LG’s manufacturing expertise to combine with NVIDIA’s robotics and physical AI technology at new Data Factory in Seoul
SEOUL: LG Electronics has announced an expansion of its robotics collaboration with NVIDIA, combining LG’s manufacturing and robotics capabilities with NVIDIA’s physical AI and robotics technology stack. LG recently hosted senior NVIDIA officials at its Data Factory, currently under construction at the company’s Yangjae R&D Campus in Seoul, South Korea. The two companies reviewed the progress of their collaboration and explored opportunities to generate further synergies in robotics and physical AI.
The facility, scheduled to become fully operational by the end of 2026, will serve as a major hub for generating, collecting and refining robot-learning data. LG has deployed its self-developed and field-proven LG CLOiD™ home robots at scale as part of the initiative.
“Through the synergy built on ‘One LG’ – bringing together core capabilities across the Group – and strategic collaboration with global partners, we will secure our competitiveness in physical AI and become a comprehensive robotics solutions provider,” said Lyu Jae-cheol, CEO of LG Electronics.
A Data Factory for Robot Learning
The Yangjae Data Factory will feature dedicated training spaces where robots can learn and repeat a range of tasks, alongside specialised areas for validating and refining the data generated during these activities.
A replicated home environment will allow LG CLOiD robots to learn household tasks such as cleaning. The facility will also include a simulated manufacturing environment modelled on LG’s washing machine plant in Tennessee, where robots will move, stack and assemble various components.
LG CLOiD robots are also being deployed across logistics automation solutions developed by LG CNS, while LG Innotek will use dedicated spaces at the facility to train robotic hands.
Data collected from these environments will be integrated with NVIDIA’s robotics stack, where it can be augmented and synthesised into high-quality training data for robot learning.
NVIDIA Technologies to Support Data-to-Deployment Robotics
The Data Factory will function as a physical learning environment and a forward base for expanding robot-learning data gathered from LG’s manufacturing, logistics and home appliance operations worldwide.
LG will use NVIDIA’s physical AI technologies across the robotics development process, including:
- NVIDIA Omniverse libraries
- NVIDIA Cosmos open world models
- NVIDIA Isaac robotics development platform
Spanning four floors — including one basement level and three floors above ground — the facility will cover approximately 10,000 square metres and is expected to house several hundred robots by the end of 2026.
LG also plans to continue upgrading its data-learning systems as it expands its robotics capabilities.
100,000 Hours of Robot Training Data Expected
By the end of the year, LG expects the combined volume of training data collected directly at the Data Factory and synthetically generated using NVIDIA Cosmos open world models to reach approximately 100,000 hours.
The volume is equivalent to roughly 12 years of data and will be used to further advance LG’s Robot Foundation Model (RFM), which supports the performance and learning capabilities of humanoid robots.
The expanded collaboration reflects LG’s broader push into physical AI and robotics, leveraging decades of manufacturing expertise alongside NVIDIA’s AI, simulation and robotics development technologies.