NVIDIA AI Podcast AI for Robotics and Manufacturing | GTC Live Washington, D.C. Chapter 5
51 snips
Nov 11, 2025 
Join industry leaders as they explore the future of manufacturing and robotics. Yong Lu, CEO of Foxconn, shares insights on transitioning to AI-driven factories in the U.S. Peter Kurta from Siemens discusses the transformative role of digital twins and sensors. Brett Adcock of Figure AI reveals innovative humanoid robotics and the manufacturing hurdles ahead. Aki Jain of Palantir explains creating ontologies to optimize AI workflows. Together, they highlight a new era of digital intelligence merging seamlessly with physical action.
AI Snips
Chapters
Transcript
Episode notes
AI-Native Factories And Digital Twins
- AI-native factories pair digital twins and pervasive sensors to optimize production in real time.
- This approach boosts speed, productivity, sustainability, and addresses US labor constraints.
Running Robot Inference Offline
- Figure trains large-scale human-like datasets and runs inference on NVIDIA GPUs locally on robots.
- Brett describes running policies on robots without any network connection for real-world operation.
Why Neural Nets Are Essential For Humanoids
- General-purpose humanoid control is intractable by hand-coded rules due to astronomical state spaces.
- Brett argues neural nets are necessary to manage the robot's combinatorial joint states.