Samsung and SK Hynix Are 90.8% of Korean Semis: Where ETF Flow Is More Sensitive

Sat, 13 Jun 2026 14:40:00 +0900

Context This note follows Korea theme ETF rebalance flow, ETF flow is leading the Korean market, the Korea Semiconductor / HBM Hub, the AI PCB and Substrate Hub, the Sam-Ha-Ma parity follow-up, and AI data-center capex bottlenecks.

TL;DR

Korea’s semiconductor value chain is extremely concentrated by market cap. As of June 12, 2026, Samsung Electronics and SK Hynix represented about 90.8% of the top-50 Korean semiconductor market-cap proxy basket.
In absolute ETF ownership, SK Hynix, Samsung Electronics, Samsung Electro-Mechanics and Hanmi Semiconductor dominate. But on market-cap-adjusted ETF sensitivity, Leeno Industrial, DB HiTek, EO Technics, Wonik IPS, HPSP and ISC matter more.
On an ETF under-owned gap framework, the first watchlist is TCK, Daeduck Electronics, Korea Circuit and Doosan Tesna. Simmtech, TES and TLB already have meaningful ETF exposure.
Strategy conclusion: separate megacap memory as core exposure, ETF-sensitive leaders as event/flow trades, and under-owned ETF gap candidates as the next ETF product / index-expansion watchlist.
This is not a buy list. It is a research-priority framework. Before using ETF flow as a trade signal, investors still need issuer PCF/PDF confirmation, index methodology, closing-auction execution, and T+1 / T+3 relative-strength evidence.

Core Point

Korean semiconductor ETF flow is not about which company is largest. It is about which stocks are most sensitive to ETF flow and which stocks ETFs have not yet owned enough. Samsung and SK Hynix are the market itself; alpha is more likely to come from sensitivity and under-owned gaps.

1. The Question: Two Stocks Dominate the Sector

The Thesis OS Alpha note defined Korean semiconductor exposure broadly: devices, memory, fabless, OSAT, equipment, materials, components, substrates and packaging. It then ranked the top 50 listed Korean names by a market-cap proxy.

The proxy is not official KRX market capitalization. It uses the local database formula:

market-cap proxy = close price × foreign-owned shares ÷ foreign ownership ratio
estimated ETF ownership = ETF NAV × constituent weight
ETF exposure / market cap = estimated ETF ownership ÷ market-cap proxy

This is not a perfect replacement for official KRX market cap. But when the same method is applied consistently across the top 50, it is useful for comparing where capital is concentrated.

Item	Value
Top-50 semiconductor market-cap proxy	about KRW 3,766tn
Samsung Electronics market-cap proxy	KRW 1,885.58tn
SK Hynix market-cap proxy	KRW 1,532.41tn
Samsung + SK Hynix share of top-50 basket	about 90.8%

Korean semiconductors look broad by stock count, but by market cap they are effectively a Samsung Electronics / SK Hynix market. That creates two separate questions:

Does the biggest ETF money still end up in Samsung and SK Hynix?
Inside the remaining 9.2%, which stocks are most exposed to ETF-flow shocks?

This post focuses on the second question.

2. Top 50 Korean Semiconductor Market-Cap Proxy

Unit: KRW tn. Market cap is a local database proxy, not the official KRX figure.

