Starcloud Secures $250M at $2.3B Valuation to Move AI Workloads into Orbit

Artificial intelligence scaling is hitting a hard terrestrial ceiling. Between multi-year waits for power grid hookups and cooling water constraints, data center operators face growing physical limits on Earth, prompting hardware giants to evaluate orbital alternatives.

Starcloud closed a $250 million funding extension that values the company at $2.3 billion, backed by semiconductor designer Nvidia and networking hardware vendor Cisco. The startup previously tested its compute design in November 2025 by deploying an Nvidia H100 GPU into orbit, successfully executing inference on a version of Google's Gemini model.

To operate at scale, Starcloud submitted an application to the Federal Communications Commission for an 88,000-satellite constellation designed to provide 20 gigawatts of compute capacity. The architecture relies on continuous solar exposure and passive radiative cooling in the vacuum of space, eliminating the local utility interconnections that delay traditional ground-based facilities.

While operating hardware off-planet sidesteps local energy permits, the strategy trades power line bottlenecks for launch manifest constraints and orbital debris management. Whether satellite clusters can compete on latency and cost per token against nuclear-backed terrestrial campuses will determine if space data centers become an operational tier or remain a niche experiment.