Elon Musk is pushing a new idea for orbital computing. Earlier this month, he said upgraded Starlink V3 satellites could be scaled into space-based data centres. The remark came in response to a post about autonomous in‑orbit assembly.
“SpaceX will be doing this,” said Elon Musk. The statement set no delivery date and offered few design details.
Rival visions are forming around the same goal. Jeff Bezos has argued that continuous solar power makes orbit a better home for heavy AI compututation.
“We will be able to beat the cost of terrestrial data centres in space in the next couple of decades,” said Jeff Bezos.
Blue Origin’s work remains largely under wraps. Yet the message is consistent, orbital infrastructure could ease land, water, and grid constraints. Competitive posturing is now shaping expectations.
Technical hurdles and regulatory paths
Engineering realities will decide timelines. High radiation, thermal cycling, and micrometeoroids demand hardened compute and storage. Latency to Earth still limits real‑time uses, so edge tasks like caching or pre‑processing look more practical. Orbital throughput would rely on laser crosslinks and high‑capacity downlinks to dense ground station networks.
Spectrum access and interference management add further complexity under the International Telecommunication Union’s coordination rules. Sustainability norms also apply, guided by the United Nations’ long‑term space sustainability guidelines.
Starlink already flies with optical crosslinks, a base for inter‑satellite networking. Turning communications nodes into compute nodes adds power, cooling, and shielding demands. It also raises mass and integration trade‑offs. Launch economics and on‑orbit assembly techniques would shape the architecture. Any scalable design must simplify replacement and deorbit at end of life.
The appeal is clear, space offers abundant solar power and passive radiative cooling. That could limit water use and local grid stress tied to terrestrial campuses. Even so, orbital data centres would not stand alone.
They would lean on ground stations, terrestrial fibre backbones, and regional interconnects. Data residency, export controls, and security reviews would frame where downlinks land and how traffic is routed. Procurement teams and regulators face a new question, how to price and permit compute that start above the Kármán line.