Jensen Huang
Robot learning, simulation and the infrastructure behind physical AI.
We follow how Huang’s public vision develops into products, developer adoption and documented use. Platform announcements establish what NVIDIA offers; customer results help establish what those tools achieve.
A timeline of the argument
Huang describes physical AI as a coming wave of AI at COMPUTEX. Source
Cosmos gives the vision a more explicit training-data and world-model component. Source
NVIDIA’s investor release describes Cosmos 3’s combined reasoning and generation approach. Source
NVIDIA publishes an account of Skild using Cosmos and Isaac Lab, adding a concrete example to the platform’s adoption story. Source
NVIDIA announces Isaac ROS 5.0, adding agent-assisted developer workflows. The software release supplies another implementation step; partner accounts still need measured customer results. This is a company update, not a new Huang quotation. Source
Published September interview: Pickle’s CTO describes Isaac simulation in warehouse testing. Builder testimony, not a new Huang statement or independently measured customer result. Source
At London Build-a-Claw, Huang argues for limited agent permissions and monitoring. NVIDIA dates the event October14; TechRadar reports its robotics analogy October15. This does not establish safe customer robot operation. Source
Berlin: Huang and Merz discuss industrial data and manufacturing AI, alongside a government–NVIDIA declaration of intent. The initiative is a plan for collaboration, not verified customer deployment. Source
Read the analysis
Germany and NVIDIA’s physical AI pact: what changes now?
What Germany’s October 19 declaration with NVIDIA establishes, how Huang frames industrial data, and which projects and customer results to watch next.
Jensen Huang’s physical AI vision: what changed from 2024 to 2026?
From Cosmos and warehouse simulation to Huang’s October agent-control argument: what NVIDIA’s physical AI tools establish, and what still needs real-world evidence.
Brett Adcock’s compute bet: what Figure still has to prove
What Figure’s Index collection, Helix 2.5 tests and Nscale commitment establish—and which measurements could show that more data produces more useful robots.
Sources & context
- Huang’s COMPUTEX 2024 addressNVIDIA · Jun 2, 2024 · Primary sourceNVIDIA Robotics Adopted by Industry Leaders
- Cosmos launch announcementNVIDIA · Jan 6, 2025 · Primary sourceOpening paragraphs and Huang statement
- Cosmos 3 launch announcementNVIDIA · Jun 1, 2026 · Primary sourceOpening paragraphs; A New Architecture; Availability
- Skild AI and NVIDIA: S1 and the physical AI development toolsNVIDIA · Sep 10, 2026 · Primary sourceDeepak Pathak statement; NVIDIA Technology Across the Development Cycle
- NVIDIA Isaac ROS 5.0 Advances Agentic, Open Source Robotics DevelopmentNVIDIA · Sep 22, 2026 · Primary sourceDated announcement; agent development workflows; Magna integration; available-now closing paragraph
- Pickle Robot CTO Ariana Eisenstein interviewJustin Hodges · Oct 7, 2026 · MediaSimulation section; interview took place September2026, exact day undisclosed; author disclosure
- Build-a-Claw Comes to UK DevelopersNVIDIA · Oct 14, 2026 · Primary sourceOctober14 London section, dated1:11pm PDT; not the separate October15 Stockholm entry.
- Jensen Huang links OpenClaw and robotics, calling for controlsTechRadar · Oct 15, 2026 · Reported by this outletMike Moore; paragraphs on London event, physical robot analogy and privileges.
- German government: cooperation on industrial AI with international partnersGerman Federal Government, Press and Information Office · Oct 19, 2026 · Primary source · de sourcePress release189; paragraphs on ministries, declaration, Industry4.0 and specialized open models.
- Germany Wants to Lead the World in Industrial AI: How NVIDIA Will HelpNVIDIA · Oct 19, 2026 · Primary sourceBerlin meeting, declaration, industrial-data argument and planned startup investments.
Timeline updated Oct 20, 2026. A selected timeline, not a complete archive of every public statement.