Samsung Electro-Mechanics joins Nvidia's substrate chain with a $5 billion capex plan and no advance named

Nvidia's GTC Berlin release quantified advances to IBIDEN, KYEC and Unimicron. Its new Korean substrate supplier carries a $5 billion capex plan of its own, and by 2028 one buyer is projected to take a significant portion of its substrate output.

Vincent JiangVincent Jiang · 3 min read
Share
Cutaway diagram of a multilayer ceramic capacitor showing its stacked ceramic dielectric and electrode layers
1 / 6Slide 1 of 6
The layers of an MLCC, the passive component whose lead times stretched from 30 to 52 weeks this autumn as SEMCO's lines filled with AI demand.

Berlin's ledger has one blank line

At Nvidia's+2.42% — Nvidia, up 2.42 percent today GTC Berlin event on October 21, a supply-chain release named six partners: packaging and test names Amkor+4.56% — Amkor, up 4.56 percent today, SPIL and KYEC, substrate makers IBIDEN and Unimicron, and Samsung Electro-Mechanics+1.13% — Samsung Electro-Mechanics, up 1.13 percent today (SEMCO), the Samsung affiliate that builds chip substrates and passive components 1.

The release attaches money to three of them: a multibillion-dollar capacity reservation advance to IBIDEN's Ono plant in Japan, a $1.1 billion advance to KYEC for new US test lines, and a capacity reservation agreement behind Unimicron's Taiwan expansion 12. For SEMCO, the newest member of the substrate chain, it offers two words: "joint investment" 13.

Nvidia wants substrate capacity up 2.5 times in three years 1. SEMCO will build dedicated lines for ABF substrates, the layered boards an AI processor sits on, at Sejong and Busan, ship from 2026, and by 2028 send Nvidia a "significant portion" of its total annual substrate output 13. Nvidia's commitments to suppliers, mostly memory and factory space, hit $279 billion on July 26, up from $119 billion a quarter earlier 4.

No advance, no slack, and the board has already voted

The supplier without a named advance has no spare capacity. Its multilayer ceramic capacitor (MLCC) lead times stretched from just over 30 weeks in September to 52 weeks in October, on Meritz Securities data cited by the Wall Street Journal; its components division ran at 91% utilization in the first half of 2026, and its Philippine expansion reaches mass production only after 2027 5. In March, chief executive Duck-Hyun Chang put substrate demand at more than 50% above capacity 6.

The money is mostly SEMCO's own. On September 28 its board approved 6.78 trillion won, about $5 billion, for substrate capacity at Sejong and in Vietnam, its largest bet on one product; the Sejong tranche alone is about 43.6% of shareholders' equity 7.

The filing says unnamed big-tech customers "will provide part of the capital" and guarantee volume, and mass production at Sejong begins in September 2028 7. Neither Berlin release names an advance 13. Seven long-term contracts since May, the latest a KRW360.5 billion memory-substrate deal for calendar 2027, already lock in KRW4.3 trillion of orders 8, up from the KRW3.6 trillion The Inference tallied on October 3.

Nvidia asks suppliers for up to 15 times more capacity; substrates 2.5 times

  • Estimate
0x5x10x15xSPIL optical engines, mid-202715xAmkor interposers, end-20284xSubstrates: IBIDEN, SEMCO, Unimicron, 3 years2.5xSEMCO's new lines feed this targetChip testing, 20272x
Data
Value
SPIL optical engines, mid-2027 (estimate)15x
Amkor interposers, end-2028 (estimate)4x
Substrates: IBIDEN, SEMCO, Unimicron, 3 years (estimate)2.5x
Chip testing, 2027 (estimate)2x
Stated capacity targets from Nvidia's October 21, 2026, release, as multiples of current capacity; supplier and Nvidia targets, not achieved output.1,4

The commitment cuts both ways

Nvidia operations chief Debora Shoquist says SEMCO "broadens our network of substrate partners and adds manufacturing capacity" 3, and the stock had more than quadrupled in 2026 before the September filing 7. The same scale binds: Nvidia's quarterly filing warns it "may be unable to reduce these commitments" if demand misses plan 4, and one buyer taking a significant share of one supplier's output concentrates them both.

Watch the split

Korean disclosure rules force SEMCO to publish any contract above 2.5% of revenue 8, so an Nvidia volume figure, or an advance, appears there if it exists; third-quarter results and the capex schedule will show who carries the Sejong lines. Three partners were paid in advance. The fourth is paid in orders.

Loading NVDA…

More about NVIDIA

NVDA · Fiscal Q2 2027
Fiscal Q2 2027 · quarter to 26 Jul 2026 · changes vs a year earlier

Revenue rose 106% to $96.2B, 92.5% of it from Data Center, “driven by the ramp of our Blackwell Ultra infrastructure”. Gains on equity stakes of $7.8B lifted net income to $59.7B.

Revenue$96.2B▲ 106%
Gross margin75.0%▲ 2.6 pts
Operating margin66.2%▲ 5.4 pts
Net income$59.7B▲ 126%
Hyperscale$48.7B▲ 102%AI clouds & enterprise$40.3B▲ 138%Data Center$89.0B▲ 117%Edge Computing$7.2B▲ 27%Revenue$96.2B▲ 106%Gross profit$72.1B75.0% marginCost of revenue$24.1BOther income$7.8Bequity-stake gainsOperating income$63.7B66.2% marginOperating expenses$8.4BNet income$59.7B▲ 126%Tax$11.8B16.5% rateR&D$7.1BSG&A$1.4B
Show as a table
LineValue
Revenue$96.2B
Gross margin75.0%
Operating margin66.2%
Net income$59.7B
Hyperscale$48.7B
AI clouds & enterprise$40.3B
Data Center$89.0B
Edge Computing$7.2B
Gross profit$72.1B
Cost of revenue$24.1B
Other income$7.8B
Operating income$63.7B
Operating expenses$8.4B
Tax$11.8B
R&D$7.1B
SG&A$1.4B

The daily newsletter

Subscribe to get your daily newsletter on the latest in AI and tech and dive deeper into each topic that interests you.

Reader comments

0 comments

    Get the newsletter

    The day's AI and market news, in your inbox every weekday.