Rank	Company	Code	Segment	Market-cap proxy
1	Samsung Electronics	005930	Devices / memory	1,885.58
2	SK Hynix	000660	Devices / memory	1,532.41
3	Samsung Electro-Mechanics	009150	Substrate / packaging	128.02
4	Hanmi Semiconductor	042700	Equipment / test	34.42
5	LG Innotek	011070	Substrate / component	24.52
6	Jusung Engineering	036930	Equipment	10.73
7	Wonik IPS	240810	Equipment	9.00
8	Isu Petasys	007660	Substrate	8.99
9	Daeduck Electronics	353200	Substrate	8.17
10	Leeno Industrial	058470	Test socket	7.97
11	DB HiTek	000990	Foundry	7.58
12	EO Technics	039030	Equipment	7.54
13	SKC	011790	Materials / composite	6.96
14	HPSP	403870	Equipment	5.88
15	FADU	440110	Fabless	5.06
16	PSK	319660	Equipment	4.92
17	ISC	095340	Test socket	4.88
18	Simmtech	222800	Substrate	4.87
19	Eugene Technology	084370	Equipment	4.49
20	Jeju Semiconductor	080220	Fabless	3.53
21	TES	095610	Equipment	3.53
22	Soulbrain	357780	Materials	3.43
23	Doosan Tesna	131970	OSAT / test	3.38
24	PSK Holdings	031980	Equipment	3.29
25	Hana Micron	067310	OSAT	3.25
26	TCK	064760	Materials / components	3.24
27	Korea Circuit	007810	Substrate	3.22
28	Hansol Chemical	014680	Materials	3.21
29	Dongjin Semichem	005290	Materials	3.01
30	TSE	131290	Test	2.99
31	Techwing	089030	Test equipment	2.42
32	VM	089970	Equipment	2.40
33	Park Systems	140860	Metrology	2.38
34	RFHIC	218410	RF / compound-semi adjacent	2.04
35	GigaVis	420770	Inspection equipment	1.85
36	KC Tech	281820	Equipment	1.72
37	Hana Materials	166090	Materials / components	1.45
38	S&S Tech	101490	Blank mask	1.36
39	Haesung DS	195870	Packaging / lead frame	1.33
40	YC	232140	Inspection equipment	1.31
41	Komico	183300	Cleaning / coating	1.31
42	GST	083450	Equipment	1.20
43	SFA Semicon	036540	OSAT	1.19
44	Wonik QnC	074600	Quartz / materials	1.11
45	TLB	356860	Substrate	0.98
46	SEMCNS	252990	Probe-card component	0.98
47	LX Semicon	108320	Fabless	0.82
48	DI	003160	Inspection equipment	0.80
49	Protec	053610	Equipment	0.80
50	Dongwoon Anatech	094170	Fabless	0.78

The key point is the gap after the third name. Samsung Electro-Mechanics is already a large independent name, but Hanmi is materially smaller, and Leeno is much smaller still. The same ETF flow therefore has a larger price-impact potential in mid-cap equipment, substrate, test and materials names.

3. Absolute ETF Exposure: Big Money Is Already in the Leaders

Estimated ETF ownership is calculated as ETF constituent weight multiplied by ETF NAV and summed across the local ETF database.

Rank	Company	ETF count	Weight sum	Max ETF weight	Estimated ETF ownership	ETF / market cap
1	SK Hynix	9	209.8%	37.6%	KRW 8,249.9bn	0.54%
2	Samsung Electronics	9	174.8%	26.2%	KRW 6,205.9bn	0.33%
3	Samsung Electro-Mechanics	4	76.2%	37.9%	KRW 3,037.2bn	2.37%
4	Hanmi Semiconductor	5	53.9%	18.1%	KRW 2,134.6bn	6.20%
5	Leeno Industrial	3	14.2%	7.3%	KRW 772.3bn	9.70%
6	DB HiTek	4	11.5%	4.7%	KRW 699.7bn	9.23%
7	Wonik IPS	2	11.1%	6.7%	KRW 636.4bn	7.07%
8	EO Technics	2	10.4%	6.3%	KRW 596.5bn	7.91%
9	LG Innotek	1	8.9%	8.9%	KRW 524.8bn	2.14%
10	Isu Petasys	3	15.8%	9.1%	KRW 447.1bn	4.98%

In absolute terms, SK Hynix and Samsung dominate. But ETF ownership is only 0.54% and 0.33% of their respective market-cap proxies. For these megacaps, ETF flow is only one part of a much larger flow stack: foreign cash equities, domestic institutions, retail leverage, futures/program flow and earnings expectations.

By contrast, Hanmi’s estimated ETF ownership is KRW 2.13tn, or 6.20% of its market-cap proxy. Leeno, DB HiTek, Wonik IPS and EO Technics are smaller in absolute ETF ownership, but much more sensitive relative to market cap.

