Doosan Group said on the 8th it will expand its collaboration with Nvidia in physical AI, Robotics, and AI factories. Doosan sees its core businesses in energy, electronic materials, and Robotics as tied to Nvidia's push for AI factories and is pursuing ways to link Nvidia's accelerated computing and physical AI platform with Doosan's product, technology, and manufacturing capabilities.
In energy, Doosan Enerbility and Doosan Fuel Cell will be the pillars of cooperation. Doosan Enerbility has gas turbines and Small Modular Reactor (SMR) technology, while Doosan Fuel Cell has Hydrogen Fuel Cell technology. Doosan is pursuing ways to link these energy solutions with Nvidia's DSX (a design architecture for AI factories integrating computing networks, storage, and power) AI factory platform. The two companies plan to discuss power supply design for AI factory build-outs, optimization of power generation facilities, and securing low-carbon power sources.
In Robotics, Doosan Robotics is developing an agentic robot operating system using Nvidia's physical AI technology. An agentic robot operating system is software that helps robots recognize and judge their work environment on their own before moving. Doosan Robotics is using Nvidia's Isaac Sim and Isaac Lab robot simulation tools, as well as the Cosmos world model, NeMo, and Jetson Thor-based edge devices.
Based on this, Doosan and Nvidia are discussing developing reference robot solutions for industrial tasks that require precision, such as depalletizing and sanding. Depalletizing is the task of a robot moving or sorting items stacked on a pallet, and sanding is smoothing a surface. The two companies will work to improve the accuracy and autonomy of industrial robots using a robot operating system with perception, inference, and simulation capabilities.
Doosan plans to extend the Robotics collaboration to Doosan Bobcat's construction, landscaping, agriculture, and logistics equipment. Doosan will integrate Nvidia's physical AI technology into the equipment and push to develop world models specialized for industrial sites. A world model is technology that helps AI learn a virtual environment similar to reality to predict equipment movements and work conditions. Doosan aims to advance technology that enables equipment to recognize diverse work environments, assess situations, and operate autonomously.
In electronic materials, Doosan's Electronics BG is seeking collaboration opportunities related to Nvidia's next-generation AI data center infrastructure. Doosan's Electronics BG produces copper-clad laminate (CCL), a key material for printed circuit boards (PCB) used in Nvidia's AI infrastructure. CCL is necessary for AI accelerators to operate stably. Doosan is building a new production base in Thailand to expand CCL output, targeting mass production in 2028.
Park Jeong-won, chairman of Doosan Group, said, "Doosan Group has been steadily advancing the technologies needed in the AI era in energy, Robotics, and advanced materials, based on manufacturing capabilities accumulated over a long period," adding, "As the AI factory era arrives, our collaboration with Nvidia will greatly help us apply AI in our business areas and seek business opportunities."
Jensen Huang, Nvidia's chief executive officer (CEO), said, "Korea is a global manufacturing hub, and physical AI will be a new growth opportunity for corporations that build, move, and power the world," adding, "By combining Nvidia DSX and physical AI with Doosan's energy, Robotics, and advanced materials businesses, Doosan Group can secure new growth engines in key AI-era fields such as intelligent robots, autonomous industrial equipment, and next-generation infrastructure."