Starcloud’s $250M Raise Pushes Orbital AI Data Centres Closer

Starcloud has raised another $250 million for orbital AI data centres, and the new funding makes one thing clearer: space-based compute is no longer only a science-fiction pitch.

TechCrunch reported that the new money extends Starcloud’s earlier $170 million Series A and values the company at $2.3 billion. The startup is building satellites designed to run AI inference in orbit, with its larger Starcloud-3 spacecraft intended for a future SpaceX Starship launch.

The appeal is simple. AI data centres on Earth need land, power, cooling, grid connections and political approval. In orbit, Starcloud is betting that solar power, cold space environments and a different physical footprint could eventually make some AI workloads cheaper or easier to run.

That does not mean orbital data centres are easy. Launch capacity is now part of the business model. Starcloud’s chief executive has said the company needs to secure a large amount of launch capacity, and that becomes more complicated if Falcon 9 availability tightens while Starship is still ramping toward routine commercial operations.

The startup has also requested permission from the U.S. Federal Communications Commission to operate a very large constellation. That alone shows the ambition, but it also hints at the regulatory and orbital-management issues ahead. Space-based data centres will need spectrum, debris planning, launch access, cooling design, radiation protection and clear rules around where data is processed.

Starcloud’s hardware story is also getting more concrete. GeekWire notes that Nvidia’s next-generation Space-1 Vera Rubin module is expected to be used on future Starcloud satellites. That gives the project a stronger AI infrastructure angle than a generic satellite story.

This connects directly to the wider debate around orbital compute. We previously looked at why orbital data centres are becoming part of the AI compute conversation, including the Musk and Altman arguments around space infrastructure, energy and the long-term cost of AI.

The best case for Starcloud is that some AI inference can be moved off Earth, especially workloads that are not latency-sensitive and can tolerate satellite connectivity constraints. The weaker case is that launch cost, maintenance risk and network delay make orbital compute useful only for narrow workloads for a long time.

There is also a geopolitical angle. If data centres move into orbit, cloud infrastructure becomes partly space infrastructure. That raises questions around jurisdiction, security, export controls, satellite vulnerability and whether governments will treat orbital AI compute like critical infrastructure.

The $250 million raise does not prove orbital AI data centres will work at scale. It does prove that investors are willing to fund the attempt. The next test is not the pitch deck. It is whether Starcloud can launch, operate and sell useful compute from orbit at a price that makes sense.