SpaceX is collaborating with Nvidia to send the Vera Rubin NVL72, a specialized AI supercomputer system, into orbit in the fourth quarter of 2027, marking a significant advance in space-optimized artificial intelligence technology.
- Vera Rubin NVL72 combines 72 Rubin GPUs and 36 Vera CPUs tailored for space AI workloads.
- First AI-specific CPU by Nvidia with 88 Olympus cores launched into orbit with SpaceX satellite.
- System launch set for late 2027; large-scale expansion planned for 2028.
What happened
SpaceX and Nvidia have jointly developed the Vera Rubin NVL72, a space-optimized AI rack-scale system that is planned for orbital deployment in Q4 2027. This system combines 72 Rubin GPUs and 36 Vera CPUs, along with advanced interconnect and data processing units, creating a powerful AI platform specifically designed for use in aerospace environments.
Elon Musk announced this collaboration, reinforcing SpaceX's ambitions to expand its AI infrastructure to support new projects like the StarMind AI satellite. This marks the first time Nvidia's Vera CPU, which is designed specifically to accelerate agent AI workloads, will be used in space.
Why it matters
The Vera Rubin NVL72 system brings a novel approach to AI in space by integrating a CPU architecture optimized for AI agent operations, which includes orchestration, code execution, and data processing. This allows AI agents to perform faster and more efficiently by keeping GPUs properly fed and fully utilized, a critical factor in high-demand AI tasks such as reinforcement learning and simulation.
By extending Earth-based AI factory computing architecture into orbit, SpaceXAI aims to leverage these advancements to enhance satellite capabilities and push forward space-based AI research and applications. This could lead to faster, more autonomous satellite operations and new AI-powered functionalities in space missions.
What to watch next
The launch of the Vera Rubin NVL72 system aboard SpaceX's first-generation StarMind AI satellite in late 2027 will be a key milestone. Observers will be watching to see how the system performs its AI workloads in orbit, including its ability to accelerate AI orchestration and agent tasks under space environment constraints.
Looking toward 2028, SpaceXAI plans a significant scale-up of this AI infrastructure, which could signal broader deployment in additional satellites or other space-based platforms. Industry watchers should also monitor how Nvidia’s AI-optimized CPU architecture influences AI agents’ development and application in orbital and possibly deep-space missions.