LG Innotek

LG Innotek started in 1970 as Goldstar Precision, a components maker building switches out of a single factory in Busan. Six name changes and five decades later, it's an LG Group affiliate with 22 sites worldwide and just over 19,000 employees — and, less flatteringly, a company that gets roughly 80% of its revenue from camera modules sold to a single customer: Apple. Every iPhone's Face ID sensor and rear camera cluster almost certainly has an LG Innotek part in it.

That concentration is the reason the rest of this story matters. Being this dependent on one customer's product cycle is uncomfortable, and LG Innotek has spent the last several years visibly trying to build a second leg to stand on. Semiconductor substrates — the unglamorous layer that sits between a chip and the board it plugs into — is the leg it picked, and glass is the newest bet within that leg.

Substrates are LG Innotek's oldest business, not its newest — the company has been making package substrates for 40-plus years and currently holds the number one global market share in two specific substrate types: RF-SiP (radio-frequency system-in-package, used in 5G mmWave antenna modules) and AiP (antenna-in-package). It also has a strong position in FC-CSP, the substrate format used in high-performance mobile application processors.

In 2022 the company decided to go after a bigger prize: FC-BGA (flip-chip ball grid array), the substrate format used under CPUs, GPUs, and AI accelerators — a market Fuji Chimera Research expected to more than double from $8 billion in 2022 to $16.4 billion by 2030. LG Innotek's first move was a $336 million investment to convert LG Electronics' old Gumi No. 4 factory into what it calls the "Dream Factory" — a highly automated FC-BGA line where autonomous robots handle nearly every step, because a single stray eyelash can ruin an ultra-fine circuit. Full-scale production started there in February 2024, and the company began shipping PC-grade FC-BGAs to North American big-tech customers by the end of that year.

Glass fits into this story as the next material generation inside the same FC-BGA roadmap, not a separate business. LG Innotek's stated plan is to internalize glass core technology — alongside redistribution-layer (RDL) circuitry and device-embedding techniques — by 2027, and it has a pilot line running at its Gumi facility aimed at exactly that. To speed up the glass-specific parts, which sit outside its own core competency, LG Innotek partnered with UTI, a Korean precision glass processor that also does work for Corning.

The company breaks its business into components (camera modules, sensors — still the overwhelming majority of revenue, still mostly Apple), substrate materials (the FC-BGA and glass-core business described above), and a growing automotive parts segment (sensors, EV charging, motors), which LG Innotek has separately targeted for ₩2 trillion in annual revenue by 2030. The substrate business specifically is targeted to become a $700 million line by 2030, alongside a broader semiconductor-components ambition — FC-BGA, RF-SiP, and vehicle AP modules combined — of over $2.2 billion a year by the same date. Glass substrates, if they arrive on schedule, would sit inside that FC-BGA number as a premium tier rather than a separate revenue line.

One recent proof point: in July 2025, LG Innotek announced it had developed and begun mass-producing the world's first "Cu-Post" (copper post) technology, filing around 40 related patents. It's a denser, better-heat-dissipating way to wire a substrate, aimed at RF-SiP and FC-CSP products — a reminder that the company's core competitive edge is genuinely in fine-circuit substrate engineering, not just capacity.

LG Innotek sits in stage 2, substrate & core fabrication, alongside Absolics and Samsung Electro-Mechanics — the three companies actually trying to turn a sheet of glass into a working substrate core with copper-filled vias running through it.

Compared to its two stage-mates, LG Innotek's approach is the most conservative on paper: Absolics is racing for 2026 mass production and Samsung Electro-Mechanics is targeting 2026–2027, while LG Innotek's own pilot line points to 2027–2028. That gap isn't necessarily a weakness — it may just be a company with 40 years of substrate-yield discipline choosing not to rush a material it doesn't yet trust at scale.

The most interesting recent data point isn't a product announcement — it's a comment. At CES 2026, CEO Moon Hyuksoo told reporters that glass substrate development is "essentially done" on the technology side, but that the market itself may open later than the industry has been assuming: not 2028, but closer to 2030, because he doesn't yet see demand large enough to absorb the capacity a company would need to build. He also flagged a real structural risk that gets less airtime than the hype: glass substrates may not expand the total substrate market so much as cannibalize the high end of LG Innotek's own existing FC-BGA business, since glass is likely to replace only the priciest, most performance-constrained slice of what FC-BGA already does.

That's an unusually candid thing for an executive to say about his own company's headline growth bet, and it's worth taking seriously precisely because LG Innotek has more at stake in FC-BGA today than any of its glass-substrate peers. This isn't a company talking its book upward; if anything, it's talking its own most-hyped product line down.