The largest orbit calculation cluster begins operations

Despite all the hype about data centers in space, there aren’t that many GPUs out there. As that begins to change, the near-term business of orbital computing is starting to take shape.

The largest compute cluster currently in orbit was launched in January by Canada’s Kepler Communications and includes about 40 Nvidia Orin edge processors on 10 operational satellites, all linked by laser communications links.

The company currently has 18 customers and announced its newest company, Sophia Space, on Monday. The startup plans to test the software for its own orbital computer aboard the Kepler constellation.

Experts predict that large-scale data centers like those envisioned by SpaceX and Blue Origin won’t emerge until the 2030s. The first step is to process data collected in orbit to improve the capabilities of space-based sensors used by private companies and government agencies.

Kepler doesn’t think of itself as a data center company, but as an infrastructure for applications in space, CEO Mina Mitry told TechCrunch. This wants to be the layer that provides network services to other satellites in space, or to drones and aircraft below.

Meanwhile, Sofia is developing a passively cooled space computer that addresses one of the key challenges for large data centers in orbit: building and building heavy, expensive active cooling systems to prevent powerful processors from overheating.

In the new partnership, Sophia will upload its own operating system to one of Kepler’s satellites and attempt to boot and configure it across the six GPUs on the two spacecraft. This kind of activity is very important in ground-based data centers, and this is the first time it has been attempted in orbit. Verifying that the software works in orbit will be an important risk mitigation exercise for Sophia ahead of its first satellite launch, planned for late 2027.

For Kepler, this partnership helps prove the network’s usefulness. It currently carries and processes data uploaded from the ground or collected by hosted payloads on its own spacecraft. However, as the field matures, the company expects to begin linking with third-party satellites to provide networking and processing services.

Mitry said satellite companies are now planning future assets around this model, pointing to the benefits of offloading processing for more power-hungry sensors like synthetic aperture radar. The U.S. military is a major customer for this type of work, as it develops new missile defense systems that rely on satellite threat detection and tracking. Kepler has already demonstrated a space-to-air laser link in demonstrations for the U.S. government.

This kind of edge processing, processing collected data to increase responsiveness, is where orbital data centers will first prove their value. This vision sets Sophia and Kepler apart from established space companies like SpaceX and Blue Origin, as well as startups like Starcloud and Aetherflux, which have raised significant funding to focus on large-scale data centers with data center-style processors.

“We believe in inference rather than training, so we want to have more distributed GPUs doing inference, rather than one super-powerful GPU with training workload capacity,” Mitry told TechCrunch. “If this consumes kilowatts of power and is only running 10% of the time, that’s not very useful. In our case, the GPU is running 100% of the time.”

And once these technologies are proven in orbit, anything can happen. Rob DeMillo, CEO of Sophia, noted that a city in Wisconsin adopted a data center construction ban last week that some members of Congress are also pushing for. Anything that limits data centers on Earth is what makes space-based alternatives more attractive in their eyes.

“There are no more data centers in this area. [city]”This is where things get weird,” Demilo muttered.