4. ETF-Flow Sensitivity: Mid-Cap Semis Move More

Rank	Company	Estimated ETF ownership	ETF / market cap
1	Leeno Industrial	KRW 772.3bn	9.70%
2	DB HiTek	KRW 699.7bn	9.23%
3	EO Technics	KRW 596.5bn	7.91%
4	Wonik IPS	KRW 636.4bn	7.07%
5	HPSP	KRW 411.2bn	6.99%
6	ISC	KRW 306.5bn	6.29%
7	Hanmi Semiconductor	KRW 2,134.6bn	6.20%
8	Soulbrain	KRW 182.5bn	5.31%
9	Isu Petasys	KRW 447.1bn	4.98%
10	Samsung Electro-Mechanics	KRW 3,037.2bn	2.37%

This group has already received meaningful ETF flow. The implication is two-sided.

First, if semiconductor ETF creation continues or thematic ETFs expand, these names can keep receiving incremental flow.

Second, if ETF redemptions or cap rebalances appear, these same names can become mechanically vulnerable. High ETF exposure is not only a positive. It means higher upside participation when flows are supportive and higher mechanical selling risk when flows reverse.

5. Daeduck and Simmtech: Local Theme Databases Can Understate Exposure

The analysis rechecked Daeduck Electronics and Simmtech across the full Naver ETF API universe of 1,137 ETFs because a local theme ETF database alone can understate their exposure.

Company	ETF count	Managers	Weight sum	Max ETF weight	Estimated ETF ownership	ETF / market cap
Daeduck Electronics	16	10	81.78%	15.39%	KRW 159.0bn	1.95%
Simmtech	17	11	61.25%	6.91%	KRW 188.5bn	3.87%

Daeduck is already in 16 ETFs. Simmtech is already in 17. So neither is an undiscovered ETF name.

But they differ:

Daeduck has many ETF inclusions, but only 1.95% ETF ownership relative to market cap.
Simmtech is already at 3.87%, meaning ETF flow is more reflected.
Daeduck could benefit if AI substrates, FC-BGA, ASIC and networking themes become more explicit in ETF product names and index methodology.
Simmtech is already broadly represented across SOL AI Semiconductor Materials/Equipment, ACE AI Semiconductor TOP3+, RISE Non-Memory Semiconductor Active, KoAct KOSDAQ Active and KODEX AI Semiconductor Core Equipment.

Daeduck Electronics: Main ETF Exposures

ETF	Manager	Weight	ETF NAV	Estimated holding
KODEX AI Semiconductor Core Equipment	Samsung AM	6.84%	KRW 473.7bn	KRW 32.4bn
ACE AI Semiconductor TOP3+	KIM	2.23%	KRW 1,296.9bn	KRW 28.9bn
RISE Non-Memory Semiconductor Active	KB AM	5.76%	KRW 345.3bn	KRW 19.9bn
WON Semiconductor Value Chain Active	Woori AM	4.36%	KRW 378.5bn	KRW 16.5bn
1Q K Semiconductor TOP2+	Hana AM	6.06%	KRW 253.5bn	KRW 15.4bn
SOL Semiconductor Back-End	Shinhan AM	15.39%	KRW 83.4bn	KRW 12.8bn

Simmtech: Main ETF Exposures

ETF	Manager	Weight	ETF NAV	Estimated holding
SOL AI Semiconductor Materials/Equipment	Shinhan AM	4.79%	KRW 1,221.7bn	KRW 58.5bn
ACE AI Semiconductor TOP3+	KIM	2.32%	KRW 1,296.9bn	KRW 30.1bn
RISE Non-Memory Semiconductor Active	KB AM	6.91%	KRW 345.3bn	KRW 23.9bn
KoAct KOSDAQ Active	Samsung Active AM	3.55%	KRW 652.8bn	KRW 23.2bn
KODEX AI Semiconductor Core Equipment	Samsung AM	4.77%	KRW 473.7bn	KRW 22.6bn
TIME KOSDAQ Active	Timefolio	2.62%	KRW 383.5bn	KRW 10.0bn

This changes the conclusion. Simmtech is not an ETF under-owned gap. It is closer to an already-ETF-owned substrate name. Daeduck is different: it is widely included but still underweight relative to market cap.

6. ETF Under-Owned Gap Reclassification

The focused gap screen covered TLB, Korea Circuit, Doosan Tesna, TES, TCK, Daeduck and Simmtech.

