Crusoe to become first cloud operator in space (www.crusoe.ai)

Crusoe announced a partnership with Starcloud to become the first public cloud operator running AI workloads in orbit, deploying a dedicated Crusoe Cloud module on a Starcloud satellite slated to launch in late 2026 and offering limited GPU capacity by early 2027. Starcloud’s orbital data center is a self-contained platform powered by dedicated solar arrays that removes dependence on terrestrial grids, cooling infrastructure and land, while Crusoe brings its “energy‑first” AI infrastructure expertise to co-locate compute with abundant solar power. The companies also plan to scale larger space-based data centers over time. Technically notable: Starcloud will be the first to fly an NVIDIA H100 GPU to space in November 2025 (which the announcement frames as ~100x more powerful GPU compute than has previously been in orbit), and the 2026 satellite will host Crusoe’s cloud stack to let customers deploy GPU workloads from space. The move signals a new paradigm for meeting AI’s enormous energy needs by tapping orbital solar, promising sustainability and potentially massive compute scale — while also raising practical constraints around bandwidth, latency, data egress, regulatory/launch risk and limited initial capacity. For AI/ML builders, this is a provocative step toward off‑earth AI factories that trade terrestrial constraints for new operational and architectural tradeoffs.