Why LG Innotek’s FC-BGA AI substrate wants to be the quiet engine behind the next GPU wave

Reviewed: ad hoc news Accessory & Components desk. Edited and checked on 2026-06-17, 11:12. Details in the imprint.

LG Innotek’s FC-BGA AI substrate is one of those parts you never see, yet it sits just millimeters below the GPUs that drive today’s AI models, quietly routing thousands of signals and soaking up brutal heat and power swings.

What LG’s AI substrate actually does

In simple terms, the FC-BGA AI substrate is the wide, flat base on which cutting-edge AI GPUs and accelerators are mounted, providing ultra-dense wiring between the chip and the server motherboard.

LG Innotek focuses here on large flip-chip ball grid array formats that can handle high input-output counts, strong power delivery and fast memory interfaces needed for AI training and inference workloads.

Built for brutal AI server conditions

AI accelerators pull enormous current, spike their power draw and then idle again, and the substrate has to survive these thermal and electrical cycles for years without warping or cracking.

According to Korean business reports, LG Innotek is expanding into high-value FC-BGA substrates for AI servers, specifically mentioning products for AI training, inference and server networks as a growth focus.

Collaboration with a North American tech giant

The most eye-catching detail is that LG Innotek is co-developing large AI substrates with a major North American Big Tech customer, aiming for tangible results by the end of this year.

Industry coverage points to large-format FC-BGA for data-center GPUs, where tight cooperation on design rules, reliability targets and manufacturing tolerances is essential for qualification.

Why FC-BGA is suddenly center stage

For years, LG Innotek was mainly known for camera modules and smartphone boards; now, the package solutions division is being pushed as a separate growth engine built around RF-SiP, FC-CSP and especially FC-BGA substrates.

AI server boards need much larger substrates with many more layers than classic PC parts, so each design win can translate into higher revenue per unit and stickier long-term supply relationships.

Ambitious profit targets for substrates

Management has publicly set a goal for the package solutions business - which includes the AI substrate line - to reach around 3 trillion won in revenue by 2030 and 1 trillion won in operating profit by 2031.

That profit ambition alone would exceed the group’s total consolidated operating profit in the most recent year, underlining how central AI-focused substrates have become in LG Innotek’s strategy.

Positioning against Japanese and Taiwanese rivals

Japan and Taiwan historically dominated advanced FC-BGA, so LG Innotek is entering a tough arena with entrenched players and tight customer relationships built over decades.

The company counters this with heavy front-loaded capex, process know-how adapted from communication substrates, and direct co-development commitments with big cloud and chip customers.

What this means for server buyers

For hyperscalers and AI startups, a broader supply base for high-end FC-BGA could ease bottlenecks that currently delay GPU shipments and inflate system prices.

If LG Innotek hits its mass-production schedule for AI training and inference substrates next year, some AI server platforms could ship with more predictable lead times and potentially lower total system cost.

Context and stock reference

LG Innotek remains best known for smartphone components, but the strategic spotlight is clearly shifting toward high-value semiconductor substrates for AI servers and other advanced computing applications. Shares of LG Innotek Co Ltd (KR7011070000) trade on the Korea Exchange.

This article was AI-assisted and editorially reviewed. Product information without guarantee; prices and availability may change at short notice. No investment advice, no buy or sell recommendation. Stock-market transactions involve risks up to total loss.