Nvidia teases next-generation Rubin platform, shares physical AI vision
To do so, physical AI has to understand the world model, so that they understand how to interpret the world, how to perceive the world, he said. They have to have excellent cognitive capabilities so they can understand us, understand what we asked, and perform the tasks in the future.
Huang went on to predict a robotic future. “All of the factories will be robotic. The factories will orchestrate robots, and those robots will be building products that are robotic. Robots interacting with robots, building products that are robotic. Well, in order for us to do that, we need to make some breakthroughs,” he said before showing off a demo video of researchers developing robots powered by physical AI.
Physical AIs use multimodal LLMs that enable robots to learn, perceive and understand the world around them and plan how they’ll act. One of the integral technologies for advancing robotics is reinforcement learning, gained from human feedback to learn particular skills.
In this vision, however, generative physical AI can learn skills using reinforcement learning from feedback in a simulated world rather than from humans. “These simulation environments are where robots learn to make decisions by performing actions in a virtual world that obeys the laws of physics. In these robot gyms, a robot can learn to perform complex and dynamic tasks safely and quickly, refining their skills through millions of acts of trial and error,” Huang said.
Annual upgrades for IA accelerators
The news of a new architecture isn’t exactly a surprise, as Nvidia recently announced it wants to go to a one-year cadence for its new architectures rather than two years. It’s extremely ambitious and leaves no room for error to update GPU technology with billions of transistors already. AMD has said it will go to a similar cadence with its Instinct line.
“Our company has a one-year rhythm. Our basic philosophy is very simple: build the entire data center scale, disaggregate and sell to you parts on a one-year rhythm, and push everything to technology limits,” Huang said.
