Artificial intelligence agents may be all the rage now, but NVIDIA (NVDA) CEO Jensen Huang has gone beyond the hype. He said on NVIDIA’s fourth-quarter earnings call yesterday (February 26) that the next phase of AI will be “physical AI.” Physical AI is a form of AI systems that can directly interact with the real world and cover products such as autonomous cars and robots.
“Now is the beginning of the proxy AI era, and you hear a lot of people talking about it. Then, after that there is physical AI,” Huang told analysts. He went on to explain: “They can’t just understand the meaning of words and language, but they must understand the meaning of the world, friction and inertia, the permanence and causes and effects of objects.”
To this end, NVIDIA revealed a “world model” platform in January that attempts to simulate the real world environment. This new platform, known as the Universe, promises to advance training in physical AI systems by providing a highly realistic virtual world. For example, self-driving cars can be tested under various simulated weather conditions, while robotics companies can check the safety of their products under different circumstances.
The universe has been adopted by 1X and NEURA robotics technology, as well as robotics companies such as self-driving car manufacturers such as Waabi and Wayve. NVIDIA Chief Financial Officer Colette Kress said yesterday that the platform is also exploring Robotaxi products with Uber (Uber). She added: “Just like the language basic model has revolutionized language AI, the universe is also a physical AI that revolutionizes robotics technology.”
The “World Model” is a concept of AI last year. Fei-Fei Li, a famous computer scientist and leader in AI research, founded the World Lab, which hopes to inject AI into “spatial intelligence” to understand and interact with the real world. The startup raised $230 million last year and unveiled its progress in generating virtual 3D scenes in 2D images.
Currently, AI developers have been focusing on agents, leaders such as OpenAI and Microsoft (MSFT) highlighting the characteristics of autonomous, action in recent AI releases. NVIDIA performed well in the Agesic AI era. The company’s quarterly revenue rose 78% year-on-year to $39.3 billion, with the vast majority coming from its data center business. Profits rose 80% to $22 billion in the quarter.
NVIDIA’s physical AI dream remains a trivial part of its business, with its automotive and robotics divisions accounting for less than 2% of the company’s total revenue in the January quarter of November 1. This share will rise in the future. “One day, there will be 1 billion cars on the road, and each of them will be robotic cars,” Huang told analysts.
Despite exceeding Wall Street’s expectations for revenue and net income, some investors are frightened by NVIDIA’s huge gross margin outlook. Nvidia’s stock fell more than 3% today.
