Publications Journal article · Orbital computing · 2026
Powering the Cloud Above the Clouds: Are Orbital Data Centers the First Real Customer for Space-Based Solar Power?
IEEE Aerospace and Electronic Systems Magazine, pp. 1–10, 2026
Abstract
Orbital data centers (ODCs) and space-based solar power (SBSP) are converging on the same orbital shell. Most ODC proposals target dawn–dusk sun-synchronous orbit because it almost never enters Earth's shadow, so a compute node can run without a large battery. That orbit is crowded, and it is poorly placed for latency-sensitive AI inference. This article asks whether SBSP can let compute leave it: sunlit relays would beam power to ODCs in other orbits, and orbital compute would become the first real in-space customer for SBSP. We frame the pairing as a fractionated architecture, the Sun-Side Cloud, and test it with an orbital, optical, and carbon model whose uncertain inputs are all set in the relay's favor. Three results bound the idea. First, relays in the dawn–dusk plane shrink an inclined client's eclipse but cannot erase it: for part of every precession cycle the middle of Earth's shadow is hidden from them, so clients keep a residual battery. Second, reaching that shadow core takes links of several thousand kilometers, where a sun-sized receiver captures only a small part of the beam. Third, once relay mass scales with the power it delivers, beamed power beats an onboard battery on carbon only above roughly 10–20% end-to-end efficiency and with tens of clients per fleet. Today's technology falls an order of magnitude short, so near-term value must come from the orbit itself. We turn these thresholds, together with the cooling and computing bottlenecks that no orbit removes, into a research agenda.
Cite
BibTeX
@article{fraire2026poweringcloud,
author = {Fraire, Juan A.},
title = {Powering the Cloud Above the Clouds: Are Orbital Data Centers the First Real Customer for Space-Based Solar Power?},
journal = {IEEE Aerospace and Electronic Systems Magazine},
pages = {1--10},
year = {2026},
doi = {10.1109/maes.2026.3742742}
}