NVIDIA focuses on space. Systems for operation in orbit are shown

NVIDIA shows more and more clearly that its ambitions go far beyond classic data centers on Earth. While GTC 2026 announced Vera Rubin Space Modulei.e. new a computing platform designed for AI work directly in orbit. It is expected to offer up to 25 times more computing power than the H100 accelerator. Moreover, 6 commercial companies are already using Green technologies in the space sector.

There will be a new module foundation for orbital data centerswhich will run large language models and advanced base models outside the atmosphere. NVIDIA is betting here Tightly integrated CPU and GPU architecture and fast, high-bandwidth connections. The whole thing was designed to process huge data streams coming from instruments placed in space in real time.

It is positioned lower in this cosmic puzzle NVIDIA IGX Thor. This system is intended for mission-critical edge environments where real-time processing, functional safety, secure boot and the ability to operate autonomously are important.

On the other hand Jetson Orin targets the smallest structures, so satellites severely limited in size, power consumption and cooling. This is where the system will be responsible for computer vision, navigation and analysis of sensor data without having to constantly send everything back to Earth.

Unfortunately, NVIDIA did not provide a specific release date for the Vera Rubin Space Module and limited itself to announcing that “the platform will hit the market at a later date”. The rest of the solutions discussed are already available. They are used by companies such as Aetherflux, Axiom Space, Kepler Communications, Planet Labs PBC, Sophia Space and Starcloud.