model into orbit

Satlyt has already demonstrated this approach using software deployed on a spacecraft operated by Momentus.

The system ran a compact model from Google's Gemma family to analyse technical information associated with the satellite's software.

During a demonstration involving simulated software faults, the AI system processed diagnostic logs and generated shorter summaries for transmission to Earth.

According to the company's reported results, the approach reduced the size of diagnostic information requiring transmission by more than 64% in the tested scenarios.

The experiment illustrated one potential benefit of onboard intelligence: reducing the volume of data that operators need to receive and interpret.

However, the result should be understood precisely.

It concerned specific diagnostic tests, not a verified reduction of 64% across every type of satellite transmission.

It also did not establish that the software could independently repair every malfunction or operate without human supervision.

Nevertheless, the demonstration provides evidence that Satlyt's technology has moved beyond software running exclusively in terrestrial laboratories.

Its applications have already been tested in the environment for which they are being developed.

The ambition to connect satellites into one computing network

Running artificial intelligence on a single satellite is only the first part of Satlyt's strategy.

The company ultimately wants to connect computing resources aboard different spacecraft, allowing them to share workloads.

Today, many satellites operate as relatively independent systems, even when they belong to the same constellation.

Satlyt is developing software intended to make those resources easier to coordinate.

Its long-term objective resembles the architecture of cloud computing on Earth, where computing tasks can be distributed across servers instead of remaining on one machine.

The difference is that the servers would be moving through orbit, subject to changing communication links, limited power supplies and the constraints of space hardware.

This introduces substantial engineering challenges.

Satellites are not continuously connected to one another. Their computing hardware varies, and their energy budgets are limited.

Software must also operate reliably despite interruptions and delays in communication.

Satlyt is working towards a demonstration involving computing across two separate satellites, which it has identified as a goal for 2027.

That milestone remains ahead. The existence of onboard software demonstrations should not be confused with proof that a complete multi-satellite cloud is already operational.

Why investors are watching orbital computing

Satlyt is emerging at a time when several major technology companies are exploring the possibilities of computing in space.

Advances in artificial intelligence are driving demand for processors, electricity and data-centre capacity on Earth.

Space-based computing offers a different set of opportunities and constraints.

Satellites already carry increasingly capable processors. Some tasks can be performed more efficiently when data is analysed directly in orbit.

Larger proposals also explore the possibility of using orbital infrastructure for more demanding AI workloads.

But these ideas face difficult questions involving launch costs, radiation exposure, thermal management, equipment replacement and communications.

Satlyt's strategy is more immediately focused on existing satellite hardware.

Rather than assuming a large new orbital computing industry must be built before it can serve customers, the startup aims to enable operators to extract greater value from spacecraft already being launched.

Non Sibi Ventures' investment reflects confidence in that narrower opportunity as well as the company's longer-term ambitions.

Bernard Harris, a former NASA astronaut, is associated with the investment firm and its assessment of space technologies.

The financing gives Satlyt resources to continue testing its software and developing commercial applications, although its eventual revenue and market share remain uncertain.

A Kenyan-founded company in the global space economy

Satlyt's development adds another dimension to the growing international presence of African technology entrepreneurs.

Much of the attention directed towards Africa's startup ecosystem has historically focused on financial technology, mobile services, e-commerce and digital infrastructure.

Space computing represents a different field, requiring specialised engineering and relationships with satellite operators, launch providers and research institutions.

The company's presence in Nairobi connects Kenya's technology ecosystem with an industry often associated with Silicon Valley, European aerospace centres and major national space agencies.

It also challenges the assumption that African participation in the space economy must be limited to purchasing satellite imagery or operating locally owned spacecraft.

Software developed by companies with African roots can contribute to the technologies used in international missions.

For Satlyt, the geographical reach is already built into its business model: its software is intended for satellite operators regardless of where their spacecraft are designed or launched.

This does not make Satlyt exclusively Kenyan. Its American headquarters, international investors and cross-border team are equally important to its structure.

Its significance lies in the participation of Kenyan-founded engineering talent in a highly specialised global technology market.

From demonstration to commercial adoption

The next challenge for Satlyt will be converting its technical achievements into a reliable commercial service.

Satellite operators need software that can function consistently on different kinds of hardware, integrate with existing systems and meet demanding reliability requirements.

They also need evidence that onboard processing produces sufficient benefits to justify implementation.

Reducing communication requirements may be valuable, but the size of those benefits will depend on the mission and the applications involved.

An Earth-observation company processing large images may have different needs from a scientific satellite monitoring a small number of instruments.

Satlyt's software must also be able to operate within the power, memory and thermal limits of spacecraft.

The company has announced plans for further deployments and new customer applications.

It has not publicly demonstrated a universal computing network linking large numbers of operational satellites.

Its future success will depend on the results of additional missions, the adoption of its tools by satellite operators and its ability to maintain reliable services over time.

An African connection to the future of space computing

Satlyt's $8 million financing is significant because it supports an approach to artificial intelligence that moves computing away from conventional data centres and into the infrastructure already orbiting Earth.

The startup is not claiming to have completed a vast network of orbital data centres. It is building the software that could eventually allow such networks to function across different spacecraft.

Its early demonstrations provide a foundation, while the larger vision remains a technical and commercial challenge.

For Rama Afullo, the company represents an opportunity to pursue an idea that did not gain traction during his time at Google and SpaceX.

For the African technology ecosystem, its rise provides another example of entrepreneurial ambition extending into advanced global industries.

And for the space sector, Satlyt is working towards a future in which satellites do more than collect information.

They could become part of the computing infrastructure that processes it.

Found this useful? Pass it on.

Send it to someone who would want the context.

The NationBrand Edit.

One considered edition, every weekday morning.

One considered edition, every weekday morning.

One considered edition, every weekday morning.

A curated perspective on culture, creativity and influence.

One email each weekday. Unsubscribe whenever you like.

© 2026 NationBrand. All rights reserved.

Remix for free

Create a free website with Framer, the website builder loved by startups, designers and agencies.