EWY Top Holdings: You Didn't Buy 'Korea' — You Bought Two Memory Stocks

This article is for informational purposes only and is not financial advice. TheGatBull may earn a commission from some links at no cost to you — see our disclosure and full disclaimer.

Not financial advice — a view from Seoul, not a recommendation. Figures verified as of July 3, 2026; markets move fast.

On July 2, SK Hynix (KRX: 000660) fell 14.6%, Samsung Electronics (KRX: 005930) 9.1%, and the KOSPI closed at 7,648.09 — down 7.9% in one day, with a sell-side sidecar triggered seven minutes after the open. The reason is simple: two stocks are more than half the index.
The top two holdings of the iShares MSCI South Korea ETF (EWY) — SK Hynix at ~27.7% and Samsung Electronics at ~23.8% — are roughly 51.5% of the fund (late June 2026). The KOSPI itself is more extreme: the duo is 57.1% of the index, or 59.7% counting Samsung’s preferred shares (Korea Exchange, June 25).
For scale: Nvidia plus Apple are about 13% of the S&P 500. Kioxia plus Toyota are about 7.4% of Japan’s TOPIX. Korea runs at three to four times the concentration Americans call the “Magnificent 7 problem.”
Goldman Sachs warns that every additional 1 percentage point of combined weight could force ~$2 billion of selling by foreign institutions bound by US 40 Act diversification rules. Going up is itself a sell trigger.
The honest question isn’t “will it crash?” — it’s whether you knew what you owned. If you bought EWY for “diversified Korea exposure,” that one word — diversified — is the thing to retire today.

The short answer

If you searched “EWY top holdings,” the answer isn’t a list — it’s one sentence: you effectively bought two stocks. SK Hynix and Samsung Electronics make up about 51.5% of the fund, and in the actual KOSPI outside the ETF wrapper, the duo is 57–60% of total market value. For the S&P 500 to feel like this, Nvidia and Apple would have to be four times their current size. So when a Nasdaq chip selloff hit Seoul on July 2 and the KOSPI dropped 7.9% in a single session, that wasn’t an anomaly. It was the structure doing exactly what the structure is built to do. Here’s what happened, why Korea got this way, and what your actual choices are.

What actually happened on July 2

The chain was short. Reports that Meta was considering selling access to its own AI computing infrastructure — a signal it may have built more capacity than it needs — reignited “AI overbuild” fears and knocked US chip stocks overnight. The next morning in Seoul:

SK Hynix closed down 14.6% at ₩2,187,000 (~$1,570), Samsung Electronics down 9.1% at ₩286,000 (~$206).
The KOSPI was down more than 6% within minutes of the open — a sell-side sidecar (a temporary halt on program sell orders) triggered at 9:07 a.m., seven minutes in — and closed at 7,648.09, down 7.9%.
Whatever the hundreds of other listed companies did that day barely mattered. Two stocks had already decided the index.

And this wasn’t a first. On June 23 the KOSPI closed down 9.99% from near record highs on a chip scare. Korea’s circuit breaker — the full market-wide halt — has fired five times in 2026, out of just eleven activations since the mechanism was introduced in 1998. The lighter sidecar mechanism has fired 29 times this year, a record. Nearly half of all circuit-breaker activations in Korean market history have happened in the past six months. That statistic alone tells you something structural is going on — and the next section is the part your brokerage app won’t show you.

The part your ETF screen doesn’t show

Open EWY in a US brokerage app and you’ll see “78 holdings” and a $24 billion fund. Reassuring. Here’s what the screen leaves out: even EWY’s ~51.5% is a softened number. The fund tracks the MSCI Korea 25/50 Index, which caps any single stock at 25% — not for risk reasons, but because US tax law (the RIC diversification rules) requires it. The real Korean market has no such cap.

And there’s a second layer: the cap only applies on rebalancing dates. Between rebalances, weights drift with prices — which is why SK Hynix currently sits at ~27.7% of EWY, above its own cap. In other words: the ETF’s “diversification” is makeup applied by regulation, and lately the makeup can’t keep up with the face. The bare face is in Seoul, where the duo approaches 60% of the index. How did one of the world’s major markets end up looking like this? Three forces, and at least one of them is uniquely Korean.

Why Korea got here: three forces

1. The AI memory supercycle, undiluted. HBM — the high-bandwidth memory inside AI accelerators — sent both companies’ earnings and share prices vertical. Their combined KOSPI weight was 34.0% at the end of 2025; it crossed 50% for the first time ever on May 27, 2026, and reached 57–60% by late June (Korea Exchange). On June 22, SK Hynix overtook Samsung Electronics as Korea’s most valuable company — the first dethroning in roughly a quarter century. The #3 stock in the entire index, SK Square (KRX: 402340), is a distant 3.4% (Korea Exchange, June 25).

