SpaceX, Nvidia Develop AI System for Space
Elon Musk stated that SpaceX and Nvidia have created a space-optimized Vera Rubin NVL72 AI system intended for launch into orbit in the fourth quarter of 2027.
Nvidia's Vera Rubin NVL72 represents a sophisticated rack-scale AI supercomputer setup. It brings together cutting-edge technologies, featuring 72 Rubin GPUs, 36 Vera CPUs, ConnectX-9 SuperNICs, and BlueField-4 DPUs in a cohesive setup.
By resharing an Nvidia post, Musk indicated that the system will experience considerable growth in 2028 as SpaceXAI boosts its AI infrastructure.
The initial CPU designed for agents will operate at scale. @SpaceX is implementing NVIDIA Vera to enhance the orchestration, code execution, and data processing that drives its next generation of agentic AI -- ensuring GPUs remain utilized and agents respond quickly.
Nvidia's Vera is its inaugural CPU specifically created for AI agents, intended to enhance CPU-heavy tasks related to model inference, such as tool utilization, code execution, data handling, orchestration, and simulation.
The CPU includes 88 Nvidia-designed Olympus cores, Spatial Multithreading technology, and high-bandwidth LPDDR5X memory, providing as much as 1.2TB per second of memory bandwidth. Nvidia stated that Vera can achieve up to 1.8 times quicker task completion than x86 CPUs in workloads including agentic AI, reinforcement learning, and data processing.
Also Read: US Teen Drops Social Media Lawsuit Against Meta, Google, Snap
"Nvidia states that SpaceXAI's upcoming first-generation Starmind AI satellite will utilize the enhanced NVIDIA Vera Rubin NVL72 rack-scale system, applying the same advanced computing framework that drives future AI facilities on Earth into outer space."
Also Read: New Zealand Plans Bill to Ban Social Media for Children Under 16
With Agentic AI applications depending more on CPUs for tool orchestration, code execution, data processing, and simulations during model calls, SpaceXAI will leverage Vera to enhance these application workloads, enabling AI agents to operate more swiftly while ensuring GPUs remain fully utilized and supplied.
Also Read: Sonova Unveils New AI-Powered Hearing Aid
"Vera provides the CPU efficiency and memory bandwidth required to handle vast quantities of orchestration, code, and data processing while allowing GPUs to excel at their primary functions," states Mike Nicolls, president of SpaceXAI.