Buoyed by a soaring stock price and a hot artificial intelligence boom, Nvidia on Friday announced several plans to further leverage AI, including an alliance to build better humanoid robots, new Blackwell computer chips and a new way to design factories. Facility using Apple’s Vision Pro spatial computing platform.
The announcement came during a two-hour keynote address by Nvidia co-founder and CEO Jensen Huang at the annual GPU Technology Conference (GTC) in San Jose, California, on Monday.
Starting with the company’s underlying technology, Huang noted that the need for AI development will exponentially increase the amount of relevant data.
“We’re going to have to build much larger GPUs,” he said, noting that the existing Hopper line is great before unveiling “a very, very large GPU.”
Huang, named after mathematician David Blackwell, said the new hardware shows that it’s not just a GPU, but a platform, but at its core is the new B200 chip.
“People think we make GPUs. That’s right. But GPUs don’t look like they used to.”
“We’re starting to do really great work on the next generation of robotics,” Huang said. “The next generation of robotics will likely be humanoid robotics. We now have the technology needed to imagine generalized human robotics.”
To get there, Huang announced Project GR00T, an acronym for the iconic Marvel Comics character. Generalist Robot 00 Technology. On stage, Huang presented the audience with several AI-powered robots, such as the Jetson Thor computing platform designed to perform complex tasks and interact naturally with humans.
Huang said Nvidia is integrating projects with several related departments within the chipmaker to develop a complete system for training humanoid robots that can learn from visual content, including video and virtual reality.
“We developed the Nvidia Isaac Lab to train GR00T at scale, a compute orchestration service that orchestrates training and inference workflows across Nvidia DGX systems for training, Nvidia OVX systems for simulation, and Nvidia IGX and Nvidia AGX systems for hardware. Built NVIDIA Osmo. “In-the-loop verification,” the company said.
Jetson Thor is part of an update to NVIDIA’s Isaac robotics platform, and Huang said they are training these robots for real-world adaptation in the Isaac Reinforcement Learning Gym.
“In some ways, I think human robotics might be easier,” Huang said. “The reason is that because we are structured in a very similar way, there is a lot more imitative training data that we can provide to the robot. “It is very likely that humanoid robotics will be much more useful in our world because we have created a world where we can interoperate and work well together.”
The virtual gym for training AI robots is called Omniverse, and Huang showed a demonstration that he described as “a little crazy, but very close to tomorrow.”
The demo showed a robotics building described as a “digital twin of a 100,000-square-foot warehouse,” populated by autonomous systems (humans and vehicles) that interact with each other. The environment will enable rapid development of both physical plant engineering and design, as well as better coordination of the work of humans and robots within it. And it’s more than theoretical.
“Siemens is the world’s largest industrial engineering and operations platform. Siemens is building the industrial metaverse, and today we announced that Siemens is connecting its leading Xccelerator to the Nvidia Omniverse.”
Apple’s new Vision Pro headset is also part of the picture.
“The Omniverse Cloud streams to Vision Pro… Vision Pro connected to Omniverse connects to the Omniverse portal. It’s really, really amazing,” Huang announced. “And all of these CAD tools and different design tools are now integrated and connected in Omniverse, so you can have these types of workflows.”
Huang said the automotive industry is an obvious place to leverage AI innovation.
“Everything that moves will be robotic. There’s no doubt about it. It’s safer and more convenient,” he explained. “One of the biggest industries will be automotive. For self-driving cars, we build the robotics stack from top to bottom, from computer systems to manufacturing.”
Huang said Jetson Thor could be deployed in self-driving cars. He added that the product will be launched next year and will be included in Mercedes and Jaguar Land Rover vehicles.
The Nvidia CEO also revealed that the company is building artificial intelligence platforms for other companies, including 1X Technologies, Agility Robotics, Apptronik, Boston Dynamics, Figure AI, Fourier Intelligence, Sanctuary AI, Unitree Robotics, and XPENG Robotics.
“Building a foundational model for a generic humanoid robot is one of the most exciting problems to solve in AI today,” Huang said. “Supporting technologies are coming together to enable the world’s best roboticists to take a giant leap forward toward artificial general robotics.”
Edited by Ryan Ozawa.
