SANTA CLARA, Calif. — Nvidia on Wednesday said second-quarter revenue rose to a record $96.2 billion, more than double the same period a year earlier, and told investors to expect $108 billion in the current quarter — guidance that would leave the chipmaker within a rounding error of the $100 billion-a-quarter club previously reached only by Amazon, Apple and Alphabet.
The results and outlook are the market's cleanest read yet on whether the artificial-intelligence buildout is still accelerating or beginning to plateau, and Wall Street took them as evidence of the former. Shares rose 4.7 percent in after-hours trading, and Nvidia's market capitalization now tops $5 trillion, according to the BBC.
Data center in the driver's seat
Data-center revenue, the segment that sells graphics processors to the hyperscalers building out AI infrastructure, reached a record $89 billion in the quarter, up 117 percent from a year earlier, according to The Verge's account of the earnings release. Profit more than doubled to $59.7 billion in the quarter, The Verge reported.
Almost every major AI infrastructure spender — Amazon, Meta, Google and Microsoft among them — buys Nvidia's chips to train and run models. Nvidia has also begun writing checks back into the ecosystem it supplies, providing funding to OpenAI, Anthropic and SpaceX to keep the buildout going, the BBC reported. The move stitches Nvidia into the balance sheets of the customers who are its largest sources of demand, a circularity that has drawn the attention of both bulls and skeptics.
"Full steam"
"AI has reached its inflection point," Chief Executive Jensen Huang said in prepared remarks, describing the infrastructure buildout as running "at full steam."
Matt Britzman, senior equity analyst at Hargreaves Lansdown, called the report "another monster set of results," telling the BBC that guidance for the current period "points to revenue comfortably above $110bn." Wall Street and outlets covering the release read the guidance as extension of the AI capital-expenditure cycle, an argument some money managers had begun to question mid-year as they worried hyperscalers were pulling demand forward.
Consumer chill
Not everything was on fire. Nvidia's "edge computing" category, which includes its consumer gaming business, brought in $7.2 billion — up 27 percent from a year earlier but flagged by the company as tempered by "slower consumer PC sales" against a backdrop of "elevated memory and systems prices," according to The Verge. The Verge noted that Nvidia had warned before the release that prices for its AI chips could also rise, a signal that customers should expect to pay more per unit of compute even as the volume of demand continues to expand.
What could break the trade
Competition is emerging. Nvidia's largest customers are designing their own accelerators, and cheaper Chinese suppliers are pressing in on the low end. So far those pressures have not shown up in the numbers, the BBC noted.
The broader risk is concentration itself. Roughly 40 percent of the U.S. stock market's value now sits in ten companies with heavy AI exposure, the BBC reported — a share that leaves the index and Nvidia moving in near lockstep. A miss on Nvidia's next guidance, or a downshift by one hyperscaler, would land harder on the broader market than it would have two years ago.
The bear case
Bears say much of the current spending is going into hardware whose useful life is being shortened by each generation of Nvidia chip, and that data-center demand is being pulled forward by customers hoarding capacity. Neither claim is refuted by Wednesday's numbers; both would show up in a guidance miss rather than a beat.
The other adversarial read, one Nvidia itself has begun to engage with, is that end-user monetization of AI is not keeping pace with the infrastructure being built to serve it. Huang's inflection-point language is the counter, and Wednesday's guidance is the evidence he is offering. Neither settles the argument.
Nvidia's fiscal third quarter closes in October. That report, and the guidance that comes with it, will be the next test of whether the AI trade is still accelerating — or whether $108 billion is where the curve begins to bend.