SpaceX and Nvidia Collaborate to Launch Space-Optimized AI System in Orbit

SpaceX and Nvidia are collaborating to develop a space-optimized Vera Rubin NVL72 AI system, which is planned for an orbital launch in the fourth quarter of next year. Elon Musk announced the development by resharing an Nvidia post, noting that the deployment will see significant scaling in 2028 as SpaceXAI expands its AI infrastructure.

The Nvidia Vera Rubin NVL72 is a rack-scale AI supercomputer system. It integrates 72 Rubin GPUs, 36 Vera CPUs, ConnectX-9 SuperNICs, and BlueField-4 DPUs into a unified system architecture. According to Nvidia, the system's deployment is intended to accelerate orchestration, code execution, and data processing for next-generation agentic AI applications.

Nvidia's Vera CPU is built specifically for AI agents to accelerate CPU-intensive tasks during model inference, including tool use, code execution, data processing, orchestration, and simulation. The processor features 88 Nvidia-designed Olympus cores, Spatial Multithreading technology, and high-bandwidth LPDDR5X memory that offers up to 1.2 terabytes per second of memory bandwidth. Nvidia reports that Vera delivers up to 1.8 times faster task completion than x86 CPUs across workloads like agentic AI, reinforcement learning, and data processing.

As part of the initiative, SpaceXAI's planned first-generation Starmind AI satellite will utilize the optimized Vera Rubin NVL72 rack-scale system. This deployment brings the accelerated computing architecture used in terrestrial AI factories directly into space. Mike Nicolls, president of SpaceXAI, stated that the hardware provides the CPU performance and memory bandwidth required to manage large volumes of orchestration, code, and data processing while maintaining high GPU utilization.

The announcement follows a period of active operations for SpaceX, which recently launched two Falcon 9 rockets from opposite coasts of the United States within a short timeframe, marking a company record for the minimum gap between orbital missions.

"The collaboration between SpaceX and Nvidia to deploy advanced rack-scale AI systems in orbit highlights a significant technical shift in how edge computing and space infrastructure intersect. By bringing high-bandwidth memory and specialized CPU architectures directly into satellite hardware, companies are addressing critical bottlenecks in orbital data processing and agentic AI workloads. This development demonstrates how terrestrial computing standards are increasingly extending into aerospace applications to handle complex telemetry and automation tasks at scale." — Dr. Shishir Gupta, Founder & CEO, StartupLanes