Nvidia, AMD, and Micron shares declined in Thursday's premarket session, extending a broader sell-off in AI-related equities as Wall Street shifts its focus from demand strength to the sustainability of the infrastructure buildout. Nvidia slipped 1.8%, AMD fell 2.5%, and Micron lost 4%, while Nasdaq 100 futures dropped about 1.3%. The moves followed hotter-than-expected US inflation data and mounting stress among highly leveraged AI infrastructure companies.
Despite robust chip demand, the market is increasingly questioning whether rising financing costs and slower enterprise adoption could undermine future spending before current order books show any weakness. The concern is that the economics of building and operating massive data centers may deteriorate, even as near-term demand remains solid.
AI Financing Risk Hits Nvidia and AMD
CoreWeave, a major Nvidia customer, serves as a clear warning sign. Its stock fell another 4.5% before Thursday's open after plunging 15% on Wednesday. The company's bonds have weakened sharply as investors demand higher yields to compensate for the risk of financing expensive data centers and GPU capacity. Bernstein analyst Madison Rezaei told Investor's Business Daily that CoreWeave's unsecured bond yields had approached 15%, levels typically associated with distress.
That stress matters upstream. Nvidia and AMD rely on customers continuing to build clusters of accelerators, while Micron benefits from the accompanying demand for high-bandwidth and conventional memory. If financing costs rise sharply, marginal data-center projects become harder to justify even when demand for AI services remains healthy. As a result, future chip orders can come under scrutiny before suppliers see any deterioration in current shipments.
Enterprise AI Adoption Puts Micron Under Pressure
The key question is whether corporate adoption is catching up with the expectations already priced into AI hardware. Barclays downgraded Dell, Hewlett Packard Enterprise, and NetApp this week after share-price gains, arguing that enterprise AI spending has not yet reached a meaningful inflection point. Analyst Tim Long wrote, "We believe the strong enterprise revenues are more of a function of memory costs than a move to enterprise AI at this time."
That warning is particularly relevant for Micron. Higher memory prices can produce exceptional near-term revenue and margins, but investors increasingly want evidence that underlying AI consumption will expand fast enough to justify those prices and the capacity being built around them. The contradiction is that hardware revenues can remain strong while investors become less confident about the quality and durability of that growth.
Strong AI Demand Still Faces a Valuation Test
The bearish case is not that the AI boom has ended. TSMC reported its sixth consecutive quarter of record profit on Thursday, with net income rising 69%. The company said AI demand remains robust and continues to exceed supply, although it struck a cautious tone on 2027 because of macroeconomic uncertainty.
Jefferies economist Mohit Kumar told Reuters that "a slowdown in the AI infrastructure build-out would imply a sharp slowdown in US growth expectations." He nevertheless expects earnings season to show that demand and profits remain robust. Nvidia, AMD, and Micron do not need AI spending to collapse for their shares to struggle; growth only needs to fall short of expectations embedded in valuations.
Higher rates make that hurdle harder. Expensive financing raises the return customers must earn from AI investments, while slower enterprise adoption lengthens the time required to generate those returns. As a result, after huge rallies across memory and AI infrastructure stocks, good numbers may no longer be sufficient to sustain share prices.
This article is for informational purposes only and does not constitute financial advice.
