SpaceX’s ‘orbital data centre’ idea: what the 1m-satellite filing really means

/ Technology / By

Summary: SpaceX has filed an application with the US Federal Communications Commission (FCC) to deploy up to one million additional satellites in low‑Earth orbit. The pitch is unusual: satellites as part of an “orbital data centre” concept designed to support the growing compute requirements of artificial intelligence.

What SpaceX is asking for

  • Permission to operate a new constellation that could reach up to 1,000,000 satellites.
  • Satellites would operate in low‑Earth orbit at roughly 500–2,000 km altitude.
  • SpaceX says the system could provide compute capacity for “billions of users globally.”

SpaceX already operates Starlink (nearly 10,000 satellites). This would be an order‑of‑magnitude step change.

What “orbital data centres” actually implies

On the ground, a data centre is mainly:

  • racks of servers (compute + storage)
  • power delivery
  • networking
  • cooling

Putting compute in orbit changes the physics:

  • Cooling is harder, not easier. In space you can’t rely on air convection; you have to reject heat via radiators.
  • Servicing and upgrades are expensive. A failed server in a warehouse is swapped; in orbit it’s a replacement launch.
  • Power is different. Solar is abundant in orbit, but high‑density compute needs consistent power and thermal stability.

So the claim isn’t “space is easy,” it’s that—at very large scale—some parts of the energy/cooling trade‑off might look better than building ever‑larger terrestrial facilities.

Why the AI angle matters

AI workloads (training and large‑scale inference) are driving:

  • new data centre buildouts
  • grid demand and long-term power contracts
  • huge spending on GPUs and networking

SpaceX is effectively saying: if compute demand keeps growing, the bottleneck becomes energy + infrastructure, and orbit could become one more place to put capacity.

The biggest concerns: congestion, debris, and astronomy

A constellation of this size would intensify existing debates around mega‑constellations:

1) Orbital safety and congestion

More satellites means more close approaches and a higher risk of collision cascades (debris creating more debris). Even with good tracking, the environment gets harder to manage as density increases.

2) Space debris risk to hardware

More objects in orbit increases the chance of impacts that can damage satellites and create long-lived fragments.

3) Impact on astronomy

Astronomers have complained that satellite networks can interfere with observations—including radio emissions that may affect sensitive instruments.

What’s notable (and what’s missing)

  • The filing reportedly does not specify a timeline.
  • The concept is framed partly as a sustainability argument (less water cooling), but the practical engineering burden remains substantial.
  • Musk has argued congestion concerns are overstated, but regulators and researchers will focus on “scale effects” if the proposal advances.

Bottom line

This is an early regulatory step, not a launch schedule—but it signals that SpaceX is thinking beyond “satellite internet” and positioning itself in the next infrastructure race: compute capacity for AI, wherever it can be placed.

Sources

n English