NVIDIA Frames AI Infrastructure as Fungible as CoreWeave Brings Vera Rubin NVL72 Into Production

NVIDIA is sharpening its argument that AI infrastructure should be treated as a flexible, long-lived platform rather than a single-purpose chip deployment, pointing to CoreWeave's first production use of NVIDIA Vera Rubin NVL72 systems and continued demand for older NVIDIA GPU generations. The company says the model depends on one architecture being able to serve training, reinforcement learning, inference, agent execution and even non-AI workloads as customer demand shifts.