LG Innotek pushes deeper into AI substrates

LG Innotek is sharpening its push for North American artificial intelligence chip customers as demand for advanced semiconductor substrates lifts earnings and turns packaging capacity into a strategic battleground.

The Seoul-based electronics components manufacturer is preparing to use the IEEE Electronic Components and Technology Conference in Orlando, Florida, from May 26 to 29 to broaden contact with chip designers, server processor developers and packaging specialists. The event, one of the semiconductor industry’s key technical gatherings, comes as AI hardware makers confront tight supply of the substrates that connect processors, memory and circuit boards inside high-performance computing systems.

LG Innotek’s pitch is backed by firmer numbers. Its package solutions division, which includes semiconductor substrates, generated 437.1 billion won in first-quarter revenue, up 16 per cent from a year earlier. The company’s consolidated revenue rose 11.1 per cent to 5.53 trillion won, while operating profit jumped 136 per cent to 295.3 billion won. The result marked its strongest first-quarter sales performance and gave investors fresh evidence that the company is building a growth engine beyond its long-standing dependence on camera modules for premium smartphones.

The company’s substrate portfolio spans RF-SiP, FC-CSP and FC-BGA products. RF-SiP substrates are used in compact communications modules, while FC-CSP supports smaller high-performance chips and memory applications. FC-BGA, the area drawing the most attention from AI and server customers, is used for larger processors that require high-density wiring, thermal stability and reliable signal transmission.

AI accelerators and custom processors are becoming larger and more complex as cloud providers, chip start-ups and established semiconductor groups race to train and deploy bigger models. That has increased the importance of advanced packaging, where substrates serve as the physical and electrical foundation linking logic chips, high-bandwidth memory and boards. Constraints once associated mainly with wafer fabrication are now appearing across the back end of the semiconductor supply chain, including interposers, substrates, build-up films and assembly capacity.

LG Innotek is trying to position itself as a credible alternative supplier at a time when leading substrate makers in Japan and Taiwan are heavily committed to existing customers. The company has been expanding high-value substrate capacity and wants the package solutions business to make a contribution to operating profit comparable to its optical solutions division within five years. That target signals a deliberate shift from cyclical consumer electronics exposure towards higher-margin semiconductor and AI infrastructure demand.

Its first-quarter performance still shows the scale of the transition ahead. Optical solutions, mainly camera modules, remained the dominant business with revenue of 4.61 trillion won, up 11.4 per cent from a year earlier. Package solutions, despite faster strategic momentum, accounted for less than a tenth of total sales. Mobility solutions, covering vehicle cameras, lighting modules and related components, posted 487.1 billion won in revenue, up 4.2 per cent, supported by an order backlog of 19.2 trillion won at the end of last year.

The Orlando conference gives LG Innotek a timely platform to engage potential customers as packaging technology moves closer to the centre of chip design decisions. ECTC’s 2026 programme covers chiplets, 2.5D and 3D integration, hybrid bonding, glass and dielectric materials, thermal management, high-density redistribution layers and large-package manufacturing. Those areas closely match the technical challenges facing AI processors, which need higher bandwidth, lower power loss and improved heat dissipation.

Competition will be intense. Samsung Electro-Mechanics, Ibiden, Unimicron, Nan Ya PCB, Kinsus and other substrate makers are also seeking to capture AI-linked demand. Some rivals have deeper experience in high-end server substrates, while others benefit from long-standing relationships with major foundries and chip designers. LG Innotek’s challenge is to prove that its technology, yields and capacity road map can meet the reliability standards required by AI hardware customers operating at scale.