2. The daejangju culture. Korean retail has a word — daejangju (the “flag-carrier” stock, the designated leader of a sector) — for its habit of concentrating flows into one champion per theme. It isn’t slang; it’s an observable liquidity pattern. And in May 2026 it got a power tool: Korea’s first single-stock leveraged ETFs on Samsung and SK Hynix listed on May 27. Since then, intraday volatility and ETF tracking gaps have widened noticeably, and brokerage analysts note retail money rotating out of index-level products into the single-stock leveraged versions — which drains liquidity from everything else in the index.

3. A regulatory feedback loop. This is the Goldman Sachs warning. Under the US Investment Company Act’s diversification rules, many foreign institutions cannot let single positions breach set limits. Goldman’s analysts (Timothy Moe and John Kwon) estimate that each additional 1 percentage point of combined Samsung–Hynix weight could force roughly $2 billion of mechanical selling, with up to ~$20 billion at risk in total. Sit with that: the more the two stocks rise, the more some holders are forced to sell them. The loop is already switched on.

Both sides of the concentration trade

The bull reading: concentration is what a bull market looks like. Investors worried the same way when the Magnificent 7 carried the S&P — and earnings justified the weight. As long as HBM profit growth at Samsung and SK Hynix is real, concentration isn’t risk; it’s efficiency. You’re getting the purest possible exposure to the defining tech cycle of the decade.

The bear reading: even if the earnings are real, the structure is fragile. Leveraged ETFs, heavy options activity, and retail margin debt amplify every downswing — Goldman’s own note flags exactly this, and 2026’s five circuit breakers and 29 sidecars are the receipts. Worse, the two stocks share one cycle — memory pricing. This isn’t two bets; it’s one factor wearing two tickers. An index with near-zero internal diversification.

Notice what both sides agree on: whichever way it resolves, this is no longer a bet on “the Korean economy.” It’s a bet on the AI memory cycle.

🎩 Under the Gat — You think you bought an index. From Seoul, what you bought is the DRAM cycle. The hundreds of other listed companies get a voice only after the two giants have finished moving. I’m not saying that’s bad — if you bought it on purpose, it’s a wonderfully pure bet. The problem is buying it while believing the word “diversified.” Delete that one word, and your decisions actually get easier.

How Korea’s concentration compares to the world

	KOSPI (Korea)	S&P 500 (US)	TOPIX (Japan)	EWY (the ETF)
Top two stocks	Samsung Electronics + SK Hynix	Nvidia + Apple	Kioxia + Toyota	SK Hynix + Samsung Electronics
Combined weight	57.1–59.7%	~13.1%	~7.4%	~51.5%
Same industry?	Yes — both memory. Effectively a single factor	No (GPUs vs consumer hardware)	No (memory vs autos)	Yes — both memory
Cap or buffer	None — raw market cap	None	None	25/50 cap, applied only at rebalancing

Not a perfect parallel — the US and Japanese top-two pairs sit in different industries, so the quality of their concentration differs. Korea’s top two ride the same memory cycle, which is the decisive difference. Note that Japan’s #1 stock is now also a memory chipmaker (Kioxia overtook Toyota this year) — yet Japan’s concentration is still about one-eighth of Korea’s. All figures late June 2026.

🎩 Under the Gat — If you’ve read US articles fretting that “the Magnificent 7 are 30% of the index,” the KOSPI is that worry at three to four times the strength — with the added condition that the top two stocks are the same trade. And yet, here’s the interesting part: whether this concentration terrifies you or tempts you, the Korean market has a door for each of you. That’s the next section.

So what are your actual options?

Three doors, framed as questions rather than advice:

① Own it knowingly. Redefine EWY in your head: not “Korea exposure” but a high-purity AI memory bet. Then the dashboard you watch isn’t Korean GDP — it’s HBM margins and the DRAM cycle.
② Pick the Korea outside the memory duo. Beyond the two giants sits a “different Korea” the index shadow hides — the AI data center power ecosystem (transformers, cables, nuclear, cooling), defense, shipbuilding. We map that terrain in Korea’s AI data center power supply chain and the Value-Up one-year checkup.
③ Judge the giants as stocks, not as an index. If you’d rather size the two names directly than own them diluted, start with why the flows around them are so strange right now: foreigners are selling Korea while buying the chips. (SK Hynix is even preparing a US listing, per Barron’s — the duo is coming to your home market either way.)

