NVIDIA and Doosan Group are widening an existing physical AI partnership into a broader infrastructure plan that now touches robotics, construction equipment, power systems, and materials for AI data centers.

The new NVIDIA blog post says the companies are expanding work across Doosan Robotics, Doosan Bobcat, Doosan Enerbility, and Doosan Corporation Electro-Materials BG. The robotics portion centers on NVIDIA Isaac Sim, Isaac Lab, Cosmos, Newton, and related simulation-to-real workflows. NVIDIA says the companies are also looking at reference use cases such as depalletizing and sanding, plus dual-arm and humanoid robot form factors.

The construction-equipment track is less concrete, but still specific: Doosan Bobcat plans to explore NVIDIA physical AI technology for equipment used in construction, landscaping, agriculture, and material handling. The goal is specialized world models that let machines perceive changing operating environments and perform tasks with more autonomy.

The infrastructure side is the more unusual part of the announcement. Doosan Enerbility is exploring support for NVIDIA AI factories and the DSX platform through gas turbines, steam turbines, small modular reactors, and fuel-cell systems. Doosan's materials unit is also being positioned around copper-clad laminate for printed circuit boards used in AI data-center equipment.

This is not a product launch or deployment milestone. Doosan had already announced a strategic NVIDIA partnership in 2025 to integrate physical AI into construction equipment, power generation, and robotics. The verified news is that the scope has expanded into a group-wide AI factory frame, tying robots, industrial machines, power supply, and data-center materials into one physical AI roadmap.