Nvidia broadens ties with Korean component makers
LG Innotek's headquarters in Gangseo District, Seoul / Courtesy of LG Innotek
LG Innotek develops sensing solutions while Samsung Electro-Mechanics ramps up substrate production
Nvidia is expanding partnerships with domestic component makers LG Innotek and Samsung Electro-Mechanics, strengthening its supply chain for physical artificial intelligence (AI) applications and next-generation AI infrastructure.
LG Innotek unveiled a lineup of sensing solutions incorporating Nvidia's latest Holoscan Sensor Bridge (HSB) technology at the U.S. chipmaker's annual GPU Technology Conference (GTC) in Berlin on Tuesday (local time).
The solutions include 3-megapixel camera modules and a multimodal vision system that combines multiple cameras and sensors, targeting applications in physical AI and robotics.
Nvidia's HSB technology allows sensor data to be sent directly to AI computers, bypassing the central processing unit and potentially speeding up how robots and autonomous vehicles process information and make decisions.
The partnership reflects growing demand for advanced sensing technologies as AI expands beyond software and data centers into physical environments, where machines must perceive and respond to their surroundings in real time.
LG Innotek said it plans to expand collaboration with Nvidia on HSB-based sensing solutions for humanoid robot vision systems and robotic hands. The companies also plan to develop sensing solutions compatible with Nvidia's next-generation platforms and showcase them at CES and GTC next year.
“This collaboration is a valuable opportunity to bring LG Innotek’s differentiated integrated sensing solutions to global markets,” the company’s CEO Moon Hyuk-soo said.
“As a total solution provider for physical AI, encompassing not only sensing but also communication modules and semiconductor substrates, we will move quickly to establish a leading position in the physical AI solutions market.”
Separately, Samsung Electro-Mechanics announced a strategic business partnership with Nvidia to expand production of advanced semiconductor package substrates for AI chips and data center network switches.
Samsung Electro-Mechanics' manufacturing site in Suwon, Gyeonggi Province / Courtesy of Samsung Electro-Mechanics
The companies will jointly fund new production capacity, with Samsung Electro-Mechanics set to begin supplying Nvidia with the substrates in 2026 under a multiyear agreement.
Samsung Electro-Mechanics will establish dedicated production lines at its facilities in Sejong and Busan to manufacture substrates for Nvidia's computing chips and network switches used to connect chips at high speeds in data centers.
The company estimates its annual substrate supplies to Nvidia to grow by double digits on average from 2026. By 2028, Nvidia-related orders are expected to account for a substantial share of the company's total package substrate production capacity.
Advanced substrates are becoming increasingly important as AI chips grow more complex and require greater computing power, tighter power control and more efficient data transmission. The company said its manufacturing technologies include fine-line circuit fabrication and embedding passive components, such as multilayer ceramic capacitors and silicon capacitors, in substrates.
"The pace of the AI infrastructure buildout requires us to expand manufacturing capacity on a global scale,” Nvidia Executive Vice President of Operations Debora Shoquist said.
“Our work with Samsung Electro-Mechanics broadens our network of substrate partners and adds manufacturing capacity to help us scale production and deliver the accelerated computing platforms our customers need."