You may have heard of data centers on the ground, but now Nvidia chips are bringing them to space.

At Nvidia GTC on Monday, the company unveiled its first space computing solutions, which, according to Nvidia, will usher in a new era of space innovation by bringing AI compute to orbital data centers (ODCs), geospatial intelligence, and autonomous space operations.

Nvidia joins Google and SpaceX, which have unveiled plans for putting datacenters and AI factories in space. In fact, SpaceX became the largest private company in the world by acquiring xAI and using space-based data centers as a major synergy of the deal. However, Nvidia is doing something different. It's enabling space-based devices, such as satellites and spacecraft, to process AI workloads in real time without sending the data back to earth-based AI factories.

“AI processing across space and ground systems enables real-time sensing, decision-making and autonomy, transforming orbital data centers into instruments of discovery and spacecraft into self-navigating systems,” said Nvidia CEO Jensen Huang.

While Nvidia’s AI-powered space infrastructure includes existing technologies such as the IGX Thor & Jetson Orin, the Nvidia Vera Rubin Space Module is the latest in Nvidia’s accelerated platforms for space. Highlights include:

• Up to 25x more AI compute for space-based inference than the Nvidia H100 GPU
• Large language models and advanced foundation models operate directly in space
• Tightly integrated CPU-GPU architecture and high-bandwidth interconnect for processing of a massive data stream from space in real-time

Space is a challenging environment for electronics due to constraints on size, weight, and power (SWaP), requiring systems to be small, lightweight, and to consume minimal power. Nvidia’s infrastructure was designed to meet those specific needs. Space companies, including Aetherflux, Axiom Space, Kepler Communications, Planet, Sophia Space, and Starcloud, are already using Nvidia's platforms for their missions.

Unlike traditional satellites that beam raw data back to Earth for processing, space computing enables real-time on-orbit analysis. This is critical for disaster monitoring, environmental hazard tracking, and autonomous operations that don't require waiting on human input from mission control.

Our Deeper View

In the post, Huang says, "intelligence must live wherever data is generated," which brings up a really interesting point. The benefits of processing data with AI on the edge are clear, as there has been a strong emphasis on running AI models on the edge for personal and organizational use cases since AI became mainstream. Local processing unlocks new levels of speed, reliability, security, and more. Now, if you apply that logic to everything, the number of workflows and infrastructure that will need to change is vast, with the most visceral example being something as complex and far-reaching as space.