The largest orbital compute cluster is open for business

In a groundbreaking development for space technology, the largest compute cluster in orbit has officially commenced operations. Launched by Kepler Communications from Canada in January, this innovative system consists of 10 satellites equipped with approximately 40 Nvidia Orin edge processors, all interconnected through advanced laser communication links. Currently, Kepler has secured 18 clients, with its latest partnership announced recently with Sophia Space, a startup dedicated to testing its unique software on Kepler's satellite constellation. Experts predict that large-scale data centers in space, as envisioned by companies like SpaceX and Blue Origin, may not come to fruition until the 2030s. The initial focus will instead be on processing data gathered in orbit to enhance the functionality of space-based sensors utilized by both private enterprises and government entities. Kepler's CEO, Mina Mitry, emphasizes that the company does not position itself as a conventional data center. Rather, it aims to serve as a vital infrastructure layer for applications in space, enabling services for other satellites and air-based drones. Meanwhile, Sophia Space is innovating with passively-cooled space computers to address one of the major challenges for orbital data centers: preventing overheating of high-powered processors without the need for bulky and costly active-cooling systems. The new collaboration involves Sophia uploading its proprietary operating system to one of Kepler’s satellites, aiming to configure it across six GPUs on two different spacecraft. This endeavor mirrors standard practices in terrestrial data centers but marks a significant milestone as it will be the first attempt of its kind in outer space. Ensuring the software's functionality in orbit is crucial for Sophia as it prepares for its inaugural satellite launch set for late 2027. For Kepler, this partnership is instrumental in validating the utility of its network capabilities. The company currently processes data transmitted from the ground or collected from payloads on its satellites. As the industry evolves, Kepler anticipates forming connections with third-party satellites to offer enhanced networking and processing services. Mitry highlights a growing trend among satellite companies to design future assets based on this model, particularly for offloading processing tasks from power-intensive sensors like synthetic aperture radar. The U.S. military is among the key stakeholders interested in this technology as it develops a new missile defense system reliant on satellite detection and tracking. Kepler has already successfully demonstrated a space-to-air laser link during a trial for the U.S. government, showcasing the potential of edge processing—an approach that processes data at its source for quicker response times. This focus distinguishes Kepler and Sophia from established entities like SpaceX and Blue Origin, as well as emerging firms like Starcloud and Aetherflux, which are pursuing significant investments for large-scale data centers equipped with traditional data center processors. Mitry asserts the importance of distributing GPU capabilities for inference tasks rather than relying on a singular, powerful GPU for training workloads. He notes that their GPUs are utilized continuously, optimizing efficiency and power consumption. As these technologies prove their effectiveness in space, the future holds limitless possibilities. Sophia's CEO, Rob DeMillo, reflects on recent developments, including Wisconsin's new ban on data center construction, which some lawmakers in Congress are advocating. This shift may make space-based alternatives increasingly appealing, as DeMillo quips, "There’s no more data centers in this country. It’s gonna get weird from here."