NVIDIA Takes AI Computing to Space with Vera Rubin Space-1 Module

NVIDIA Launches Space Computing at GTC 2026

NVIDIA has unveiled a suite of hardware designed to bring data-centre-class AI performance to orbit, headlined by the Space-1 Vera Rubin Module. Announced during CEO Jensen Huang's keynote at the GTC 2026 conference in San Jose, the module delivers up to twenty five times the AI compute of the H100 GPU for space-based inferencing workloads.

Built for the Demands of Orbit

The Space-1 Vera Rubin Module pairs a Rubin GPU with a tightly integrated CPU architecture and high-bandwidth interconnect, all engineered for the size, weight, and power constraints of spacecraft. The platform enables large language models and foundation models to run directly in space, processing massive data streams from space-based instruments in real time.

A Growing Ecosystem of Partners

Six companies are already working with NVIDIA's space computing platforms, including Aetherflux, Axiom Space, Kepler Communications, Planet Labs, Sophia Space, and Starcloud. Starcloud, which successfully launched a test satellite carrying an NVIDIA H100 GPU in November 2025 and became the first company to train a large language model in orbit, is now building purpose-designed orbital data centres.

Available Now and Coming Soon

NVIDIA's IGX Thor and Jetson Orin platforms are available immediately for edge AI inferencing aboard satellites. On the ground, the RTX PRO 6000 Blackwell Server Edition GPU delivers up to one hundred times faster geospatial intelligence processing compared to legacy CPU-based systems. The Space-1 Vera Rubin Module itself will be available at a later date.

Challenges Ahead

A key technical challenge remains cooling AI chips in the vacuum of space, which Huang acknowledged during his keynote. The announcement arrives as the commercial space industry faces growing demand for real-time data processing in orbit and companies explore large-scale satellite-based computing infrastructure.