A selloff in AI supplier shares on Tuesday is the strongest sign to date that the mood on the trade of the decade is shifting, with growing worries that rising interest rates could scupper a spending spree bigger than the historic US railroad boom.
Nearly four years since the launch of OpenAI’s ChatGPT, cash has streamed into semiconductor companies, data center owners, cable companies and anyone reaping revenue from an unprecedented race to build computing infrastructure.
Cracks in those bets began in June and are now widening across markets, with a data center IPO collapsing last week in Australia and credit investors becoming picky on AI debt.
In Asia, selling of South Korean chipmaker shares extended into a fifth straight day on Tuesday, with the KOSPI index falling more than 6%, while in Tokyo memory-maker Kioxia, down 13%, led AI-related losses.
In the US, the Philadelphia Semiconductor Index was on track for a sixth straight decline, down 1.5%, and Micron Technology was down 5%. The SOX was up 68% for the year headed into Tuesday’s trading, while Micron stock has tripled in 2026.
The falls are the heaviest since the same sort of stocks collapsed at the end of July. The KOSPI is on its longest losing streak for more than two years, down a quarter from July highs, while Shanghai’s tech-focused STAR Market has lost more than a third from July peaks.
“The most important development has been not just bond yields going up, but the volatility for bonds … and that’s starting to work its way out through asset pricing across the board,” said Damien Boey, portfolio strategist at Wilson Asset Management in Sydney.
“It’s a very uncomfortable situation on many fronts. … Maybe AI capex just comes off the boil a little bit, and maybe a few players are casualties in that,” Boey said.
The SOX dropped 2.7% on Monday and even heavier selling drove Seoul-listed chipmakers Samsung Electronics and SK Hynix down by 10% and 8%, respectively.
On Monday, the research arm of BlackRock, the world’s biggest asset manager, downgraded its US stocks outlook to neutral, citing “potential volatility around the AI buildout” and turned bullish on short-term Treasuries.
“The consequences of an AI correction could be global,” the International Monetary Fund warned on Tuesday in its latest financial stability report.
The IMF called out AI’s contribution to growth but said there was growing evidence that capability investments and economic gains may not be as big as expected.
COST OF CAPITAL
Caution is also coursing through primary markets.
Last week, Firmus, a data center operator backed by Nvidia, shelved a $5 billion IPO in Australia as investor interest shrivelled, underscoring how suddenly risks are becoming the chief focus in allocating capital to the sector.
Tuesday’s dollar and dim sum bond sale by Tencent Holdings in Hong Kong was the latest deal in debt markets’ sights, with guidance for 10-year dollar notes about 1.15 percentage points on top of Treasury yields, which sit at 5.2%.
Japan’s tech investor SoftBank last month sold the biggest high-yield debt pile ever and promised to pay as much as 9.75% on $4.5 billion it raised at a 7-1/2-year tenor.
On Tuesday, fears about the risks tied to the debt-funded expansion of AI infrastructure and related sectors were evident in the US as well. The yield on a bond sold in June by data center operator CoreWeave hit 15%, up from 9.625% at issue. CoreWeave shares were down 4% on Tuesday and are 49% off their 52-week high.
The mood shift was front of mind at a slew of top financial conferences in Singapore, most recently on Monday when Broadcom’s CEO said capital was the biggest constraint on growth.
To be sure, few are calling time on the AI trade altogether, and on Monday Meta shares jumped 3.4% on the release of a new AI model, but many see semiconductors as a warning sign.
Marc Velan, head of investments at Lucerne Asset Management in Singapore, believes the “brutal session” in South Korea says more about positioning and the cost of capital than about any fundamental deterioration in AI demand.
“I suspect we’re entering the next phase of the AI cycle, where markets increasingly focus on monetisation and returns on invested capital — I wouldn’t call this the end of the AI trade,” Velan said. “But Korea could be an early warning of a broader shakeout.”
(Reporting by Tom Westbrook)