Whichever door: the starting point is the same. Decide whether the exposure you actually want is “Korea” or “memory.” Everything else follows from that one call. — a view, not advice.

Sources

CNBC / Korea JoongAng Daily / Seoul Economic Daily — July 2, 2026 session: KOSPI 7,648.09 (−7.89%), SK Hynix −14.57% (₩2,187,000), Samsung Electronics −9.06% (₩286,000), sell-side sidecar at 9:07 a.m.; Meta AI-compute reports as trigger
Korea Exchange via The Herald Business (June 26, 2026) — Samsung 28.67% + SK Hynix 28.44% = 57.11% of KOSPI (June 25), 59.69% including preferred shares; 34.04% at end-2025; first 50% cross on May 27; single-stock leveraged ETF listing and volatility effects (NH Investment & Securities)
iShares (BlackRock), EWY fund page — 78 holdings, ~$24.1B net assets, MSCI Korea 25/50 benchmark, IT sector 57.0% (late June 2026); holdings detail via Finnhub/StockAnalysis as of June 26: SK Hynix 27.69%, Samsung Electronics 23.82%, top 10 = 67.0%
Goldman Sachs Research (Timothy Moe, John Kwon) via press reports (July 1, 2026) — ~$2B of forced foreign selling per +1%p of combined weight under 40 Act diversification rules; ~$20B total at risk; leveraged ETF / options / margin fragility
The Asia Business Daily (June 26, 2026) — sidecar triggered 29 times in 2026 (15 buy / 14 sell), record; circuit breaker 5 times in 2026 of 11 total since 1998
Korea Exchange / The Herald Business — June 23 close −9.99%; March 4 −12.1%
The Korea Economic Daily / KED Global (June 22, 2026) — SK Hynix overtakes Samsung Electronics as Korea’s largest company by market cap, first time in about a quarter century
Slickcharts — Nasdaq-100 weights (Nvidia ~8.7%, Apple ~7.1%); S&P 500 weights via KRX comparison reporting (Nvidia 7.06%, Apple 6.04%, June 25)
MSCI — MSCI 25/50 index methodology (RIC concentration constraints applied at rebalancing)
Barron’s (June 26, 2026) — SK Hynix US listing plans

(₩ converted at roughly ₩1,390 per US dollar, July 2026.)

Disclaimer: TheGatBull provides information and commentary for educational purposes only. This is not financial advice, not a recommendation to buy or sell any security or fund, and not a price target. Index and ETF weights change daily — the figures here are verified as of July 3, 2026 and should be re-checked against primary sources (iShares, Korea Exchange) as of your trade date. Do your own research and consult a licensed advisor before making investment decisions.

Frequently Asked Questions

Is EWY a diversified ETF?

By holding count it looks like one — 78 stocks. By weight, no: the top two holdings, SK Hynix and Samsung Electronics, are roughly 51.5% of the fund, and the top ten are about 67% (as of late June 2026). On top of that, the top two are in the same industry — memory semiconductors — so they tend to move together.

Why is the KOSPI (57–60%) more concentrated than EWY (~51.5%)?

EWY tracks the MSCI Korea 25/50 Index, which caps any single stock at 25% to satisfy US tax (RIC) diversification rules. But the cap only applies at rebalancing dates — SK Hynix has already drifted to about 27.7% between rebalances. The uncapped, real-world KOSPI concentration is closer to 60%.

Why did the KOSPI fall so hard on July 2, 2026?

Reports that Meta was considering selling access to its AI computing infrastructure sparked fears of overbuilt AI capacity, hitting US chip stocks overnight. When Seoul opened, SK Hynix fell 14.6% and Samsung Electronics 9.1% — and because those two are more than half the index, the KOSPI closed down 7.9%, with a sell-side sidecar (a temporary halt on program sell orders) triggered seven minutes after the open.

What happens if the concentration gets even worse?

Goldman Sachs estimates that each additional percentage point of combined Samsung–SK Hynix index weight could force roughly $2 billion of selling by foreign institutions bound by US Investment Company Act (40 Act) diversification limits — with up to about $20 billion at risk in total. Rising weight itself becomes a sell trigger.

How can investors get Korea exposure outside the memory duo?

Individual stocks or theme-level approaches are the usual routes. Beyond the two giants sit Korea’s AI data center power supply chain (transformers, cables, nuclear, cooling), defense, and shipbuilding — sectors this site maps in its AI power ecosystem series. That is a structural observation, not advice.