Driven by Earth-observation data growth, terrestrial energy limits and tightening sovereign data regulations, orbital data centers are emerging as a serious extension of the global compute stack. ABI Research Principal Analyst Andrew Cavalier explains why factors like in-orbit processing and zero-water cooling could reshape both the space economy and the future of data infrastructure.
Q: What’s driving the need for orbital data centers? What’s the potential value of moving data processing to space?
A: The demand is driven by the collision of three exponential trends: the downlink bottleneck of Earth observation (EO), the energy crisis of terrestrial AI and the rise of strict “sovereign data” laws.
From an EO data perspective, we are generating petabytes of data in orbit, but the download throughput is a challenge. It could be potentially more efficient to process at the edge in space. Second, terrestrial grids are hitting a breaking point. Starcloud and Crusoe Energy are pioneering a model where high-density AI training loads are offloaded to orbit to harvest “stranded” solar energy stuck in orbit—effectively bypassing Earth’s carbon limits.
Finally, there is a regulatory arbitrage play. Axiom Space and others are exploring “orbital data vaults” that turn regulated data into “unregulated insight” before it ever touches a border. This allows a space-based data center operator to be not restricted by regional and local data center compliance but also take advantage of global connectivity.
In the long term, this looks like an ecosystem play. It’s not just about storage or processing; it’s about an orbital economy. SpaceX reduces the launch cost to deploy these assets. Starcloud and Crusoe are already partnering to offload high-density AI workloads to orbit to bypass the constraints of Earth’s terrestrial energy grid. Meanwhile, Intuitive Machines is building the lunar data relay network (NSNS contract) that will eventually pipe data from the lunar surface back to these orbital nodes, creating a continuous “Earth-Moon digital highway.”
Q: What makes orbital data centers potentially more efficient than ground processing? How would you envision these facilities being used?
A: To leverage the potential of physics. Space has “free” resources with continuous solar energy and passive cooling. Terrestrial data centers use a significant amount of power, with upwards of 40% or more of total energy consumed on active cooling (fans and chillers). In orbit, we can radiate waste heat directly into the cold void of space. While this requires significant radiator surface area, it eliminates the massive electricity cost of terrestrial cooling, theoretically offering higher energy efficiency.
I could see a number of use cases for these facilities from edge AI for EO and LLM training runs before being sent down to Earth, to sovereign compute (data vaults) like what Axiom space is exploring, to telemetry and navigation support lunar logistics like with Astrolab’s FLEX rover in support with Intuitive Machines data relay.
Q: How realistic is the vision of space-based data centers achieving energy or sustainability advantages over terrestrial data centers?
A: The zero-water advantage is highly realistic because the baseline on Earth is unsustainable. A medium-sized terrestrial data center cycles through millions of gallons of water annually for cooling. An orbital data center consumes zero water. There is also the energy perspective, which Google’s Project Suncatcher research suggests that a constellation of solar-powered satellites using optical links could effectively create a “virtual” gigawatt-scale data center without adding a single gram of carbon to Earth’s atmosphere.
Even so, we would need to evaluate the logistics of getting a substantial amount of data center capacity into orbit, if the objective is to operate data centers in space as a substantial complement to data centers on Earth. Nonetheless, data center computational capacity will be needed for in-orbit operations, lunar operations, and potentially even deep space operations.
Q: What are the potential hurdles to deployment?
A: Again, physics. I did a post on LinkedIn about this and the biggest hurdle isn’t getting there; its keeping the “chips” alive. The thermal management issue in particular is a challenge since dissipating heat is notoriously difficult for high-density GPUs. Furthermore, radiation—cosmic rays—rapidly degrades silicon, creating a shorter hardware cycle than on Earth. This necessitates in-orbit servicing models since maintaining the racks in orbit is a challenge. We anticipate a need for autonomous logistics vehicles—mirroring the model Astrolab is deploying for lunar surface logistics—to robotically swap degraded server components and maintain the constellation.
Despite these engineering hurdles, ABI Research identifies this as a critical growth vector in the New Space economy. We are tracking significant market momentum in this sector and will be publishing our comprehensive analysis on the orbital compute market in H1 this year.
