Despite significant anticipation surrounding space-based data centers, the presence of Graphics Processing Units (GPUs) in orbit remains limited. However, this landscape is evolving, and the nascent business of orbital computing is beginning to solidify.
Canada's Kepler Communications launched the most extensive compute cluster currently orbiting Earth in January. This system features approximately 40 Nvidia Orin edge processors distributed across 10 operational satellites, interconnected via advanced laser communication links.
Kepler Communications has expanded its client base to 18 customers, with the latest addition announced on Monday: Sophia Space. This startup plans to utilize Kepler's constellation to test the software for its innovative orbital computer.
Industry experts project that large-scale data centers, akin to those conceptualized by SpaceX or Blue Origin, are unlikely to materialize before the 2030s. The immediate progression involves processing data gathered in orbit to augment the performance of space-based sensors employed by both private enterprises and government entities.
Kepler Communications does not define itself as a data center provider, but rather as foundational infrastructure for in-space applications, as stated by CEO Mina Mitry to TechCrunch. The company aspires to establish a networking layer, offering services to other satellites in orbit and to drones and aircraft within Earth's atmosphere.
Conversely, Sophia Space is innovating passively-cooled space computers, a development poised to address a critical hurdle for large-scale orbital data centers: preventing powerful processors from overheating without the need for cumbersome and costly active-cooling systems.
Under this new collaboration, Sophia will upload its proprietary operating system to one of Kepler's satellites, aiming to launch and configure it across six GPUs spanning two spacecraft. While such an operation is standard practice in terrestrial data centers, this marks its inaugural attempt in orbit. Ensuring the software's functionality in space represents a crucial de-risking step for Sophia in advance of its own satellite launch, scheduled for late 2027.
For Kepler, this partnership serves to validate the efficacy of its network. Presently, the company manages and processes data transmitted from Earth or gathered by hosted payloads on its own spacecraft. However, as the space sector evolves, Kepler anticipates establishing interconnections with third-party satellites to extend its networking and processing capabilities.
Mitry notes that satellite companies are now designing future assets around this operational model, recognizing the advantages of offloading processing tasks for energy-intensive sensors, such as synthetic aperture radar. The U.S. military is a significant client for these capabilities, particularly in its development of a new missile defense system reliant on satellite-based threat detection and tracking. Kepler has already successfully showcased a space-to-air laser link in a demonstration for the U.S. government.
This form of edge processing — handling data at its point of collection to enable swifter responsiveness — is precisely where orbital data centers are expected to initially demonstrate their utility. This strategic focus distinguishes Sophia and Kepler from established aerospace giants like SpaceX and Blue Origin, as well as startups such as Starcloud and Aetherflux, which are securing substantial funding to pursue large-scale data centers equipped with conventional data center processors.
"Our conviction is that the demand will lean more towards inference than training, hence our preference for a greater number of distributed GPUs optimized for inference, rather than a single, high-capacity GPU designed for training workloads," Mitry informed TechCrunch. He elaborated, "If a system consumes kilowatts of power but operates only 10% of the time, its utility is significantly diminished. In contrast, our GPUs are active 100% of the time."
Once these advanced technologies are validated in orbit, a myriad of possibilities could emerge. Sophia CEO Rob DeMillo highlighted Wisconsin's recent ban on data center construction, a sentiment echoed by some lawmakers in Congress. He suggests that any terrestrial restrictions on data centers inherently enhance the appeal of space-based alternatives.
"There's no more data centers in this country," DeMillo mused, anticipating future developments. "It's gonna get weird from here."
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.