According to a report on June 5th by Fast Technology, NVIDIA's CEO Huang Renxun recently put forward a compelling view in his speech: the new wave of AI will be physical AI, also known as "embodied AI" or Physical AI.

Physical AI refers to those physical systems that can perform tasks similar to intelligent organisms, capable of co-evolving the body's control, morphology, motor execution, and perception.

Huang Renxun pointed out that many current AI systems do not truly understand physical laws and cannot be based on the material world. However, generating images, videos, 3D graphics, and simulating physical phenomena all require AI that is physics-based and has a deep understanding of physical laws.

Leading global electronics companies have begun to respond to this trend by integrating NVIDIA's autonomous robot technology into their factories. Utilizing the simulation capabilities of NVIDIA's Omniverse platform, these companies are testing and validating a new generation of AI for the physical world, which currently covers over 5 million pre-programmed robots worldwide.

In this wave, many leading global robotics companies, including BYD Electronics, Siemens, Teradyne, and Intrinsic (a subsidiary of Alphabet), are incorporating technologies such as NVIDIA Isaac acceleration libraries into their software frameworks and robot models to improve efficiency in factories, warehouses, and distribution centers.

Huang Renxun said, "The era of robots has arrived, and in the future, all moving objects will operate autonomously." He emphasized that NVIDIA is accelerating the application of generative physical AI by promoting the development of the robotics stack.

Of particular note, Huang Renxun mentioned two high-volume robotic products: self-driving cars and humanoid robots.

In the field of self-driving cars, NVIDIA has provided customers with solution stacks and plans to cooperate with fleets such as Mercedes-Benz and Jaguar Land Rover in the near future.

As for humanoid robots, Huang Renxun said that the field has made significant breakthroughs in cognitive abilities and understanding of the world in recent years, and the development prospects are promising. He is particularly excited about the potential of humanoid robots, as they are most likely to adapt to the world built by humans.

However, he also pointed out that training humanoid robots requires vast amounts of data compared to other types of robots, but because they are similar in size to humans, demonstrations and videos provide them with a wealth of valuable training material.