Kepler Communications Opens The Largest Commercial Orbital Compute Cluster Ever Built
Kepler Communications just announced Sophia Space as its 18th customer, the first time a startup will attempt to upload, launch, and configure its own operating system across GPUs on satellites in orbit.
Canada’s Kepler Communications launched the world’s largest commercially operational orbital compute cluster aboard a SpaceX Falcon 9 in January 2026, just days after Falcon’s successful first launch, and as of April 13, it is now open for business.
The cluster comprises approximately 40 NVIDIA Jetson Orin edge processors across 10 satellites, interconnected via real-time laser optical inter-satellite links. This setup lets computing tasks move between satellites and scale across the network instead of being limited to a single machine.
Kepler has signed 18 customers and named Sophia Space, a Pasadena-based orbital computing startup, as its latest partner, marking the first attempt to upload, deploy, and run a proprietary operating system across live GPUs in orbit.
What Sophia Space Is Actually Testing in Orbit
Sophia Space raised a $10 million seed round in February 2026 to build TILE, Thermal Integrated LEO Edge, modular one-meter units with built-in solar panels and a passive heat-spreading design that removes the need for active cooling systems.
As TechCrunch reported, this directly tackles a core challenge highlighted by NVIDIA CEO Jensen Huang: in orbit, there is no airflow, so the processor heat must dissipate through conduction.
Sophia’s design places processors against a passive heat spreader, enabling high-density compute within strict satellite size, weight, and power limits.
In its partnership with Kepler, Sophia will deploy its proprietary operating system to a Kepler satellite and attempt to configure it across six GPUs on two spacecraft.
This kind of remote deployment is standard procedure in terrestrial cloud environments, but has never been executed in orbit. For Sophia, this test is a key de-risking step ahead of its first planned satellite launch in late 2027, when it aims to begin delivering TILE modules to customers.
Why Kepler Sees Itself as Infrastructure, Not a Data Center
The positioning is both technical and commercial. As TechCrunch reported, Kepler CEO Mina Mitry said the company does not see itself as a data center operator but as a network infrastructure layer.
Its goal is to provide processing and connectivity to other satellites, as well as to drones and aircraft operating below.
As the market evolves, Kepler plans to connect third-party satellites to its optical network. Mirroring NVIDIA’s $4 billion optical expansion, it offers networking and processing as a managed service instead of requiring customers to launch their own hardware.
GuruFocus reinforced this view, noting the partnership allows Sophia to use Kepler’s existing constellation to test and validate its hardware and software well before its own satellites launch.
For Kepler, each partnership demonstrates real-world demand for its network. For Sophia, using a live orbital constellation as a test bed shortens what would otherwise be a years-long validation timeline into months of real-world data.
Where Orbital Compute Is Actually Useful Right Now
As TechCrunch reported, industry experts do not expect large-scale space-based data centers, comparable to those supported by Elon Musk’s Austin Terafab project, until the 2030s.
In the near term, orbital computing focuses on processing data in space instead of waiting for satellite downlinks to Earth, reducing latency for time-sensitive use cases.
Mitry told TechCrunch that future satellites are already being designed around this model, especially for power-hungry sensors like synthetic aperture radar that produce more data than can be sent down in raw form.
The U.S. military is one confirmed customer type, building missile defense systems that rely on real-time satellite tracking. Kepler has also demonstrated a space-to-air laser communication link in a government-focused test.
Sophia CEO Rob DeMillo pointed to a broader shift, noting that some U.S. states are restricting new data center construction, and similar proposals are emerging at the federal level. He added that new data centers are being restricted in the country, and the situation is becoming more unpredictable.
