Nvidia's Madison Huang visits LG's robotics data hub in Seoul, deepening physical AI alliance

Madison Huang, senior director of Omniverse and robotics product marketing at Nvidia and eldest daughter of CEO Jensen Huang, visited the LG Data Factory at LG Electronics' Yangjae R&D Campus in Seocho-gu, Seoul, on Tuesday morning.

The visit came just four days after LG Group Chairman Koo Kwang-mo and Jensen Huang signed an MOU at Nvidia's headquarters in Santa Clara, California, on Aug. 13 (local time), pledging cooperation across three strategic business areas: physical AI, AI infrastructure and mobility.

Madison Huang leads Nvidia's push to expand its physical AI business. Her visit to the LG Data Factory — the nerve center of LG Group's robotics research and development — signals that the two companies' robotics partnership is advancing at an accelerating pace.

LG Electronics CEO Ryu Jae-cheol, LG CNS CEO Hyun Shin-kyun and LG Science Park President Jeong Su-heon — along with other top executives from LG affiliates — hosted Madison Huang and key Nvidia officials at the LG Data Factory on Tuesday to review the direction of the two companies' collaboration and discuss ways to generate synergies.

"We will secure physical AI competitiveness and reinvent ourselves as a total robotics solution provider through synergies grounded in 'One LG' — which unites the group's core capabilities — and through strategic collaboration with global partners," Ryu said.

Having forged what has been described as an "AI alliance" in the United States, the two companies quickly shifted the stage to Korea, where their executives moved to give concrete shape to robotics business cooperation.

Particular attention has been drawn to the choice of the Yangjae Data Factory as the meeting venue — a facility LG Electronics is building with the goal of bringing it to full operation by year-end. LG Electronics reviewed the factory's operational status with Madison Huang and reaffirmed its commitment to the partnership.

The Data Factory is being built across one below-ground floor and three above-ground floors, with a total floor area of 10,000 square meters. About 50 units of LG Electronics' humanoid robot CLOiD are currently deployed there, generating, collecting and learning from data, according to sources familiar with the matter.

CLOiD performs cleaning tasks in a space designed to replicate a real home, and carries out parts-handling, stacking and assembly training in a manufacturing environment that mirrors the production line at LG Electronics' washing machine factory in Tennessee. Robots are also deployed in LG CNS's logistics automation solution space and in LG Innotek's robotic hand training area.

To secure high-quality robot training data, LG Electronics plans to increase the number of robots deployed at the facility to several hundred by year-end and to continuously upgrade its data learning framework.

LG Electronics aims to use the Yangjae Data Factory to build a world-class data flywheel — a system that repeatedly trains on and refines high-quality data in a self-reinforcing cycle.

The factory combines decades' worth of high-quality data that LG Electronics has accumulated from manufacturing and logistics sites and home appliances around the world with Nvidia's robotics solutions to synthesize and amplify that data.

The process draws on three Nvidia tools: the Omniverse Library, the Cosmos open-world model and the Isaac open robotics development platform.

LG Electronics expects the total volume of training data — combining data collected directly at the Yangjae Data Factory with virtual data generated through amplification and synthesis — to reach 100,000 hours by year-end, equivalent to roughly 12 years' worth of data.

Through this effort, LG Electronics plans to advance its robot foundation model, which underpins humanoid robot performance, and to continuously strengthen its robotics competitiveness.

In its semiannual report this year, LG Electronics mentioned its collaboration with Nvidia for the first time, saying it is "pursuing a joint project aimed at combining our robot hardware technology and manufacturing capabilities with Nvidia's physical AI technology stack, advancing manufacturing-site robots from proof of concept through to real-world deployment."