SKHY Fell 21% in Three Sessions. What Broke Wasn't the AI Memory Trade

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.

The largest ADR listing in history just had a brutal first week — and almost every US headline drew the wrong lesson from it. This is not financial advice — a view from Seoul, not a recommendation to buy or sell anything. (Won-to-dollar conversions use roughly ₩1,480/$, the July 16–17 range; verify at the time of reading.)

SK Hynix (Nasdaq: SKHY) fell from a Tuesday high of $193.92 to $152.31 by Thursday — about 21% in three sessions. New York’s fast read: the AI memory trade broke; the IPO was the top. But put the Seoul tape next to the New York tape and a different picture appears. Most of that 21% wasn’t the price of a company. It was the price of a premium — the extra that New York was paying over Seoul for the exact same stock. That premium ran near 38% on Tuesday and compressed to about 22% by Thursday. The company didn’t reprice. The premium did.

The myth: “first-week crash = the AI memory trade is dead”

On a US screen it’s simple. Debut at $168.01 (+12.8%) on July 10, a run to $193.92 on Tuesday, then $152.31 by Thursday. Add Michael Burry’s short headlines and the story writes itself: “biggest ADR ever = biggest top ever.” The Philadelphia Semiconductor Index is down more than 19% from its late-June high, so the mood music fits.

The data: the same stock had two prices

Start with the structure. SKHY trades as an ADR — an American depositary receipt, a US-listed certificate standing in for foreign shares — and ten of its ADSs (the individual share units) equal one ordinary share in Seoul (KRX: 000660). Thursday’s Seoul close was ₩1,842,000 (≈$1,245; down 11.53%). Divide by ten and convert at ~₩1,480/$ and one ADS is worth about $124 in Seoul terms. That same day, Nasdaq closed SKHY at $152.31 — meaning US investors were paying roughly a 22% premium for the identical asset. On Tuesday, that premium was near 38%.

Date	SKHY (Nasdaq)	Seoul 000660	Premium
Jul 10 (debut)	$168.01 (+12.8%)	—	—
Jul 14 (Tue)	$193.92 (52-wk-high area)	rebounding	~38%
Jul 15 (Wed)	falling	Seoul catches up	~30%
Jul 16 (Thu)	$152.31	₩1,842,000 (≈$1,245, −11.53%)	~22%
Jul 17 (Fri)	traded without a Seoul anchor	market closed (holiday)	—

Of the ~21% drop from Tuesday’s $193.92 to Thursday’s $152.31, a large share is premium compression. Measured against the $149 IPO price, Thursday’s close was only about +2% — most of what “fell” was the extra New York had tacked on, not the company.

Where the myth is right: the chip pullback is real

The premium story doesn’t erase sector risk. On July 16 the KOSPI (Korea’s main stock index) fell 6.37% (to 6,820.60) and a sell-side sidecar — an automatic brief halt on program sell orders — triggered. That same day, Samsung and SK Hynix shares plus the sixteen single-stock leveraged and inverse ETFs built on them accounted for a staggering 96.3% of total market turnover — the whole market had effectively become a derivatives playground for two memory names. When chips wobble, the index lurches. In the US, the Philadelphia Semiconductor Index (SOX) is down 19% from its high. “The expectations stacked on memory are cooling” is a claim with real support.

🎩 Under the Gat — Just don’t reverse the order of events. While Seoul rebounded Tuesday and caught up Wednesday, the premium shrank partly because Seoul rose toward New York as much as New York fell toward Seoul. On one screen it’s a crash; overlay both and it’s a convergence.

Where the myth is wrong: that premium was plumbing, not conviction

Why could the same stock carry a 38% premium at all? The answer isn’t demand — it’s plumbing.

First, new ADR supply is limited. Citibank took 17.79 million ordinary shares and created 177.9 million ADSs; the path to convert more Korean stock into new ADRs is narrow (the reverse direction, ADR→ordinary, is unrestricted). Big US demand meeting a thin supply pipe means the same asset can trade richer at the New York window. If that structure is unfamiliar, the prospectus breakdown lays out the 10:1 ratio and the share math first.

Second, the arbitrage clock was locked. The conversion-request window is expected to open only after the July 29 domestic-listing step (subject to Citibank’s guidance). Until then, “buy in Seoul, sell in New York” is hard to execute — the premium was visible, but there was no hand free to erase it.

