At the Consumer Electronics Show, Nvidia CEO Jensen Huang introduced a new concept that could define the next decade of technology: “Physical AI.” While chatbots live on screens, Physical AI lives in the real world, interacting with objects and people. Huang declared that the “ChatGPT moment” for this technology has arrived, with self-driving cars leading the charge.
The vehicle for this revolution is Alpamayo, Nvidia’s new reasoning software. Huang explained that for machines to act safely in the real world, they must understand it. Alpamayo combines visual sensors with language-like reasoning, allowing robots and cars to comprehend context, anticipate outcomes, and make nuanced decisions in complex environments.
This technology is expected to be a game-changer for robotaxis. Huang noted that these autonomous fleets will be among the first to benefit, as they require high levels of safety and adaptability to function in cities. The ability to reason through rare scenarios allows these vehicles to operate without the constant need for remote human assistance, making the business model of robotaxis finally viable.
Powering this vision is the new Vera Rubin chip platform. These chips are optimized to generate the “tokens” used by AI systems ten times more efficiently than before. This efficiency is critical for physical machines, which need to process vast amounts of data without draining batteries or requiring massive cooling systems.
Huang’s presentation made it clear that Nvidia views the car as just the beginning. The principles behind Alpamayo and the power of the Rubin chips are scalable to all forms of robotics. Nvidia is laying the groundwork for a future where intelligent machines are commonplace assistants in our physical lives, all powered by their technology.