Orbiting data centers could be powered by near-continuous sunlight in sun-synchronous orbits, using radiators to reject heat into space and potentially avoiding terrestrial cooling constraints like water use and transformer bottlenecks.

Critical engineering and risk factors include protecting hardware from radiation, achieving precise inter-satellite alignment and fault tolerance, managing debris and collision risks, and addressing environmental and regulatory concerns highlighted by oversight bodies.

Economic viability hinges on launch costs; if prices drop below roughly $200 per kilogram to LEO by the mid-2030s, orbital compute could approach terrestrial data-center energy costs per kWh, though cost and reliability barriers remain.

Communications pose a major bottleneck, requiring inter-satellite links at tens of terabits per second and downlink performance that is weather-dependent, which drives the need for hybrid optical-radio networks and robust ground infrastructure.

A central technical challenge is radiator area in space, with 1 MW systems potentially needing thousands of square meters of radiators and scaling to hundreds of thousands of square meters for 100 MW.

Analysts note that a 100-megawatt hyperscale AI data center consumes the electricity of roughly 100,000 households, with global data-center electricity demand projected to rise from about 415 TWh in 2024 to around 945 TWh by 2030.

Early viable workloads for orbital data centers are likely space-based data processing at collection points, such as imaging and radar, to reduce Earth-bound data transmission and enable in-space inference rather than full model training.

The article cautions that orbital data centers are not a guaranteed energy or cooling solution; initial deployments will likely be small, space-adjacent processors and sensors, advancing only as networking, thermal management, and economic viability prove reliable.