Third, on Friday Seoul was simply closed. On a Korean market holiday, with only Nasdaq open, the ADR floated without its anchor — which is the kind of setup that produces an anchorless, whippy session. (Reported intraday and closing prices for that Friday vary by source and aren’t corroborated here, so we’ll leave the specific figures out; the point is structural, not the decimal.)

Valuation says the same thing the plumbing does. On Barron’s numbers SKHY’s forward P/E is about 5.71x — essentially level with Micron (MU) at 5.93x. That’s normalization to a peer multiple, not a collapse. And on the fundamentals, at least one Korean brokerage estimates DRAM supply is meeting only 75–80% of demand, sliding toward the 60%s by 2027 — a shortage signal, not a glut.

🎩 Under the Gat — When the same company has two prices, one of them isn’t the company — it’s the plumbing. Most of the −21% printed in US accounts was the price of the plumbing, not the price of HBM (high-bandwidth memory — the chips inside AI accelerators). The catch: about 22% of that plumbing charge is still on the meter.

The verdict

What broke in week one was a scarcity premium, not the AI memory thesis. It’s not a clean sweep for the skeptics either — the sector pullback and leverage froth are real, and the remaining ~22% still means you’re holding the same stock ~22% richer than Seoul does. Once the conversion window opens and the two tapes trade side by side again, that 22% looks like a hard number to sustain.

What it means for investors: July 29 is a double exam

At 9 a.m. Seoul time on July 29, Q2 earnings land. If Korea’s two-step earnings rhythm (preliminary, then final) and SK Hynix’s quirk of skipping the preliminary are unfamiliar, the earnings-season guide is worth reading first. Around the same date, the conversion window is expected to open. Earnings test the numerator of the premium (the company’s value); conversion tests the denominator (the plumbing constraint). Two exams, one day.

The flows behind it are moving too. Margin loans this week stood at ₩34.37 trillion (~$23.2B) — down more than ₩4 trillion from the record ₩38.63 trillion (~$26.1B). Reading the bottom as a de-leveraging rate rather than an index level — the frame from our margin-debt piece — that rate is actually climbing. Meanwhile assets in 2x-leveraged SK Hynix funds swelled to about $7.78B, and from August 5 regulators triple the minimum deposit for single-stock leveraged ETFs to ₩30 million (~$20,000) — a move to shrink the amplifier.

Four checkpoints: (1) the July 29 Q2 print (9 a.m. KST, per company IR); (2) how fast the premium shrinks once conversion opens; (3) whether margin loans keep falling; (4) turnover concentration around the August 5 leverage rule — does that 96.3% come down?

🎩 Under the Gat — If you bought at $190 during debut week, this piece isn’t a comfort. But knowing whether you bought the company or the premium changes the quality of your next decision. July 29 is the day that distinction shows up as a number. — A view, not advice.

This is not financial advice. Won-to-dollar conversions use ~₩1,480/$ (July 16–17 range). Prices, premiums, margin-loan balances, fund assets, and the July 29 earnings and conversion timing should all be re-verified at the time of reading against primary sources (company IR, Nasdaq, KRX, the Korea Financial Investment Association, and the Korea Securities Depository). SK Hynix’s Q2 date is per company IR — not a consensus-calendar guess.

Frequently Asked Questions

Why does SK Hynix’s Nasdaq ADR trade at a premium to its Korean shares?

New ADR creation is limited and the conversion window between the ADR and the Seoul-listed ordinary shares had not yet opened, so US demand met a thin supply pipe. That scarcity — not the company’s fundamentals — is what let the same stock trade richer in New York (as much as ~38%, later ~22%). This is not financial advice.

Is one SKHY share equal to one SK Hynix Seoul share?

No. Ten ADSs equal one ordinary share (KRX: 000660). To compare, divide the Seoul price by ten and convert at the won/dollar rate. Skip that step and the “premium” will look far larger than it is.

Was SKHY’s first-week drop a sign of weak earnings?

It looks more like a price-structure event than an earnings event. On a forward-P/E basis SKHY sits near Micron — roughly 5.7x versus 5.9x — which reads as normalization to a peer multiple, not a fundamental break. The earnings answer itself comes on July 29. Not financial advice.

If Seoul shares are ~22% cheaper, why not just buy those?

In theory you save the premium; in practice the friction is the premium. Direct KRX access, currency conversion, settlement lag, and different tax treatment all sit in the way — and until the conversion window opens, arbitrage can’t easily close the gap.

When does the premium disappear?

There’s no fixed date, but the first gate is the conversion window expected around the July 29 domestic-listing step. Even then, conversion fees, settlement lag, and issuance limits can leave part of it in place. Verify current terms.