NVIDIA CEO Jensen Huang announced at GTC 2026 that “space computing is here,” officially launching the space computing platform. The core product is the Vera Rubin Space-1 system, designed specifically for satellite and orbital data centers, equipped with IGX Thor (GPU) and Jetson Orin (CPU).
(Background: Jensen Huang’s full GTC 2026 speech: AI demand reaches trillions of dollars, computing power jumps 350 times, OpenClaw turns every company into AaaS)
(Additional context: Computing power is king! Jensen Huang discusses how AI will reshape the global value chain and when robots will become widespread)
In his keynote at GTC 2026, Jensen Huang officially announced NVIDIA’s entry into the space computing field, with the core message: “Intelligence must live wherever data is generated.”
Satellites capture images of Earth and signals daily, transmitting raw data back to ground stations for processing. This process faces high latency and bandwidth bottlenecks. NVIDIA’s solution is to bring computing power directly into orbit.
Vera Rubin Space-1 consists of two chips: the IGX Thor (GPU) responsible for AI inference, and the Jetson Orin (CPU) for general processing. The entire system is redesigned for the “small size, light weight, low power” space environment, making it deployable on satellites or future orbital space stations.
Partner Axiom Space is building private space stations, while Planet operates a large constellation of Earth observation satellites. Both are direct applications of edge computing in space.
Jensen Huang openly admitted that the main engineering challenge for space computing centers is cooling. Earth-based data centers rely on air convection and liquid cooling, but in space, there is no air. Heat can only be dissipated through solid conduction or thermal radiation. This means Vera Rubin Space-1 must be equipped with “quite large heat sinks,” and the entire cooling architecture needs to be redesigned.
This is why NVIDIA announced a complete “space computing system platform,” not just a single chip.