Starcloud-1
-
Nvidia-Backed Starcloud Trains Its First AI Model in Space Using Orbital Data Centers
Starcloud‑1, launched on 2 Nov 2025, carries an Nvidia H100 GPU and runs the open‑source LLM Gemma in orbit—the first high‑performance AI model operating beyond Earth. The demonstration proves that orbital data centers can deliver complex AI inference with up to ten‑fold lower energy costs than ground facilities, offering applications such as real‑time disaster detection and satellite telemetry processing. Starcloud plans a 5‑GW, solar‑powered orbital compute constellation and a 2026 launch featuring additional GPUs and Nvidia’s Blackwell chips. Risks include radiation, debris, maintenance challenges, and regulatory uncertainty.