SK hynix Looks Cheap After a 10% Drop-But Investors Are Still Paying for a Perfect AI Quarter

The 10% drop reflected expectation management, not weak results

A 10% drop after bumper quarterly results suggests investors were less focused on whether SK hynix had a strong quarter than on how much success was already priced in.

What came in was hard to dismiss. The company reported 79.3187 trillion won in revenue, 60.5426 trillion won in operating profit, and a 76% operating margin. Those are exceptional numbers. But once a stock has benefited from an AI-driven run, investors start looking for limits instead of rewarding standout execution.

Why bulls and bears are reading the same quarter differently

Bulls see a company still monetizing the AI buildout with unusual strength. Bears see a memory supplier exposed to the industry's normal volatility, especially with delays in some advanced-product shipments and fresh concerns that AI infrastructure spending could cool.

That caution spread beyond SK hynix. The stock's decline came alongside pressure on memory-chip stocks moving lower in sympathy, which suggests the market was reacting to broader worries about how durable the current memory upcycle really is.

The sell-off, then, is not a verdict on current profitability. It is a debate about how much of the next few quarters is already spoken for.

Q1 and Q2 together show the operating strength-and the accounting noise

Q1 already showed the trend before the Kioxia gain

A quick clean-up of the quarter helps explain the disagreement. In Q1, SK hynix already posted 52.5763 trillion won in revenue, 37.6103 trillion won in operating profit, and a 72% operating margin. That makes the Q2 result look like part of a continuing trend rather than a one-off spike.

Q2's net income overstated the core business

In the second quarter, the key accounting point is that net income of ₩93.9 trillion was much larger than operating profit of ₩60.54 trillion because of the Kioxia sale gain. That does not weaken the case that SK hynix had a strong quarter. It does mean investors should separate operating performance from a one-time balance-sheet gain.

HBM leadership supports durability, but memory demand is still cyclical

The more durable piece of the story is AI-memory leadership. SK hynix had already commanded a 61% share of the HBM market, and Q1 results highlighted higher sales of high-value-added products such as HBM and server DRAM modules. That points to real mix improvement inside the AI stack.

The cyclical risk is still there. Management said delays in some advanced-product shipments limited DRAM price gains. If those delays ease, margins may hold. If they persist, part of the premium investors are paying becomes harder to justify.

A cheap trailing multiple can still be expensive near a cycle peak

Pricing power helped this quarter, and that matters for valuation

Near a cycle peak, last quarter's earnings are often driven as much by pricing as by execution. SK hynix benefited from enormous price increases tied to supply shortages, while Samsung was expected to post its third straight record-high operating profit for the same reason. In that environment, a low multiple based on the latest quarter can still be misleading.

Long-term agreements improve visibility, but they do not remove cyclicality

What matters now is whether SK hynix can lock in enough of the AI demand wave to outlast a future supply catch-up.

On that score, the evidence is meaningful. The company said it has long-term agreements with around 10 key customers, supported by multi-year contracts tied to technological innovation. That does not eliminate memory-market cyclicality. It does suggest more visibility than a pure commodity story would offer.

What to watch before treating the pullback as a buy

After a 10% slide following bumper quarterly results, the setup looks more interesting, but it is not a clean buy. The next few signals matter more than the headline beat.

Signals that would strengthen the case

• Memory-group behavior. If the sell-off proves short-lived after pressure on memory-chip stocks moving lower in sympathy, that would point more to positioning than to a fundamental reversal.
• Memory pricing. If pricing stiffens again after shipment delays ease, the operating-leverage case improves.
• HBM4 progress. SK hynix said HBM4 achieves customer-required operating speeds. Continued leadership there would matter more than another quarter of extreme headline numbers.
• Customer demand. Management has said major customers are still requesting more memory supply. That message needs to hold as the year progresses.

Signals that would weaken it

• Persistent shipment delays. If delays in some advanced-product shipments continue to limit DRAM price gains, some of the valuation premium is still tied to cycle timing.
• Softer AI-spending language. The market already reacted to concern that AI infrastructure delays pose biggest risk. If that concern gains traction, valuation can compress before earnings do.
• Faster supply expansion. If global memory supply growth arrives sooner than expected, the current shortage-driven profit outlook becomes less durable.

For now, the more disciplined stance is to wait for the market's reaction to do its job. The case improves when investors stop anchoring to one elite quarter and start paying for demand visibility.