NVIDIA says AI factory returns hinge on productivity, durability and interchangeability

NVIDIA said in an official blog post that maximizing returns from AI factories depends on three factors: productivity, durability, and interchangeability. The company defined productivity as achieving the highest throughput per megawatt and the lowest cost per token, with the goal of lifting annual revenue capacity. Durability refers to NVIDIA GPUs and systems continuing to generate revenue years after shipment. Interchangeability, in NVIDIA’s framing, means the same infrastructure can run a wide range of AI and non-AI workloads, broadening demand. The post cited SemiAnalysis AgentX data showing that NVIDIA’s Vera Rubin NVL72 system delivers more than 30 times higher throughput per megawatt than the GB300 NVL72. It also said cost per million tokens on the DeepSeek V4 Pro model can be reduced by as much as 45 times. NVIDIA added that its A100 GPUs, first shipped in 2020, are still in commercial service, and that CoreWeave has extended reservations for some of those units through 2029.