Company	Market-cap proxy	ETF count	Estimated ETF ownership	ETF / market cap	Recent active ETF change	Interpretation
TCK	KRW 3.24tn	5	KRW 14.4bn	0.44%	+0.39%p	Best ETF under-owned gap
Daeduck Electronics	KRW 8.17tn	16	KRW 159.0bn	1.95%	+0.09%p	AI substrate ETF expansion candidate
Doosan Tesna	KRW 3.38tn	6	KRW 53.6bn	1.58%	-1.27%p	Needs OSAT/test ETF trigger
Korea Circuit	KRW 3.22tn	5	KRW 73.2bn	2.27%	+0.59%p	Quiet weight-increase candidate
Simmtech	KRW 4.87tn	17	KRW 188.5bn	3.87%	+2.00%p	ETF flow already meaningful
TES	KRW 3.53tn	14	KRW 126.6bn	3.59%	-0.63%p	Many ETF inclusions, some recent trimming
TLB	KRW 0.98tn	7	KRW 34.2bn	3.47%	-0.60%p	High SOCAMM purity, but ETF exposure already exists

As a strategy framework, this becomes a four-bucket funnel:

Bucket	Names	Interpretation
Advance to deeper work	TCK, Daeduck Electronics, Korea Circuit	ETF under-owned gap plus possible theme expansion
Watch / trigger needed	Doosan Tesna	Under-owned, but recent active ETF change is weak
ETF already reflected	Simmtech, TES, TLB	Already widely included and ETF-sensitive
Flow-sensitive leaders	Leeno, DB HiTek, EO Technics, Wonik IPS, HPSP, ISC, Hanmi	Already exposed to ETF flow; event / flow-trade profile

7. Idea Card: TCK

Actionability: Advance to deeper work. This is not an immediate buy signal; it needs ETF product expansion and flow confirmation.

Variant Wedge: TCK has a KRW 3.24tn market-cap proxy, but only KRW 14.4bn of estimated ETF ownership, or 0.44% of market cap. That is very low for a representative semiconductor materials / consumables name.

Why Now: If semiconductor ETFs expand from memory and equipment into materials / consumables, TCK is one of the cleanest under-owned candidates. Recent active ETF change is also positive at +0.39%p.

First Rejection: The gap can persist if materials / consumables ETFs do not scale, or if TCK fails liquidity, methodology or theme-label screens.

What Would Make It Investable: Issuer PCF/PDF evidence of inclusion or weight increase, plus confirming foreign / institutional flow.

What Would Kill It: If operating momentum or margins weaken while the ETF gap remains unfilled, the “ETF will eventually buy it” argument is not enough.

Next Workflow: Map the semiconductor materials / consumables ETF universe, index methodology, liquidity and free-float constraints.

8. Idea Card: Daeduck Electronics

Actionability: Advance to deeper work. AI substrate ETF expansion candidate.

Variant Wedge: Daeduck is already included in 16 ETFs, but estimated ETF ownership is only 1.95% of its market-cap proxy. The stock is included widely, but not heavily owned.

Why Now: If AI substrates, FC-BGA, ASIC and networking become more explicit in ETF product labels and index construction, Daeduck’s weight can rise. It is much smaller than Samsung Electro-Mechanics and less ETF-reflected than Simmtech.

First Rejection: Do not overstate the “undiscovered” angle. It is already present across ten managers and 16 ETFs.

What Would Make It Investable: Higher weight in AI substrate / FC-BGA indices, relative strength above the 20-day line, and confirming foreign / institutional flow.

What Would Kill It: If AI substrate value capture remains concentrated in Samsung Electro-Mechanics and Daeduck does not show FC-BGA / high-speed substrate earnings contribution.

Next Workflow: Compare Daeduck, Simmtech, Korea Circuit and TLB by ETF weight, relative strength and estimate revisions.

9. Idea Card: Korea Circuit

Actionability: Advance to deeper work / watchlist.

Variant Wedge: Korea Circuit has a KRW 3.22tn market-cap proxy, KRW 73.2bn estimated ETF ownership and 2.27% ETF / market cap. It is not untouched, but it is still less reflected than Simmtech, TES and TLB.

