Samsung Electro-Mechanics Plans $5 Billion Substrate Expansion with Customer Co-Investment to Lock in Long-Term Orders
nashnova research
Samsung Electro-Mechanics announced roughly $5 billion in FCBGA substrate capacity expansion, with customers putting up part of the capital and committing to multi-year purchase volumes — a rare deal structure signaling that downstream demand for advanced substrates is urgent enough to pay upfront for guaranteed supply.
Where does the $5 billion go?
The Sejong, South Korea project takes about ₩4.27 trillion, the largest single-product investment in Samsung Electro-Mechanics' history. Spending runs from September 2026 through May 2028; mass production starts September 2028.
The Vietnam subsidiary project takes about ₩2.51 trillion, targeting completion by June 2030.
This means → capacity arrives in two waves — 2028 for near-term relief, 2030 for a second layer of insurance. Samsung Electro-Mechanics is betting advanced substrate demand has at least four more years of upside.
Why are customers willing to co-fund?
Samsung Electro-Mechanics did not name the co-investing customers or their share of the capital. The commercial logic is straightforward: customers absorb part of the capex in exchange for priority access to future capacity.
In plain terms = it is like diners paying a deposit to reserve a table at a restaurant that has not opened yet — a sign that "money alone cannot secure supply" anxiety is already strong.
For Samsung Electro-Mechanics, demand visibility rises sharply while its own capital risk exposure falls — a deal where both sides are wagering that demand will persist.
How tight is existing capacity?
Samsung Electro-Mechanics' substrate lines ran at an average 89% utilization in H1 2026, up from 70% in 2025. LG Innotek's Gumi substrate line hit 94%.
TrendForce data: ABF substrate lead times have stretched to 48–56 weeks, versus roughly 12 weeks in equilibrium — the widest gap among the six infrastructure components the firm tracks.
This means → utilization is near full and lead times are four times normal. Existing capacity has almost no room to absorb incremental demand. This expansion is not a luxury — it is a necessity to prevent supply disruption.
What do Korea's substrate exports reveal?
In the first eight months of 2026, Korean PCB exports totaled $3.563 billion, up 14.4% year-on-year, while export weight rose only 1.8%.
This reflects a continuing shift toward higher-value substrates — the same weight fetches significantly more revenue, meaning unit prices are climbing fast.
The Korea Economic Daily projects full-year exports could challenge the 2022 record of $5.9 billion.
What are peers doing?
LG Innotek unveiled an FCBGA product for AI accelerators with a single-side dimension exceeding 100 mm — roughly six times the size of its PC-grade FCBGA samples. It plans 2027 mass production for high-value FCBGA and 2028 mass production for glass-core substrates.
LG Innotek's packaging-solutions unit posted H1 2026 revenue of ₩935.6 billion, up 18% year-on-year; operating profit surged 94%, lifting the margin from 6.5% to 10.6%.
Mid-cap supplier Daeduck Electronics is also scaling up: Q2 2026 revenue jumped 63.1% year-on-year to ₩401 billion, with an operating margin of 17.5%.
What is the biggest risk in this expansion cycle?
Two verification checkpoints matter most: whether customer co-funding commitments convert into actual orders when Sejong ramps in 2028, and whether next-generation substrate technologies such as glass core can take over smoothly around the same timeframe.
In plain terms = the money is already spent. The bet is that demand will still be there — and the technology will still be relevant — four years from now. If AI compute demand growth slows, or glass-core technology leapfrogs current approaches ahead of schedule, the return on this massive investment shrinks.
These two variables will determine the slope of the next upcycle for the advanced substrate supply chain.
市场有风险，内容仅供研究参考，不构成投资建议。