LG Innotek CEO pushes physical AI as new growth engine

LG Innotek CEO Moon Hyuk-soo speaks to reporters after the company’s shareholders' meeting at its headquarters in Seoul, Monday. Courtesy of LG Innotek

LG Innotek is accelerating its transition from a component supplier to a full-fledged solutions provider that integrates hardware and software capabilities, as it bets on physical artificial intelligence (AI), including robotics and autonomous driving, as a key driver of future growth.

During the company’s shareholders' meeting at its headquarters in Seoul on Monday, CEO Moon Hyuk-soo outlined a strategy to place physical AI as a core pillar of LG Innotek’s future business with a goal of delivering meaningful results in the next few years.

“A business model where we simply win bids to supply developed components is losing its competitiveness,” he said, speaking to the press after the meeting.

“Based on the innovative technologies and product lineup we have built over time, LG Innotek aims to shift its business paradigm toward becoming a company that provides optimal ‘solutions’ tailored to customer needs.”

LG Innotek holds scalable technologies, such as sensing, substrate and control, that have been refined through years of collaboration with global tech clients. Identifying these technologies as the company’s key growth asset, Moon has been accelerating its push into the physical AI market, placing particular emphasis on emerging opportunities in autonomous driving and robotics to expand these capabilities.

Moon said that LG Innotek is already in talks with major customers in the United States and Europe to supply its technologies for robot applications.

“We are actively discussing complex sensing modules, including light detection and ranging (LiDAR) and cameras, with major customers in the U.S. and Europe,” he said.

LG Innotek's radio frequency system-in-package / Courtesy of LG Innotek

He said the company is also ramping up production of humanoid robot parts, with mass production of robot components expected to begin around 2027 or 2028.

“We have already entered the early mass production phase for physical AI applications such as humanoid robots, supplying them in the hundreds, and we are working to identify and resolve issues arising in the field,” he said.

“We expect it will take about three to four years before the business will generate (revenue) figures that are meaningful.”

Earlier this year, in his New Year’s address, Moon pledged to strengthen the high-margin package solutions unit, bringing its operating profit contribution to the level of the optical solutions business within five years.

The company’s package solutions division saw its operating profit surge 82 percent year-over-year in 2025 to 128.9 billion won ($85 million), driven by strong demand for semiconductor substrates.

To ride favorable market conditions, Moon said the company will actively respond to the rising demand with plans to expand capacity.

“The existing semiconductor substrates that use glass fiber, such as radio frequency system-in-package, are already running near maximum capacity, while capacity for flip chip-ball grid array and other substrates used in servers is expected to be expanded around the second half of next year,” he said.

“The company is preparing for this in advance, and once the expansion goes ahead, total capacity will be increased to about double the current level.”