Zhan What competitive advantage do you think Nvidia has in building these technologies?

[rifat28dddd]

The team I currently lead is called GER, which stands for Generalist Embodied Agent Research. Simply put, our team’s work can be summed up in three words: “action generation.” Because we build embodied intelligent agents, and those agents take actions in different worlds. If these actions take place in a virtual world, it’s AI and simulation, if they take place in the physical world, it’s robotics. In fact, at the GTC conference in March of this year, Jensen showed off a project called Project Groot in his keynote speech.



This is an important effort by Nvidia to build a basic model cameroon phone numbers of humanoid robots, and that’s what the GER team is currently working on. . We hope to build AI brains for humanoid robots and beyond. Stephanie Jim Fan This is a very good question. First, there’s no doubt about the computing resources. All of these basic models require significant computing resources to scale.



We believe in the “Law of Scaling,” which is similar to the LLM scaling law, but the Scaling Law of Embodied Intelligence and Robotics has yet to be studied, so we are investing efforts in this area. Nvidia’s other strength is analog technology.Nvidia was a graphics company before it became an AI company, so we have years of experience building simulations (like physics simulations, rendering) and GPU acceleration in real time. So when we build robotics, we use simulation a lot.