SpaceXAI is going all in on NVIDIA’s next-generation AI architecture....from Grok on Earth all the way to AI compute in space SpaceXAI will deploy NVIDIA Vera CPUs for its next generation of agentic AI workloads while expanding the infrastructure behind Grok with Vera Rubin And Vera is built specifically for what AI agents actually need When an agent is working, the GPU is only part of the story Between model calls, the system has to: • Execute code • Use tools • Process huge amounts of data • Run simulations • Orchestrate multiple tasks • Keep feeding work back to the GPUs That increasingly becomes a CPU problem Vera has 88 NVIDIA-designed Olympus cores and up to 1.2 TB/s of memory bandwidth, delivering up to 1.8× faster task completion than x86 CPUs across workloads including agentic AI, reinforcement learning and data processing SpaceXAI plans to use that performance to make AI agents faster while keeping its massive GPU infrastructure better utilized And this gets much bigger SpaceXAI is scaling toward gigawatts of AI computing capacity using NVIDIA Vera Rubin And it doesn’t stop at terrestrial data centers SpaceXAI’s first-generation Starmind AI satellite is planned around an optimized Vera Rubin NVL72 system Basically, the same NVIDIA computing architecture being built for giant AI factories on Earth is now being adapted for orbital AI compute The direction is becoming pretty clear: Vera CPUs → faster, more capable AI agents Vera Rubin → gigawatt-scale AI factories Starmind → AI computing infrastructure in orbit SpaceXAI is building an AI architecture that starts on Earth and eventually extends into space That is a completely different scale of ambition