Why Now: It has a possible SOCAMM + FC-BGA dual catalyst, and recent active ETF change is +0.59%p. The key phrase is “quiet weight increase.”

First Rejection: Direct theme exposure can look less obvious than Simmtech or TLB, and ETF flow alone is not enough without AI substrate / module revenue proof.

What Would Make It Investable: Continued active ETF weight increase and matching real-money institutional or foreign flow.

What Would Kill It: If SOCAMM / FC-BGA remains a price narrative without revenue, margin or order evidence.

Next Workflow: Track Korea Circuit ETF inclusions, active ETF changes and relative flow versus Simmtech and TLB.

10. Idea Card: Doosan Tesna

Actionability: Watch / trigger needed.

Variant Wedge: Doosan Tesna has a KRW 3.38tn market-cap proxy and only KRW 53.6bn estimated ETF ownership, or 1.58% of market cap. For an OSAT / test candidate, ETF representation is still low.

Why Now: If AI back-end and test demand extends from substrates and sockets toward OSAT / test, Doosan Tesna can become a next ETF inclusion / weight candidate.

First Rejection: Recent active ETF change is -1.27%p, which is weak. Under-owned does not automatically mean actionable.

What Would Make It Investable: OSAT / test ETF weight recovery, consensus upgrades, and real-money institutional re-entry.

What Would Kill It: If test demand is priced mainly into Leeno, ISC and TSE while Doosan Tesna does not show margin or utilization leverage.

Next Workflow: Compare Doosan Tesna, Hana Micron, SFA Semicon, Leeno and ISC across ETF ownership, operating leverage and flow.

11. How to Treat Names That Already Have ETF Flow

Leeno, DB HiTek, EO Technics, Wonik IPS, HPSP, ISC and Hanmi are not under-owned gap names. They are ETF-flow-sensitive leaders.

Condition	Action
ETF creation plus foreign / institutional buying	Trend-following can work
Strong rebalance signal but one-day price spike and volume fade	Do not chase; confirm T+1 / T+3
ETF redemption or cap trim plus foreign selling	Short-term risk rises
Price holds despite ETF pressure	Confirms stronger real demand

Hanmi is especially important. Estimated ETF ownership is large, but the real stock thesis is not ETF flow. It is HBM TC bonders, HBM4 customer diversification and actual order announcements. ETF flow can move the price; it is not the thesis itself.

12. Execution Framework: Three Baskets

12.1 Core Semiconductor Exposure

Samsung Electronics and SK Hynix

For these two, Korea market flow, foreign cash equities, HBM earnings durability and relative PER versus Micron matter more than ETF sensitivity. Do not treat ETF cap-trim screens as sell signals. But because the sector is so concentrated, investors should monitor total portfolio beta when volatility rises.

12.2 ETF Flow-Sensitive Leaders

Leeno, DB HiTek, EO Technics, Wonik IPS, HPSP, ISC and Hanmi

These names already have ETF flow. They can move quickly when the event is right, but they can also be more sensitive to ETF redemptions and cap rebalances. Timing and confirmation matter more than the simple label.

12.3 ETF Under-Owned Gap Candidates

TCK, Daeduck Electronics, Korea Circuit and Doosan Tesna

These are candidates for the next ETF product label or index-expansion wave. But “ETFs do not own enough” is not sufficient. The gap must be paired with operating evidence.

Company	Evidence Needed
TCK	Materials / consumables demand, customer utilization, margin recovery
Daeduck Electronics	AI substrate, FC-BGA, ASIC / networking revenue contribution
Korea Circuit	SOCAMM + FC-BGA repeat orders and profitability
Doosan Tesna	OSAT / test volume growth and operating leverage

13. Failure Conditions

Naver ETF surface data differs materially from issuer PCF/PDF constituent data.
Under-owned candidates do not meet the actual index methodology or liquidity criteria for inclusion / weight increase.
ETF AUM does not grow, and thematic ETFs see redemptions.
Rebalance signals were already priced, and T+1 / T+3 relative strength breaks.
Under-owned gap candidates fail to deliver operating evidence.
Samsung and SK Hynix flow rolls over, causing broad semiconductor ETF redemptions.

