NVIDIA, U.S. Manufacturers Forge Physical AI for Reindustrialization

NVIDIA has announced a strategic partnership with U.S. manufacturing and robotics leaders to deploy physical AI and Omniverse digital twins, aiming to accelerate American reindustrialization and address critical labor shortages in the industrial sector. The company detailed these efforts in a press release, highlighting how digital twins and collaborative robots are transforming factory operations.

In a keynote at GTC Washington, D.C., NVIDIA CEO Jensen Huang declared, "AI is transforming the world's factories into intelligent thinking machines — the engines of a new industrial revolution. Together with American's manufacturing leaders, we're building physical AI, Omniverse digital twins and collaborative robots that will drive productivity, resilience and competitiveness across the U.S. industrial base."

NVIDIA expanded its "Mega" Omniverse Blueprint to include factory-scale digital twin capabilities, with Siemens as the first to develop supporting software (beta, part of Siemens Xcelerator). This technology enables engineers to design and operate large-scale digital twins that combine realistic 3D models with live operational data. FANUC and Foxconn Fii are among the first robot manufacturers to provide 3D, OpenUSD-based digital twins of their equipment, allowing manufacturers to drag and drop robotic systems into virtual factory models.

Major manufacturers are already implementing these technologies: Caterpillar uses Omniverse for digital twins of factories and supply chains to enable predictive maintenance and dynamic scheduling, while Lucid Motors employs the platform for real-time factory optimization and AI robotics training. Toyota leverages idealworks' iw.sim technology, which integrates Omniverse capabilities, to create digital twins of its Georgetown, Kentucky, facility for exploring complex automation scenarios. TSMC is using Omniverse to accelerate semiconductor fabrication design and the NVIDIA Isaac platform for robotics development at its Phoenix facility.

Robotics companies including Agility Robotics, Amazon Robotics, Figure, and Skild AI are adopting NVIDIA's three-computer architecture to build collaborative robot workforces. Figure, for instance, is collaborating with NVIDIA to develop humanoid robots using the Isaac platform for simulation and training, with the goal of creating large-scale fleets capable of industrial support tasks. Amazon Robotics uses Omniverse libraries to shorten development cycles for manipulation systems and mobile robots running on NVIDIA Jetson platforms.

NVIDIA showcased Foxconn's use of Omniverse to design, simulate, and optimize a new 242,287-square-foot Houston facility dedicated to manufacturing NVIDIA AI infrastructure systems. This project exemplifies how digital twins accelerate physical factory deployment. According to NVIDIA, the U.S. industrial sector has seen $1.2 trillion in announced investments toward expanding domestic production capacity in 2025, led by electronics providers, pharmaceutical companies, and semiconductor manufacturers.

Industrial software provider Belden has implemented Accenture's Physical AI Orchestrator, combining NVIDIA Omniverse libraries, the NVIDIA Metropolis platform, and agentic AI, to create virtual safety fences and real-time quality-inspection systems. This integration enables instant hazardous zone monitoring and quality control in manufacturing environments. Wistron is using Omniverse to build digital twins for factory optimization, demonstrating the platform's broad applicability across the manufacturing ecosystem.

This initiative represents a fundamental shift from traditional industrial automation toward closed-loop physical AI systems. As one industry analyst noted, "The most meaningful shift here is not the hardware scale, but the integration discipline behind it. Robotics only becomes industrially credible when perception, policy learning, simulation fidelity, and deployment orchestration operate as one system." The three-computer architecture enables continuous learning from physical environments through digital twin feedback, moving beyond the "train in the cloud, hope in the factory" model.

The convergence of high-fidelity simulation (Omniverse), centralized training, and robust edge deployment creates a new MLOps paradigm for industrial robotics. This approach addresses the critical challenge of model drift in real-world environments by continuously validating digital twins against physical operations. As the U.S. seeks to rebuild manufacturing competitiveness amid global supply chain disruptions, NVIDIA's platform provides the infrastructure for factories that can adapt in real time to changing conditions and labor constraints.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn