Jensen Huang and LG’s chairman Kwang Mo Koo shook hands in Santa Clara on August 13, and what came out of that meeting is one of the broader physical AI commitments we’ve seen from a legacy electronics company. LG announced it’s building a next-generation bipedal humanoid robot on NVIDIA’s Isaac GR00T foundation model, with a public debut planned for Q1 2027.
The robot runs on NVIDIA Jetson Thor for onboard compute and uses NVIDIA Halos for safety. LG is pulling in hardware from across its own subsidiaries — actuators and sensors from LG Innotek, batteries from LG Energy Solution. So this isn’t a clean-room prototype. LG is treating it as a full-stack internal engineering project.
More than a robot deal
The MOU between LG and NVIDIA spans three areas. First, robotics: the humanoid plus a wheel-based CLOiD robot being deployed to LG’s Tennessee washing machine plant for real-world validation this year. Second, AI infrastructure: LG plans to build an AI factory reference site using NVIDIA’s DSX architecture and Vera Rubin GPUs by H1 2027, followed by an 80 MW facility in Cheonan by 2028. Third, automotive: a new high-performance compute platform for AI-defined vehicles built on NVIDIA DRIVE Hyperion.
Why this matters for the industry
LG is not the first company chasing humanoids. Figure, Physical Intelligence, Agility Robotics, and Boston Dynamics are all further along on actual robot deployments. But LG has something most pure-play robotics startups don’t — manufacturing scale, supply chain control, and real factory floors to test in. The Tennessee plant validation is the detail worth watching. Real production environments break robots in ways labs never do.
The NVIDIA GR00T angle also signals where foundation models for robotics are heading. LG says it will use GR00T while building its own Robot Foundation Model in parallel. That’s a smart hedge — adopt the industry baseline now, build proprietary capability over time.
