Nvidia delivered a dazzling set of results, but investors punished the stock anyway: shares fell more than 5% on Thursday, marking the company’s worst trading day since April. The sell‑off reflected market anxieties that go beyond a single quarter — investors are increasingly focused on whether the AI spending boom that has powered Nvidia can be sustained and converted into predictable cash flow.
The company reported another quarter of rapid growth driven by its data‑centre business, which accounted for 91% of sales. Data‑centre revenue came in at $62.3 billion, above Street estimates of $60.69 billion, and Nvidia issued an unusually ambitious revenue guide of about $78 billion for the next fiscal period. Those figures underline how central GPU‑accelerated training workloads have become to corporate AI capacity building.
Yet strong top‑line numbers did not calm nerves. Analysists and portfolio managers pointed to three interlocking worries: sky‑high valuations that already price in near‑perfect execution, the sustainability of cloud providers’ AI capital expenditure, and the uncertainty surrounding a potential multibillion‑to‑hundreds‑of‑billions‑dollar strategic investment from OpenAI.
Nvidia’s own 10‑K added to fretful sentiment. The filing said talks with OpenAI over an investment and cooperation agreement were ongoing but not guaranteed to close — language that investors read as a realistic warning that a headline‑grabbing deal may not materialise. Market participants also fretted that if large cloud customers pause or trim AI spending, that would show up in Nvidia’s revenues within a few quarters.
The market reaction was not confined to Nvidia. Broadcom slid more than 3% and TSMC dropped about 2.8%, as traders reassessed the outlook for the chip ecosystem that supplies hyperscalers and hardware partners. Some investors argued the drop reflected mood more than fundamentals, noting that 61 of 66 analysts covering Nvidia still rate the stock a buy and that consensus target prices imply roughly 37% upside from current levels.
Strategic concerns extend beyond demand. Industry discussion is shifting from training to inference: specialised chips optimised for low‑latency, high‑volume inference workloads could erode Nvidia’s training‑centric dominance. Nvidia’s Rubin architecture — explicitly designed for inference — aims to blunt that threat, but rivals and system‑level innovation mean the moat is being tested.
For investors the question is simple but hard: can Nvidia convert its leadership and product momentum into durable, predictable cash flows as the market moves from an investment phase into one that prizes margin of safety and return on capital? The answer will determine whether the current volatility is an opportunity or an early sign of a re‑rating across the AI hardware space.
In the near term, Nvidia’s share price will likely remain sensitive to three things: concrete guidance on future demand from hyperscalers, clarity on any investment or contract with OpenAI, and evidence that inference workloads will not materially compress Nvidia’s addressable market. Until those questions are answered, even stellar quarterly numbers may not be enough to soothe a market that has started to price in both exuberance and risk.