SK Hynix Fell 6% as Memory Stocks Rattled-The Real Risk Is an AI Demand Multiple Squeeze

Record profit was not enough

SK Hynix still delivered record results, but the market was looking ahead, not backward. The company posted 60.54 trillion won of operating profit on 79.32 trillion won of revenue, yet both figures missed expectations. That is why the selloff looked less like a business break and more like a multiple reset.

The market is repricing AI-memory expectations

The real issue was valuation, not demand. After falling as much as 15% during the session and closing 9.6% lower on Wednesday, SK Hynix showed that exceptional growth alone is no longer enough if it does not clear a very high bar.

That sentiment worsened when SK Hynix's U.S.-listed shares closed at $143.02, below the $149 ADR listing price. For a stock marketed as a core AI winner, trading below its own IPO reference price signals that investors are revising expectations in real time.

The selloff also reflected broader chip weakness

Part of the drop was likely sector-driven. Reuters described a broad regional rout that included an 11% fall in SK Hynix shares on Thursday after U.S. chip stocks weakened. Still, once a stock loses the listing-price level after a strong earnings print, the burden shifts back to management: higher growth is not enough if the market starts questioning how long AI spending power will last.

The debate is about timing, not whether demand exists

The bullish case still rests on real fundamentals. Reuters quoted management saying the memory market is heading for the worst-ever supply shortage in 2027, with demand expected to remain above supply well into the next decade. SK Hynix also remains one of the most direct ways to play that tight market. It held a 61% share of the HBM market, more recent reporting put its lead at roughly 58%, and it has signed about 10 long-term supply deals.

Why the long-term case still has support

If the tight-market view holds, SK Hynix remains well positioned. A leading HBM position matters because AI customers are unlikely to source from just any supplier. Management's view that demand can stay above supply beyond 2030 matters because it frames the current environment as more than a single quarter of strength. And long-term supply agreements could help smooth some of the usual memory-cycle volatility.

Why bears still control the timing argument

The bearish argument is less about demand and more about pacing. SK Hynix said delays in shipments of some advanced products limited price gains for DRAM, while analysts pointed to HBM4 chips increase from the second quarter not yet showing up at scale. That does not disprove the long-term AI story, but it does suggest the expected near-term volume ramp may be arriving later than investors hoped.

Sector conditions made that debate harder. The Philadelphia Semiconductor index has shed more than 11%, and funds tracking U.S. semiconductor stocks saw around $11 billion in outflows. Even if part of the move is valuation digestion rather than a broken thesis, it is still a real near-term headwind.

What would change the story from here

The next decision point is contractual proof, not just another strong quarter. SK Hynix has already concluded talks on about 10 long-term supply deals. What matters for valuation now is whether those agreements expand into more durable commitments that can reduce exposure to spot pricing and sudden swings in sentiment.

Signals that would support the bull case

• The 2027 supply-tightness outlook holds or tightens further.
• Long-term supply deals translate into a steadier product mix and better pricing visibility.
• Advanced-shipment constraints ease, allowing expected HBM volume growth to show up more clearly.

Signals that would strengthen the bear case

• Shipment delays begin to weigh more broadly on DRAM pricing.
• Customers slow infrastructure spending more quickly than supply can adjust.
• U.S. chip weakness and recent fund outflows turn into sustained sector de-risking.

How to read the next one to two quarters

Treat the next few quarters as a proof window. One or two additional long-term agreements, plus clearer HBM execution, should help refocus the market on operating performance. If those signs fail to appear and ADR levels remain below the reference price, further multiple compression is more likely before sentiment stabilizes.