Nvidia introduced a specialized AI chip designed for space-based data centers.

Nvidia is advancing its presence in the emerging space-based computing market with the announcement of the Vera Rubin Space Module at its GTC event. The module is designed to run AI models from orbit and contains a GPU using Nvidia's latest "Rubin" architecture. The company promises an up to 25-times performance leap compared to the H100 GPU, which arrived in 2022. However, Nvidia has not announced a specific launch date for the chip.

The Vera Rubin Space Module appears engineered for specialized workloads on satellites or space stations rather than general-purpose cloud computing from orbit. Nvidia noted that the module can process data streams from "space-based instruments in real time," creating a path to "on-orbit analytics, autonomous scientific discovery, and rapid insight generation." The technology addresses a critical challenge: cooling AI chips in orbit where there is no air. Nvidia's CEO Jensen Huang has taken a cautious stance on the broader opportunity, stating "the economics are poor today, but it is going to improve over time" on an earnings call last month, though he acknowledged that GPUs in space could excel at certain tasks such as high-resolution satellite imaging.

This approach contrasts with SpaceX, which has been more bullish on orbital data centers. SpaceX's CEO Elon Musk has expressed confidence that space-based data centers will eventually surpass terrestrial data centers in both costs and efficiency in the near future. SpaceX has filed a regulatory request to operate up to 1 million satellites to support its orbiting data center project. Despite SpaceX's prominence in the field, Nvidia did not mention the company in connection with the Vera Rubin Space Module announcement.

Nvidia is developing the Vera Rubin Space Module in partnership with multiple organizations including Aetherflux, a space-based solar-power developer; Axiom Space, a space-station maker; Planet Labs, a satellite-imagery provider; and StarCloud, a startup developing orbiting data centers. In November of last year, StarCloud launched an Nvidia enterprise GPU, the H100, into space aboard a test satellite, and successfully connected, trained, and ran AI models using the GPU. StarCloud has since filed a request to launch up to 88,000 satellites. Additionally, Nvidia recently began recruiting for an "Orbital Datacenter System Architect" position, signaling continued expansion into this domain.