NVIDIA's Huang says AI beats paper, pencil

The provocative part is the phrasing, but the useful signal is cultural: NVIDIA is pushing AI from optional assistant to default engineering infrastructure.

• –In chip design, verification, and simulation, the value of AI is in accelerating workflows that are already too complex for purely manual iteration.
• –A reported $2B token budget suggests frontier teams will start budgeting AI like cloud spend: by throughput, reuse, and ROI.
• –That changes how engineering leaders justify AI internally: fewer demos, more metrics like cycle time, defect rate, and productivity per token.
• –NVIDIA also has a clear incentive to normalize this mindset because every new internal use case validates the ecosystem it sells.