The processor uses 88 Nvidia-designed Olympus cores alongside the company’s Spatial Multithreading technology and LPDDR5X memory. Nvidia says the CPU provides as much as 1.2TB/s of memory bandwidth and can complete tasks up to 1.8 times faster than x86 CPUs across agentic AI, reinforcement learning and data-processing workloads.
SpaceXAI’s plans extend beyond deploying Vera as a standalone CPU. The company intends to grow the AI infrastructure supporting Grok using Nvidia’s Vera Rubin architecture as it moves toward gigawatts of computing capacity.
Vera Rubin combines Nvidia compute hardware with NVLink interconnects, Spectrum-X Ethernet networking, BlueField data processing and Nvidia software. Rather than treating those components as separate pieces of infrastructure, the platform is designed around the performance and power requirements of large AI systems.
Nvidia sees CPU performance becoming increasingly important as AI applications move beyond generating responses and begin carrying out more complicated sequences of actions. “Agentic AI requires a new kind of computing system — one built not only to generate answers, but to take action,” said Ian Buck, Nvidia vice president of hyperscale and high-performance computing. “Vera gives AI agents the CPU performance to act in real time — executing code, processing data and coordinating complex tasks. SpaceXAI is taking this architecture from massive AI factories to the next frontier of computing in orbit.”
That orbital component is another part of the companies’ work together. SpaceXAI is developing its Starmind satellite around an optimized version of the Vera Rubin NVL72 rack-scale system. Nvidia and SpaceXAI are adapting the architecture for an environment with different constraints around power, cooling, bandwidth, reliability and physical integration.
The approach would give SpaceXAI a common Nvidia computing foundation across several parts of its AI infrastructure. Vera CPUs would handle CPU-intensive agent workloads, Vera Rubin would support the infrastructure behind Grok and the company’s larger AI computing deployments, and an adapted version of the same architecture would power its planned Starmind system in orbit.
For Nvidia, the SpaceXAI deployment puts Vera into a large-scale agentic AI environment where CPU orchestration and GPU computation are designed to work together. SpaceXAI, meanwhile, is adopting the architecture across both its terrestrial AI infrastructure and its planned expansion into orbital computing.
This analysis is based on reporting from NVIDIA.
Image courtesy of NVIDIA.
This article was generated with AI assistance and reviewed for accuracy and quality.
