NVIDIA says SpaceXAI will deploy its Vera CPUs to power the next generation of agentic AI workloads behind Grok, linking the chipmaker’s new CPU architecture to SpaceXAI’s push toward gigawatt-scale computing. The plan also reaches beyond terrestrial data centers: SpaceXAI intends to carry an optimized Vera Rubin NVL72 foundation into orbit through its first-generation Starmind AI satellite.
Vera is designed for the CPU-heavy work that surrounds model inference, including tool orchestration, code execution, data processing and simulation. Those tasks take place between model calls, so faster CPU processing can help agents act sooner while keeping attached GPUs occupied. NVIDIA says the processor uses 88 company-designed Olympus cores, Spatial Multithreading and high-bandwidth LPDDR5X memory capable of up to 1.2TB/s.
NVIDIA claims Vera can complete tasks up to 1.8 times faster than x86 CPUs across agentic AI, reinforcement learning and data-processing workloads. The company presents it as the first CPU purpose-built for AI agents, though the claim comes from NVIDIA. SpaceXAI president Mike Nicolls said the performance and memory bandwidth are intended to handle large volumes of orchestration, code, and data work while leaving GPUs focused on their primary jobs.
At the system level, SpaceXAI plans to expand the infrastructure supporting Grok around the Vera Rubin platform. The architecture combines accelerated computing, NVLink, Spectrum-X Ethernet, BlueField data processing and NVIDIA software across training, reasoning and inference. NVIDIA’s pitch is that a common stack can serve both large AI factories on Earth and systems adapted for orbital conditions, where power, heat, bandwidth, reliability and physical integration create different limits.
The proposed Starmind satellite would use an optimized Vera Rubin NVL72 rack-scale system, with NVIDIA and SpaceXAI adapting it for orbital computing while retaining the same software ecosystem. That would give SpaceXAI one computing foundation spanning agent workloads, Grok infrastructure, and planned space-based AI. NVIDIA did not announce a deployment date, and it cautioned that the products and features remain in development, will be offered only if and when available, and could change before release.