Nvidia on humanoid robotics and AI brains (with Ed Lello)

Nvidia's VP of Robotics states that while the physical hardware for humanoid robots is largely ready, the industry is still waiting for a breakthrough in AI 'brains'—a general-purpose, accurate model. Nvidia has been working on the Jetson platform for over a decade to provide the necessary compute, but the 'ChatGPT moment' for robotics has not yet arrived.

Yields

NDX

RUT

Oil

Metals

USD

Nvidia 1.0
Information Technology
Deepu Talla 9.0

Nvidia on humanoid robotics and AI brains (with Ed Lello)

Nvidia wants to make humanoid robots safer. The physical hardware is largely ready, but the industry is hunting for a breakthrough in AI brains.

Ed Lello
Deepu Talla

Robotics is a three-computer problem: the computer inside the robot (the brain), the computer for training the brain, and the computer for testing in simulation. We've been working on the third computer (Jetson) for over a decade.

It needs high compute for larger models, real-time performance, safety, energy efficiency, and general-purpose programmability. We have over 2.5 million developers and 10,000 companies building robots on the platform.
What is the biggest barrier to progress in humanoid robotics: software or hardware?

Ed Lello
Deepu Talla

The number one problem remains the brain. It needs to be general purpose and accurate. We haven't had the 'ChatGPT moment' for robotics yet, but it is right around the corner.

Tremendous mechatronics miracles have been done, except for the hands which are still being worked on.

SignalTube

markets at a glance