Details
- NVIDIA launched Space-1 Vera Rubin Module, IGX Thor, and Jetson Orin platforms at GTC for SWaP-constrained space environments, delivering up to 25x more AI compute than H100 GPU for orbital data centers, geospatial intelligence, and autonomous operations.
- Partners include Aetherflux, Axiom Space, Kepler Communications, Planet Labs PBC, Sophia Space, and Starcloud, integrating NVIDIA tech for real-time satellite processing and ground analysis.
- Space-1 Vera Rubin provides data-center-class AI with Rubin GPU supporting LLMs in orbit; IGX Thor and Jetson Orin enable edge AI inference; RTX PRO 6000 Blackwell GPU accelerates ground processing up to 100x vs. CPUs.
- Builds on NVIDIA's edge-to-cloud acceleration, contrasting legacy CPU systems with hyperscale AI for space-to-space and space-to-ground workflows.
- Rubin platform specs verified: 288GB HBM4 per GPU at 22TB/s bandwidth, 50 petaFLOPS NVFP4 inference, Vera CPU with 88 Olympus cores and 1.5TB LPDDR5X, launching late 2026.
Impact
NVIDIA's space-grade platforms enable real-time orbital AI, reducing data downlink needs and powering autonomous missions amid booming satellite constellations. This positions NVIDIA ahead of competitors like AMD in high-bandwidth space compute, accelerating commercial orbital data centers. Expect broader adoption in geospatial intel and exploration, extending AI factories beyond Earth.
