Artificial intelligence and cloud computing are accelerating data center construction, but developers are increasingly constrained by limited power, water availability, rising costs and community pushback. Those pressures are prompting serious consideration of a concept once relegated to science fiction: placing data centers in orbit.
A recent JLL report, discussed with Connect CRE by Andrew Batson, the firm’s global head of data center research, examines how orbital data centers could eventually complement traditional campuses on Earth. In this model, space-based facilities would focus on energy-intensive workloads such as AI training, scientific simulations and satellite data processing, allowing terrestrial sites to prioritize latency-sensitive applications.
The idea builds on existing space-based computing. The International Space Station already operates CPUs and GPUs, though on a much smaller scale than the single orbital data centers now being proposed, which Batson notes are envisioned as larger than the entire ISS. Major technology companies are testing pieces of this architecture: SpaceX has requested FCC approval for a constellation of data center satellites, AWS Ground Station is building infrastructure to link Earth and orbital workloads, and Microsoft’s Azure Orbital and Google’s Project Suncatcher are exploring cloud connectivity and AI model training in orbit.
Orbital data centers could run on continuous solar power, bypassing grid interconnection, backup generation and lengthy permitting. JLL cites estimates from aerospace startup Starcloud that a 40-megawatt orbital cluster could save about $138 million in energy costs over 10 years versus an equivalent Earth-based facility. However, the economics depend heavily on launch pricing, with Batson noting that orbital data centers only become cost-competitive if launch costs fall below $500 per kilogram.
Engineering challenges remain substantial. Cooling must be achieved through large radiator systems because the vacuum of space eliminates conventional air-based solutions, increasing equipment size and upfront capital requirements. Operators would also need to manage hardware obsolescence and the safe disposal of retired assets to avoid adding to orbital debris.
Processing data in space offers another potential advantage. Instead of downlinking massive volumes of raw satellite imagery and sensor data, orbital facilities could perform edge processing and transmit only refined outputs, particularly for use cases that do not require immediate response times.
Batson expects orbital facilities to augment rather than replace traditional data centers, but notes that the shift could create new demand patterns for commercial real estate. JLL points investors toward potential growth in aerospace manufacturing, logistics space near launch sites and related support infrastructure, alongside emerging needs such as space insurance and debris management. A key indicator for the sector may be Google’s Project Suncatcher; if it validates AI model training performance and cost targets by late 2027 or early 2028, Batson anticipates faster investment moves from AWS, Microsoft Azure and other hyperscalers.
Whether orbital data centers become commonplace remains unresolved, but the analysis frames them as a potential response to the physical and political limits of building ever more capacity on Earth.
