In March 2026, ABB Robotics and NVIDIA announced a partnership that could make “physical AI” a real factory technology rather than a conference buzzword. ABB is adding NVIDIA Omniverse libraries to its RobotStudio software and packaging the result as RobotStudio HyperReality, planned for release in the second half of 2026. In NVIDIA’s own definition, physical AI is AI that helps machines perceive, understand, reason, and act in the physical world. In plain English, this means robots that are not just precisely programmed, but trained to handle the messy, changing reality of actual production lines. (blogs.nvidia.com)

The heart of the deal is the attempt to close manufacturing’s stubborn “sim-to-real” gap. Factories have long used simulation, but virtual models often fail when lighting changes, materials behave differently, or product variations appear on the real line. ABB says HyperReality combines photorealistic simulation, synthetic data generation, and ABB’s virtual controller, which runs the same firmware as the physical robot, to achieve up to 99% correlation between virtual training and real deployment. The company also claims setup and commissioning time can be cut by up to 80%, engineering and production costs by up to 40%, and time to market by as much as 50%. Those are ambitious numbers, but even a partial improvement would be significant for manufacturers facing labor shortages, shorter product cycles, and rising pressure to automate. (blogs.nvidia.com)

The most interesting impact may be flexibility. Traditional industrial robots are excellent at repeating stable tasks, but they struggle when parts, lighting, or layouts change. By training robots in physically accurate digital twins and feeding them synthetic data, manufacturers can prepare systems for variation before a single robot is installed. Foxconn is already piloting the ABB-NVIDIA technology in consumer electronics assembly, where delicate components and frequent design changes make automation difficult. Another pilot, with U.S.-based Workr, suggests that this approach may not be limited to giant corporations; smaller manufacturers could also gain access to smarter automation with less specialist programming. (blogs.nvidia.com)

Seen more broadly, the ABB-NVIDIA partnership points to a new model for manufacturing: fewer rigid machines, more adaptive robotic systems. NVIDIA said at GTC on March 16, 2026 that ABB is part of a wider ecosystem of major robotics companies building on Omniverse, Isaac, and Jetson technologies for production-scale physical AI. If that trend continues, the factory of the future may be designed less like a fixed machine and more like a trainable intelligence system—one that learns in simulation first, then performs in the real world. (nvidianews.nvidia.com)