NVIDIA Showcases Physical AI Breakthroughs at National Robotics Week 2026

NVIDIA used National Robotics Week 2026 to showcase a series of breakthroughs in physical AI — the company's term for AI systems that operate in and interact with the real world. Highlighted across the week of April 7, the announcements span robot learning, simulation infrastructure, and physical foundation models that are rapidly closing the gap between virtual training and real-world deployment.

Central to NVIDIA's physical AI push is its Isaac simulation platform, which enables robots to train in photorealistic virtual environments before being deployed on physical hardware. The company demonstrated significant improvements in sim-to-real transfer — the ability of a robot trained in simulation to perform reliably on actual hardware — a long-standing bottleneck in robotics research that has slowed commercial adoption.

Among the highlighted robots was Moya, a biomimetic humanoid developed by the Shanghai Engineering Research Center of Humanoid Robots, featuring artificial muscles, a flexible spine, and a skeletal structure modeled on human anatomy. Unlike conventional motor-driven humanoids, Moya's design aims to replicate the fluid, energy-efficient movement of biological systems — a direction NVIDIA's simulation tools are being optimized to support.

NVIDIA also highlighted the growing role of foundation models in robotics, with general-purpose models increasingly being fine-tuned for specific manipulation and navigation tasks. The company projects that by 2027, a significant portion of industrial robotics deployments will leverage AI foundation models trained on NVIDIA infrastructure, marking a profound shift in how robots are programmed and deployed at scale.