The Rise of Orbital Computing: Space Data Centers Begin to Take Shape
For all the hype about data centers in space, there just aren’t very many GPUs up there. As that starts to change, the near-term business of orbital compute is beginning to take shape.
Leading the Way: Kepler Communications
Canada’s Kepler Communications has made a significant leap forward in the realm of space computing with the launch of the largest compute cluster currently in orbit. Initiated in January, this pioneering project houses approximately 40 Nvidia Orin edge processors onboard ten operational satellites, all interconnected through advanced laser communications links.
Growing Clientele and Innovative Partnerships
Kepler’s innovative efforts are already attracting attention from the private sector, with 18 customers utilizing their services. A notable partnership unveiled recently is with Sophia Space, a startup focused on testing software for unique orbital computers within Kepler’s satellite constellation. This collaboration exemplifies how private enterprises are beginning to explore the potential of space-based data processing.
Long-Term Vision and Current Capabilities
Experts predict that while large-scale data centers akin to those envisioned by giants like SpaceX or Blue Origin may not materialize until the 2030s, the initial steps are being taken. The focus will shift towards processing data collected in orbit, enhancing the functionalities of space-based sensors vital for both private industries and governmental operations.
Kepler’s Vision: Infrastructure for Orbital Applications
Kepler Communications perceives itself not merely as a data center but as essential infrastructure for various applications in space. CEO Mina Mitry emphasizes, “We want to be a layer that provides network services for other satellites in space, or drones and aircraft in the sky below.” This foundational approach underlines Kepler’s goal to create an interconnected network that fosters broader technological advancements.
Sophia Space: Tackling Cooling Challenges in Orbit
Sophia Space, in contrast, is focused on overcoming a critical obstacle for large-scale orbital data centers: the challenge of keeping powerful processors from overheating. The company is developing passively-cooled space computers that could eliminate the need for heavy active-cooling systems, which are costly to build and launch.
Groundbreaking Testing Ahead
In their recent partnership, Sophia will upload its proprietary operating system onto one of Kepler’s satellites, aiming to launch and configure it across six GPUs on two separate spacecraft. This endeavor is a significant first, as similar operations are commonplace in terrestrial data centers, yet untried in the challenging environment of orbit. Ensuring the software functions correctly in space will serve as a crucial risk-management step for Sophia, especially before its inaugural satellite launch planned for late 2027.
Proving Utility and Future Collaborations
Kepler’s partnership with Sophia also demonstrates the practical utility of its network. Currently, the satellite network processes data uploaded from Earth or collected by onboard payloads. As operations mature, Kepler anticipates collaborating with third-party satellites to offer networking and processing services, paving the way for enhanced satellite functionality.
Strategic Developments with Military Applications
With satellite companies now designing future projects around this operational model, the benefits of offloading processing for energy-intensive sensors, like synthetic aperture radar, have become apparent. Notably, the U.S. military is among the early adopters of these services, particularly as it advances missile defense capabilities that rely on satellite detection and threat tracking. Kepler has already demonstrated a successful space-to-air laser link in a government demonstration.
Future Outlook: Adapting to Market Needs
This edge processing—managing data right at the point of collection—will likely underscore the initial value of orbital data centers. Kepler and Sophia’s approach significantly distinguishes them from established space firms like SpaceX and Blue Origin, and from startups like Starcloud and Aetherflux, which are targeting large-scale data centers utilizing conventional data center-style processors.
A Shift in Paradigm
Mina Mitry argues for a paradigm shift in processing needs. “Because we believe it’s more inference than training, we want more distributed GPUs that handle inference rather than relying on one super-power GPU for heavy training workloads,” he reveals. This operational efficiency means that their GPUs can be fully utilized, addressing energy concerns inherent in traditional models.
Looking Ahead: Space as the New Frontier
As technology progresses, the orbital sector could become an increasingly attractive alternative for data center operations on Earth. Sophia CEO Rob DeMillo notes a recent ban on data center constructions in Wisconsin, reflecting broader regulatory trends. “There’s no more data centers in this country,” DeMillo quips, hinting at an uncertain but intriguing future for space-based computing.
As the landscape of orbital computing evolves, it opens up a realm of possibilities, marking the dawn of a new chapter in data processing and application development in space.
Image Credit: techcrunch.com
