Frequently asked questions about data centres in space

Yes — as of December 2025 there is one operational orbital data centre: Starcloud-1. Launched in November 2025, Starcloud-1 is a single satellite deployed to low Earth orbit (≈325 km) carrying an Nvidia H100-class GPU and systems designed for AI training and inference. Early reported workloads include NanoGPT training runs and Gemma inference.

Potential uses include processing data from Earth-observation satellites in orbit, providing compute close to space infrastructure (like lunar bases or deep-space missions), military and intelligence applications, specialised AI training and inference near space sensors, and experimental high-availability systems that are physically off-planet.

It’s a common misconception. Space is a vacuum, which makes convection cooling impossible. On Earth, your data centre dumps heat into the air or water. In space, you mostly rely on radiation via large radiators, which is technically challenging, heavy, and expensive.

A mix of startups, national space agencies, defence organisations, and cloud providers could all play a role. For example, China has already launched an initial cluster (12 satellites in May 2025) described as the start of a 'Three-Body Computing Constellation' but that cluster still needs to scale to thousands of nodes to reach the kind of distributed supercomputer those announcements implied. Several private startups (Aetherflux, OrbitsEdge, Orbit AI, etc.) have public timelines — Aetherflux targets Q1 2027 for a first node, while OrbitsEdge and Orbit AI have indicated 2026 launches — but many of these remain proposals or early prototypes.

It’s complicated. You could theoretically power orbital infrastructure with abundant solar energy, but launches have an environmental cost, and building/maintaining hardware in space is resource-intensive. Whether this is net-positive depends on the scale, technology, and lifecycle accounting versus terrestrial alternatives.

Latency depends on orbit and use case. For many typical web or app workloads, a data centre in space would introduce more latency than a well-placed terrestrial region. Space data centres make more sense for workloads tied to space assets or specialised tasks where physical location is an advantage.

It’s already happened at small scale: Starcloud-1 (Nov 2025) is operational. broader, production-grade fleets or distributed in-orbit supercomputers are still expected to take several years (multi-stage scaling and regulatory issues complicate timelines).

No. datacentresinspace.com is an independent project created to track and explain the idea of off-planet compute. If more genuine space data centres are launched, this counter and the FAQ will be updated.

Analysts project rapid market growth driven by AI and edge compute needs — some forecasts put the in-orbit data centre market around $1.78 billion by 2029 and expanding towards ~$39 billion by 2035. today, however, deployments are largely experimental and single-node demonstrations.

There are over 11,700 active satellites orbiting Earth. despite that large number, only Starcloud-1 is currently being marketed and recognized as an operational data centre as of December 2025.