SK Hynix Plunges 10% as AI Funding Fears Trigger Asia Chip Rout

By Elena Vasquez

SK Hynix lost more than 10% of its market value in a single session Thursday, as a sell-off in Asian technology stocks tracked Wall Street losses and put the financing of the AI build-out at the center of investor anxiety.

The numbers came fast and they came brutal. Samsung Electronics fell 7.1%. LG Innotek sank 8.1%. Seoul Semiconductor shed 7.7%, and Samsung SDI declined 6.8%. The damage stretched across the Sea of Japan before the morning was out: SoftBank Group tumbled 7.9%, Advantest dropped 6.9%, and both Tokyo Electron and Renesas Electronics lost 5.4%.

This wasn't a random dip. It was the second act of a rout that started in New York on Wednesday, and it landed hardest on the companies that have ridden the AI investment boom the highest.

What drove the sell-off?

Two fears collided this week. The first is macro: hotter-than-expected U.S. inflation data fueled concerns that the Federal Reserve could raise interest rates further. The second is specific to the AI trade: investors are increasingly worried that higher borrowing costs could weigh on the massive capital spending required to build out AI infrastructure.

Wall Street absorbed the first blow on Wednesday. The tech-heavy Nasdaq Composite dropped 1.88% to close at 26,692.87, while the S&P 500 shed 1.11% to end at 7,718.94. The Dow Jones Industrial Average slipped just 0.09% — a split that tells you exactly where the pain was concentrated.

Technology shares led the broader market losses. Asian markets, which reopen after Wall Street closes, inherited the pressure and amplified it.

The stakes here are straightforward. Chipmakers like SK Hynix and Samsung Electronics supply the memory and silicon at the heart of the AI infrastructure boom. If financing costs climb, the hyperscalers building data centers may pull back on capital expenditure — and the demand that has lifted these stocks for two years could soften.

What did Broadcom and Intel say?

The market didn't invent this fear on its own. It heard it from the top.

Chief executives of Broadcom and Intel warned on Monday that financing could become a major headwind to the AI boom. Broadcom's Hock Tan went further, saying financing could be a bigger challenge to AI expansion than shortages of chips or electricity.

That is a striking claim. Until recently, the binding constraints on AI growth were physical: not enough advanced chips, not enough power. Tan's comment reframes the problem. The constraint may now be the cost of money itself.

U.S. Treasury yields climbing above 5% are the mechanism. Higher yields raise the cost of financing the data centers, GPU clusters and fab capacity that the AI economy requires. For companies carrying heavy debt or signing long-dated contracts, the math gets harder with every basis point.

Is AI demand actually cracking?

Not according to Dan Ives, tech analyst at Yorkville Ives. In a note issued as the sell-off deepened, Ives argued that the market is conflating a financing problem with a demand problem.

"Higher yields raise the cost of financing these builds, and investors are right to focus on balance sheets and contract structure, but rates are a financing question, not a demand question," Ives said.

He added that underlying demand for AI computing power remains robust — and that the latest pullback presents a buying opportunity. In his view, the AI infrastructure boom is still in its early stages despite the mounting concerns over funding costs.

Ives was specific about what would actually change his mind.

"What would change our view is a demand signal, not a rate signal," he said.

He pointed to three warning signs investors should watch: potential cuts in hyperscaler capital expenditure plans, canceled contracts, or falling GPU rental prices. None of those, he noted, has appeared.

Why the chip sector took the worst of it

The asymmetry of Thursday's losses tells the story of sector concentration. SK Hynix fell more than 10% while the Dow slipped 0.09% the day before. The companies hit hardest — memory makers, chip equipment suppliers, semiconductor testers — sit directly in the path of AI capital spending.

Advantest, which makes semiconductor testing equipment, dropped 6.9%. Tokyo Electron, which supplies chipmaking equipment, lost 5.4%. SoftBank Group, the technology investment giant with vast AI-related holdings, tumbled 7.9%. These are leveraged plays on the AI build-out continuing at its current pace.

That pace depends on capital staying cheap enough to fund it. With the Federal Reserve potentially raising rates further, and Treasury yields already above 5%, the financing question Tan raised on Monday has moved from executive talking point to market-moving force within three trading days.

The rout also exposes a vulnerability in the AI investment thesis itself. Much of the boom has been built on the assumption that spending on infrastructure would keep accelerating. Broadcom and Intel's leadership saying out loud that financing could be the bottleneck shifts the debate from whether AI demand exists to whether AI demand can be funded at current interest rates.

What comes next

For now, the divide is between those who see a financing problem and those who see the start of a demand rollover. Ives places himself firmly in the first camp, telling clients the pullback is a buying opportunity while the boom remains early.

The test he laid out gives investors a concrete checklist. Watch hyperscaler capital expenditure plans. Watch for canceled contracts. Watch GPU rental prices. If those hold steady while rates climb, the sell-off will look like a valuation correction. If they crack, Thursday's 10% drop at SK Hynix may mark the beginning of a harder reappraisal of the entire AI hardware trade.