Podcast Notes | NVIDIA CEO Jensen Huang: Young People Should Learn to Ask AI Questions; NVIDIA Hopes to Leave Behind More Than Just Gaming and Chips

NVIDIA CEO Jensen Huang said on the Huge If True podcast that young people should learn to work with AI, framing prompt-craft as the core skill for the next generation, and argued the company's legacy should extend beyond gaming and chips to broad scientific impact. He traced NVIDIA's path from game GPUs to CUDA—a bet he said drew more than a decade of Wall Street skepticism—and to the 2012 AlexNet breakthrough, when two GeForce GTX 580 cards beat traditional image-classification methods and prompted NVIDIA to rebuild its full computing stack around AI. Huang described physical AI as requiring both world models and physics simulation, pointing to Cosmos as a foundation model for the physical world and Omniverse as a Newtonian simulation engine that generates synthetic training data so robots can rehearse millions of edge cases before deployment. He predicted that within a decade all moving objects—autonomous cars, buildings, robot mowers and humanoids—will have robotic capabilities, and that every person will have a lifelong digital companion that evolves across glasses, phones, PCs, cars and home robots. Rejecting the case for hard-wiring Transformer architectures into dedicated ASICs, he said fixing a computer to one algorithm turns it into a microwave, and NVIDIA will keep building programmable accelerated computing; he also noted that in RTX 50-series 4K rendering only about 500,000 of roughly 8 million pixels are ray-traced while AI predicts the rest.