What actually happened last week
US tech stumbled and the market's question changed from Will AI chips sell to Who pays for all the concrete, power and servers. The Nasdaq 100 posted its weakest week since April 2025, and US semiconductor stocks recorded an eight day slide, the longest since 2008, leaving the sector's benchmark down about 16%.
At the same time, crude climbed to $110 a barrel, a level not seen since May, reigniting inflation concerns that could keep interest costs elevated. The chill spread to Asia: a Goldman Sachs basket of regional AI hardware stocks fell 11% in its sharpest weekly drop since mid July, even after Taiwan Semiconductor Manufacturing Co. guided to stronger revenue.
Who in Asia got hit and why
Asia's winners of the AI boom - chipmakers and the suppliers behind printed circuit boards, precision parts and factory tools - led the retreat. Nan Ya Plastics Corp., this year's top performer on the MSCI Asia Pacific Index, fell 12% from a record. Unimicron Technology Corp., the index's second best performer, dropped 14% on the week. Samsung Electro Mechanics Co. slid 23%.
This matters because TSMC, SK Hynix Inc. and Samsung Electronics Co. earn their keep from steady investment in the systems that use their chips and memory, and component and equipment makers ride the same spending cycle with less clarity on project timing. Collectively, TSMC, Samsung and SK Hynix account for almost one-third of the MSCI Asia ex Japan Index and roughly one-fifth of the wider Asian benchmark, and supplier firms further increase that weighting.
Parts makers have also dominated the leaderboard: eight of the 10 best performers in the MSCI Asia Pacific Index this year are component firms, each up at least 236%. Outsized gains like that can be vulnerable if spending growth cools or schedules slip.
When tech and capital trends feel uncertain, sign up for the free Market Briefs newsletter to get five-minute daily clarity
Why financing is the swing factor
The conversation now centers on if hyperscalers and neocloud players can continue financing the buildout needed to deploy AI chips, and whether the paybacks will warrant the outlays. Debt issued by neocloud provider CoreWeave Inc. slid into distressed levels last week amid investor doubts over whether AI demand and revenue would support its borrowing, while its stock dropped 12% on Friday and 25% for the week, the worst since November.
Zurich-based GAM Investment Management fund manager Jian Shi Cortesi said, "Financing costs are like gravity for the AI boom." "Rising borrowing costs could make infrastructure projects harder to finance and delay spending. Even a delay could trigger a reset in Asian suppliers' earnings expectations and valuations." Paris-based Financiere de l'Echiquier portfolio manager Kevin Net added, "AI funding stress is a bigger challenge for downstream beneficiaries than for the core compute ecosystem," and cautioned, "Basically, the further away you get from the actual chip, the greater the risk you might see timing delays."
Industry voices are flagging the same pressure point. Hock Tan, the Chief Executive Officer of Broadcom Inc., has argued that capital - more than memory supply or fab capacity - will be the primary brake on AI expansion. Gartner Inc. has warned that by 2030, AI infrastructure outlays exceeding $3.5 trillion could face impairment risk if spending runs ahead of the revenues required to sustain it.
Looking ahead, hyperscaler capital spending is expected to top $3 trillion over the next three years. Note: Oracle's fiscal year ends May 31; Microsoft's fiscal year ends June 30; others referenced report on a calendar year basis.
What this could mean for your portfolio
At Federated Hermes Ltd. in London, Kunjal Gala, who leads the global emerging markets team, said, "We are not worried about the ongoing chip selloff," and pointed to strong AI demand. His fund holds an overweight position in Korea and owns TSMC shares.
The bottom line: higher rates, $110 oil and a stressed neocloud borrower have put the funding side of AI under a spotlight. That leaves Asia's AI hardware plays especially sensitive to credit conditions and any pullback or pause in hyperscaler spending.
Markets and supply chains affect everyday plans, so join the free Market Briefs for quick daily context across markets and economy
