Nvidia announced computing platforms for orbital data centres during its GTC 2026 conference, with CEO Jensen Huang declaring that "space computing, the final frontier, has arrived." The move positions the chipmaker at the centre of an emerging race to extend AI infrastructure beyond Earth as terrestrial data centres face mounting power and cooling demands.
The Vera Rubin Space-1 Module delivers up to 25 times more AI compute for space-based inferencing compared with the H100 GPU. The module includes the IGX Thor and Jetson Orin platforms, engineered for size-, weight- and power-constrained environments. The system is designed for processing data gathered from orbital sensors and spacecraft rather than for processing data for ground-based systems.
Six commercial space companies including Aetherflux, Axiom Space, Kepler Communications, Planet Labs, Sophia Space and Starcloud are using Nvidia accelerated computing platforms to power next-generation space missions. Aetherflux claims it intends to launch its first datacenter satellite into orbit during the first quarter of 2027.
The orbital push forms part of a broader strategic repositioning at Nvidia. Huang said purchase orders between Blackwell and Vera Rubin chips are expected to reach $1 trillion through 2027, up from the prior $500 billion revenue opportunity estimate. During the keynote, Huang stated "computing demand has increased by 1 million times in the last two years."
Nvidia's next major architecture is Feynman, which will include a new CPU called Rosa, named for Rosalind Franklin whose X-ray crystallography revealed the structure of DNA. The Feynman family includes the LP40 LPU built jointly with Groq, the Rosa CPU, BlueField 5, and Kyber-CPO scale-up using co-packaged optics. Feynman targets 2028.
The space venture reflects growing pressure to relocate computing infrastructure. Soaring electricity costs from the AI data centre buildout have pushed companies to explore alternatives beyond terrestrial infrastructure. Google announced Project Suncatcher in November to explore computing in space, while SpaceX acquired xAI in a merger with implications for orbital data centre deployment.
Yet significant obstacles remain unresolved. Huang acknowledged that in space there is no conduction and no convection, only radiation, requiring engineers to develop new cooling approaches. Gartner analyst Bill Ray said in February that companies are wasting money on orbital data centre development because "the economics do not work" due to prohibitive launch costs and immense cooling challenges in the vacuum of space.
The Vera Rubin Space-1 Module will ship at a later date, while the IGX Thor and Jetson Orin platforms are available today. Nvidia provided no release date for the Space Module. For Australian technology observers, the announcement underscores how thoroughly the global AI infrastructure race has shifted beyond Earth, with questions about viability and sustainability still outstanding.