NVIDIA and SpaceXAI Take Agentic AI From Earth to Orbit in Massive Grok Expansion
NVIDIA and SpaceXAI are pushing artificial intelligence toward a new frontier — and this time, the destination is space.
NVIDIA has announced that SpaceXAI will deploy its next-generation NVIDIA Vera CPUs to power increasingly sophisticated agentic AI workloads, while expanding the infrastructure behind Grok using the powerful NVIDIA Vera Rubin platform.
But the biggest twist is what comes next.
SpaceXAI is also planning to take NVIDIA-powered AI computing beyond Earth, with its first-generation Starmind AI satellite expected to use an optimized NVIDIA Vera Rubin NVL72 system.
The announcement signals an ambitious vision: AI infrastructure that can operate across massive data centers on Earth and eventually extend into orbit.
NVIDIA Vera Is Built for the Age of AI Agents
Traditional AI systems primarily generate answers. Agentic AI takes things a step further.
AI agents can execute code, process information, use digital tools, coordinate tasks and perform complex operations between model interactions. Those workloads place enormous demands on CPUs.
That is where NVIDIA Vera comes in.
The company says Vera is its first CPU specifically designed for AI agents. It features 88 NVIDIA-designed Olympus cores, Spatial Multithreading technology and up to 1.2TB/s of LPDDR5X memory bandwidth.
According to NVIDIA, Vera can deliver up to 1.8 times faster task completion than x86 CPUs across workloads such as agentic AI, reinforcement learning and large-scale data processing.
For SpaceXAI, the goal is simple: let AI agents work faster while allowing GPUs to remain focused on the massive AI workloads they handle best.
“Agentic AI requires a new kind of computing system — one built not only to generate answers, but to take action,” said Ian Buck, NVIDIA’s vice president of hyperscale and high-performance computing.
Grok’s AI Infrastructure Is Getting a Major Upgrade
SpaceXAI is also expanding the infrastructure supporting Grok with NVIDIA Vera Rubin, a platform designed for massive-scale AI training, reasoning and inference.
The architecture combines accelerated computing with NVIDIA technologies including NVLink, Spectrum-X Ethernet, BlueField data processing and NVIDIA’s AI software ecosystem.
As SpaceXAI moves toward gigawatts of computing capacity, the company is betting that a common NVIDIA architecture can help scale increasingly demanding AI workloads while improving efficiency.
Mike Nicolls, president of SpaceXAI, said Vera will provide the CPU performance and memory bandwidth needed to handle massive amounts of orchestration, code execution and data processing.
The result could be more capable AI agents — and more computing work delivered from every watt of power.
The Next AI Data Center Could Be in Space
Perhaps the most eye-catching part of the announcement is SpaceXAI’s plan to develop AI computing infrastructure for orbit.
Space-based computing comes with challenges that traditional data centers do not face, including extreme thermal management requirements, power constraints, bandwidth limitations, reliability concerns and the difficulty of physically integrating powerful computing hardware into satellites.
SpaceXAI’s proposed first-generation Starmind AI satellite aims to tackle those challenges using an optimized NVIDIA Vera Rubin NVL72 system.
If successful, it would extend NVIDIA’s accelerated computing architecture from terrestrial AI factories into orbital infrastructure.
That means the same broader computing ecosystem could eventually support:
- AI agents powered by NVIDIA Vera CPUs
- Grok infrastructure running on Vera Rubin
- Gigawatt-scale AI factories on Earth
- Advanced AI computing systems operating in orbit
AI Is Officially Reaching for the Stars
The NVIDIA-SpaceXAI partnership represents a dramatic expansion of the AI computing race.
The industry is already building larger data centers, deploying increasingly powerful chips and consuming unprecedented amounts of electricity to train and operate next-generation AI systems.
Now, SpaceXAI is looking beyond Earth.
The vision is enormous: a unified AI computing architecture capable of powering intelligent agents, massive AI factories and potentially entire orbital computing networks.
For NVIDIA, Vera Rubin may become more than the foundation for the next generation of data centers.
It could become part of the infrastructure powering AI from Earth to orbit.
And if SpaceXAI succeeds, the next major AI expansion may not happen in Silicon Valley, Texas or another massive data-center hub.
It could happen hundreds of miles above the planet.