NVIDIA announced SpaceXAI will deploy its new Vera CPU, built specifically for AI agents, to power Grok’s infrastructure on Earth and the first-generation Starmind AI satellite in orbit.
NVIDIA announced that SpaceXAI will deploy NVIDIA’s new Vera CPU to accelerate its next generation of agentic AI applications, bringing what NVIDIA calls the first CPU built specifically for AI agents to one of the world’s most ambitious AI deployments. The announcement, made at Hot Chips 2026 in Palo Alto, also detailed plans to extend the same underlying architecture from Grok’s terrestrial AI infrastructure all the way into orbit.
Why Agentic AI Needs a Different Kind of CPU
NVIDIA’s pitch centers on a specific technical bottleneck: agentic AI applications increasingly depend on CPUs to orchestrate tools, execute code, process data, and run simulations between model calls, rather than just feeding GPUs. SpaceXAI will use Vera to accelerate these application workloads, aiming to help AI agents act faster while keeping GPUs fully utilized. “Agentic AI requires a new kind of computing system — one built not only to generate answers, but to take action,” said Ian Buck, vice president of hyperscale and high-performance computing at NVIDIA. “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.”
88 Cores, 1.2TB/s of Bandwidth: Inside the Vera CPU
NVIDIA Vera is built around 88 NVIDIA-designed Olympus cores, NVIDIA Spatial Multithreading technology, and high-bandwidth LPDDR5X memory delivering up to 1.2TB/s of bandwidth. NVIDIA says Vera enables up to 1.8x faster task completion compared with x86 CPUs across workloads including agentic AI, reinforcement learning, and data processing. “Vera gives us the CPU performance and memory bandwidth to run enormous amounts of orchestration, code and data processing while keeping GPUs doing what they do best,” said Mike Nicolls, president of SpaceXAI. “That means higher-performance AI agents and more useful work from every watt of compute.”
Grok’s AI Infrastructure Is Scaling Toward Gigawatts of Compute
Beyond the standalone Vera CPU, SpaceXAI plans to expand the AI infrastructure behind Grok using the broader NVIDIA Vera Rubin platform as it scales toward gigawatts of computing capacity. Vera Rubin is codesigned across compute, networking, and software, combining NVIDIA accelerated computing, NVLink interconnect technology, Spectrum-X Ethernet networking, BlueField data processing, and NVIDIA software into one integrated architecture intended to maximize performance and power efficiency while driving down cost per token.
Starmind: The First AI Satellite Built on NVIDIA’s Rack-Scale System
The most striking element of the announcement is where this architecture is headed next. SpaceXAI’s planned first-generation Starmind AI satellite will be based on an optimized NVIDIA Vera Rubin NVL72 rack-scale system — the same class of hardware powering next-generation AI factories on Earth, adapted for orbit. NVIDIA notes that orbital computing brings dramatically different constraints than conventional data centers, including power availability, thermal management, bandwidth, reliability, and physical integration, and says it is working with SpaceXAI to adapt the Vera Rubin foundation to those requirements while preserving a common NVIDIA architecture and software ecosystem. According to independent reporting on the announcement, the Starmind satellite is targeted to be operational in orbit starting in the fourth quarter of 2027.
One Architecture, Three Environments
NVIDIA frames the broader strategy as a single computing foundation spanning very different environments: Vera CPUs accelerating increasingly sophisticated AI agents, Vera Rubin powering Grok’s AI infrastructure and gigawatt-scale AI factories on Earth, and the same NVIDIA accelerated-computing architecture extending into orbital AI infrastructure through Starmind.