Can SK Hynix Rebound? The Question Left by a Roller-Coaster Week

Here is SK Hynix’s price path over the last four trading days: -12.6% on July 13, +8.8% on the 15th, -11.5% on the 16th. Over the same stretch, the same company’s Nasdaq ADR (SKHY) printed a one-day gain of +27.3%. When one company’s price moves like this, it isn’t a fundamentals problem — it means the market has not yet agreed on which story to read this company through.

So the question is simple: can Hynix rebound? Let’s lay out the bull case and the bear case, then identify what actually decides it.

Disclaimer: This is general analysis, not investment advice. Always do your own research before making any financial decision.

What happened in four days

Date	SK Hynix	KOSPI	Trigger
Jul 13 (Mon)	-12.6%, into the 1.9M-won range	-8%, circuit breaker (7th this year)	Hormuz blockade, cross-listing arbitrage, Korea Investment’s estimate cut
Jul 14 (US)	ADR +27.3% ($193.92)	—	US sentiment ran the opposite way
Jul 15 (Wed)	+8.8%, back above 2.08M won	+6.2%, 7,284 (retook 7,000)	Cooling June US CPI; foreigners bought 650B won of Hynix
Jul 16 (Thu)	-11.5%, around 1.84M won	-6.4%, 6,820	CoreWeave weighing memory-price hedges, China’s CXMT expansion, Buffett’s AI-bubble warning

The July 13 shocks — geopolitics, post-Nasdaq-listing arbitrage — are old news now (see our crash breakdown). What matters is the 16th: the narrative that took over that day is “the memory cycle has peaked” — and that is the real battle line for any rebound.

The bull case: the numbers still side with the cycle

First, the absolute earnings are still staggering. Brokerages cut their Q2 operating-profit estimates — to 60–64 trillion won (Mirae Asset 62.3T, iM Securities 63.7T, Korea Investment 60.4T). By any historical standard those are record-class numbers, and iM kept its 3.5M-won price target even while trimming. As Samsung proved six days earlier with a record 89.4T-won quarter, the cycle’s earnings power itself has not broken.

Second, American investors keep buying. The ADR jumping 27% while Seoul dumped means the US market values the same asset far higher. On the 15th that gap worked in Seoul’s favor — foreigners net-bought 650 billion won of Hynix, the day’s top pick. Price discovery across a fresh dual listing is still in progress, and it doesn’t only pull downward.

Third, the HBM bottleneck hasn’t gone anywhere. Even Mirae Asset, while cutting estimates, stated flatly that “HBM growth continues.” As long as AI datacenter buildout doesn’t fully stop, only a handful of companies can supply leading-edge HBM.

The bear case: the story has changed

First, peak-out evidence is getting concrete. Both triggers on the 16th were specific. CoreWeave — a major AI cloud — reportedly weighing memory-price hedges: a whale customer preparing to bet that today’s prices are the top. And China’s CXMT announcing capacity expansion: the counter-material to the shortage narrative. Add an overnight rout in US memory names, and foreigners plus institutions sold 3.7 trillion won in a day.

Second, AI-bubble warnings are gaining force. Warren Buffett’s bubble warning and datacenter construction delays surfaced together. As we covered in the AI bubble debate, a structure investing $13 for every $1 of AI revenue eventually demands a correction — and when it comes, memory sits in the front row.

Third, the flow structure is still unfavorable. The “buy ADR, short Seoul” arbitrage created by the dual listing, plus dilution from new-share issuance — structural pressures that only time unwinds.

The decider: earnings in late July

The near-term catalyst converges on one event: Q2 results, expected in late July (last year’s pattern points to around July 24). Three things to check:

Does operating profit hold the lowered consensus (60–64T won)? Below 60T and the peak-out narrative wins; above 65T and “the cuts went too far” takes over.
HBM volume and price guidance. Commentary on next year’s supply contracts is nearly the only card that can offset the CoreWeave-driven “price top” fear.
Shareholder returns. How the Nasdaq-raised cash is used — investment vs. returns — answers the dilution worry.

Bottom line: can it rebound?

Short term — yes, it can, but until earnings this is an event-waiting zone. As the 15th showed, give this stock one clear day and it has the energy to jump 9%. As the 16th showed, one peak-out headline erases that in a session. When a single news item moves the price ±10%, calling the direction in advance isn’t investing — it’s a coin flip.

Structurally — the question has narrowed to “is the memory supercycle over?” Absolute earnings, the HBM bottleneck, and US sentiment (the ADR) say not yet; CXMT’s expansion, CoreWeave’s hedging, and bubble warnings say closer than you think. The late-July earnings call — especially HBM guidance — will be the first official verdict in that argument.

The roller coaster continues for now. But remember what these four days actually proved: not that Hynix is in crisis, but that this market has not yet decided what Hynix is worth.

Sources

This article may contain affiliate links. Nothing here is financial or investment advice.