$600 billion wiped out in a month, cracks exposed in SK Hynix chips

A nearly $600 billion rout in just a little over a month has flipped SK Hynix. from one of the world's hottest artificial intelligence trades to one of the biggest portfolio question marks.

Shares of the South Korean memory chipmaker have plunged 47% from their all-time high in June on concerns about overcrowding and a surge in leverage-induced volatility. Its loss in market value over that span rivals that of even SpaceX.

The stock tumbled almost 15% Tuesday after The Information reported on Chinese progress with deep ultraviolet (DUV) lithography machines, fueling concerns over a potential flood of new capacity amid broader worries over the AI rally.

Record earnings expected
While SK Hynix is likely to report another quarter of record earnings on Wednesday as the AI boom drives chip prices higher, investors are turning skittish. The primary fear is that rising memory costs will force customers to reduce usage and turn to cheaper alternatives.

"The debate now is whether memory is taking too much of the pie," said Andy Wong, head of multi-asset at Pictet Asset Management HK, whose fund has reduced its position in SK Hynix in recent weeks. The market wants to see if "anything can shift the perception that SK Hynix is squeezing too much margin out of the supply chain."

Also Read: Global Market Today: Oil jumps on renewed Middle East tension, US stock futures dip

The chipmakers's shares have greater implications for global investors since the listing of its American depositary receipts, which have fallen below their July 9 offering price. The proliferation of leveraged exchange-traded funds tied to the stock has also grabbed attention due to the unprecedented volatility they have unleashed.

Triple-digit annual gains
The stock is still on track for another year of triple-digit gains, buoyed by the firm's early lead in high-bandwidth memory for AI. This month's pullback tracks broader caution over the tech hardware sector as doubts multiply over the durability of hyperscaler outlays.