SpaceXAI will deploy Nvidia Vera CPUs to power its next generation of agentic AI workloads as the company expands the computing infrastructure behind Grok and prepares to take Nvidia systems into orbit.
The company plans to build its future AI infrastructure around Nvidia’s Vera Rubin platform as it scales toward gigawatts of computing capacity.
Vera CPUs will handle tasks including orchestration, code execution, data processing, and simulation while GPUs focus on model inference.
SpaceXAI also plans to use an optimized Vera Rubin NVL72 system for its first generation Starmind AI satellite, extending the same computing architecture used in terrestrial AI data centers into space.
Elon Musk said SpaceX and Nvidia have designed a space optimized Vera Rubin NVL72 system that is scheduled to launch into orbit in the fourth quarter of 2027, with significant deployment scale planned for 2028.
Musk said the orbital design is simpler, cheaper, denser, and lighter than a traditional rack.
The effort would extend Nvidia’s accelerated computing infrastructure beyond conventional data centers, where orbital deployments face tighter constraints around power, cooling, bandwidth, reliability, and physical size.
Nvidia’s Vera CPU is designed specifically for AI agent workloads. It includes 88 Olympus cores and LPDDR5X memory delivering up to 1.2 terabytes per second of bandwidth.
Nvidia says Vera can complete certain agentic AI, reinforcement learning, and data processing workloads up to 1.8 times faster than x86 CPUs.
SpaceXAI President Mike Nicolls said the additional CPU performance and memory bandwidth should allow the company to run more orchestration and data processing while keeping its GPUs focused on accelerated computing.
The partnership gives SpaceXAI a common Nvidia architecture spanning Grok infrastructure on Earth and its planned orbital AI systems as the company pushes toward larger scale computing deployments.