The most important failure condition is number three. ETF rebalancing is internal weight change, not the same as ETF creation. If money leaves the ETF, the internal buy candidates may underperform expectations.

14. Final View

The easiest mistake in Korean semiconductor ETF analysis is to look only at the largest names. SK Hynix and Samsung have the largest absolute ETF holdings, but their market-cap-adjusted sensitivity is low. Leeno, DB HiTek, EO Technics, Wonik IPS, HPSP and ISC are more sensitive to ETF flow.

But high sensitivity also means higher redemption risk. The next alpha is not simply in the most ETF-sensitive names. It is in places where ETF under-ownership, operating theme and flow turn overlap.

Current priority:

Rank	Candidate	View
1	TCK	Strongest ETF under-owned gap; watch materials / consumables ETF expansion
2	Daeduck Electronics	AI substrate ETF expansion candidate; widely included but still low market-cap-adjusted ownership
3	Korea Circuit	Quiet weight-increase candidate; confirm SOCAMM + FC-BGA dual catalyst
4	Doosan Tesna	OSAT / test under-owned candidate; needs trigger after weak recent active ETF change
5	Leeno / DB HiTek / EO Technics / Wonik IPS / HPSP / ISC	ETF-sensitive leaders; better framed as event / flow trades than under-owned gaps

One-line conclusion:

In a market where Samsung Electronics and SK Hynix are 90.8% of the semiconductor market-cap proxy, ETF alpha comes from separating market-cap-adjusted ETF sensitivity from ETF under-owned gaps.

Evidence Ledger

Item	Detail
Source research note	`Thesis OS source research note`
Price / market-cap proxy	`full_results_20260613_061000.csv`
Price as-of date	2026-06-12
Local theme ETF DB	`theme_etf_rebalance_flow.db`
Active ETF flow DB	`active_etf_flow.db`
Full ETF constituent API	`https://m.stock.naver.com/api/stocks/etf`, `https://m.stock.naver.com/api/etf/{ETF_CODE}/basic`
Full ETF scan	1,137 ETFs scanned, 1,137 successes, 0 failures
KRX direct API	`data.krx.co.kr` JSON/PDF endpoints returned `LOGOUT`
Raw support artifacts	`/tmp/daeduck_simmtech_all_naver_etf_20260612.csv`, `/tmp/semicap_gap_targets_all_naver_etf_20260612.csv`

Fact / Inference / Speculation / Blocked

[Fact]

The top-50 Korean semiconductor market-cap proxy basket totals about KRW 3,766tn.
Samsung Electronics and SK Hynix represent about 90.8% of that basket.
Based on the full Naver ETF API scan, Daeduck Electronics is included in 16 ETFs and Simmtech in 17 ETFs.
Estimated ETF ownership is KRW 159.0bn for Daeduck and KRW 188.5bn for Simmtech.
Among the seven focused gap candidates, TCK has the lowest ETF ownership ratio: KRW 14.4bn, or 0.44% of market cap.

[Inference]

In absolute ETF ownership, SK Hynix, Samsung Electronics, Samsung Electro-Mechanics and Hanmi are the largest beneficiaries.
In rebalance sensitivity, Leeno, DB HiTek, EO Technics, Wonik IPS, HPSP and ISC matter more.
Daeduck and Simmtech can be understated in local theme ETF data; the full ETF scan shows both are already widely included.
From an ETF under-owned gap perspective, TCK, Daeduck, Doosan Tesna and Korea Circuit are more interesting than Simmtech.

[Speculation]

If FC-BGA, ASIC and AI networking themes become more explicit in ETF labels and index methodology, Daeduck’s weight can rise.
If SOCAMM / LPDDR themes persist, TLB and Korea Circuit can receive incremental inclusion / weight expansion after Simmtech.
If semiconductor materials / consumables ETFs expand, TCK is the most direct under-owned candidate.

[Blocked]

KRX direct ETF PDF / JSON endpoints returned LOGOUT.
Market cap is a local proxy, not official KRX market cap.
Issuer PCF/PDF cross-checks have not been completed for every ETF. Key candidates need individual verification.

Leeno Industrial on Korea Invest Insights