Nvidia just reported the kind of quarter most companies never see in a decade. Revenue for its fiscal third quarter came in at $115.8 billion, up 103% from a year earlier, according to the company’s November 17 release. Net income more than doubled. Guidance for the current quarter landed above what analysts had expected. And the stock went down anyway. TechTimes reported that shares closed Wednesday, November 18, down 2.89% at $150.96, and fell further on Thursday, November 19, closing at $145.98, down 3.3%. 24/7 Wall St. put the one-month decline at roughly 27%; from Wednesday’s close of $150.96, the drop over the prior month is closer to 30%. When a company doubles its sales and investors still head for the exit, the interesting question is not whether the quarter was good. It is what the market thinks it found underneath.
The Headline Numbers Were Not the Problem
Start with what went right, because plenty did. Data center revenue, the engine of the entire AI trade, reached $107.6 billion, up 110% year over year, per Nvidia’s press release. Consensus revenue had been near $109 billion, based on LSEG estimates cited by Blockspace, so the beat was not a rounding error. Colette Kress, now interim CEO while keeping her CFO role, ran her first earnings call since Jensen Huang’s resignation last week. She told analysts the company expects fiscal 2028 revenue growth of about 75%, up from a prior view of 70%, a raise reported by Blockspace, TechTimes, and SiliconANGLE alike. She also said Rubin, the next generation platform, brought in nearly $20 billion in its first quarter of production shipments.
That is not a demand story in trouble. If you only read the top of the release, you would wonder what anyone was worried about. So the worry had to live somewhere else. In this report, it lived in the fine print.
The $118 Billion Line
The figure that drew the most attention was not revenue. It was guarantees. According to Blockspace, whose analysis was syndicated on Yahoo Finance, and to TechTimes, Nvidia’s maximum gross exposure under data-center-related guarantees rose to $118 billion from $108.5 billion a quarter earlier.
Most of that, about $105 billion, backs roughly 4.25 gigawatts of capacity across nine buildings at SB Energy’s PORTS-Pike Technology Campus in Ohio, which is leased to OpenAI for 20 years. Each guarantee becomes active only when its building is ready for service, with the first expected in fiscal 2029. The remainder is $3.5 billion of land, power, and shell guarantees for smaller cloud operators and $9.5 billion of credit support tied to power and systems. Blockspace reported that Nvidia says it has made no payments on any of them.
Read that carefully, because both sides of the debate have a point. A guarantee is not a loss. Nothing has been paid, and most of it does not activate for years. But a guarantee is a promise to stand behind someone else’s obligation, and that is a different kind of business from selling chips. When the most important chipmaker in the world underwrites the buildings its biggest customers need in order to run its chips, investors are entitled to ask how much of today’s revenue leans on Nvidia’s own balance sheet.
Kress pushed back on the idea that any of this was new. Asked by Bernstein’s Stacy Rasgon about the conditions on the guarantees, she said: “Is it new? No. Did we change anything? No.” Fair enough. The market’s reaction suggests that familiarity is not the same thing as comfort.
Receivables, Commitments, and Who Carries the Risk
Two more numbers filled out the picture. TechTimes reported that accounts receivable climbed to $81 billion, with days sales outstanding rising to 64 from 60. The explanation offered was the timing of Rubin shipments and extended payment terms on large multi-quarter agreements. Neither is alarming on its own. Customers buying at enormous scale often negotiate longer terms. Still, when sales get booked faster than cash comes in, investors notice, especially after a year in which the financing of AI infrastructure has become a debate of its own.
Then there are Nvidia’s own obligations. Supply and capacity commitments, mostly multiyear memory purchases, rose to $390 billion from $279 billion, according to both Blockspace and TechTimes. That is Nvidia locking in the components it needs for years of production, which is a statement of confidence in demand. It also raises the stakes if demand ever cools.
There was one small sign of restraint. Blockspace reported that contingent commitments to so-called neocloud operators, the specialized companies that rent out GPU capacity, fell to $32 billion from $36 billion. Kress said Nvidia would be “disciplined and selective” in supporting them. JPMorgan, per Blockspace, flagged that new GPU financing deals could take longer or require higher compensation without Huang in the room, though the bank did not assume existing commitments would change.
The Extra Week and the Margin Step Down
The forward guidance also looked better at a glance than on a second read. Nvidia guided fourth quarter revenue to $130 billion, plus or minus 3%, and its release notes that the quarter includes an additional week. TechTimes ran the math, citing an Investing.com analysis. Spread evenly, a 13-week quarter at $115.8 billion works out to about $8.9 billion a week, while a 14-week quarter at $130 billion comes to about $9.3 billion a week. That is roughly 4% growth per week of selling, not the 12% the headline comparison implies.
The guidance also assumes no data center compute revenue from China. Kress said Nvidia shipped about $2.4 billion of licensed Hopper chips there in the third quarter and is no longer doing so, according to TechTimes and Yahoo Finance. That revenue is not coming back next quarter, and nobody should model it as if it were.
Margins are stepping down too. Gross margin was 74.8% on a GAAP basis in the third quarter, and Nvidia guided the fourth quarter to between 71% and 72%. Kress said she expects the fourth quarter to mark the bottom, with margins of 72% to 73% in fiscal 2028, per TechTimes. A business with gross margins in the low seventies is still extraordinary. But a stock priced for perfection reacts to direction, and this quarter the direction was down.
What a Careful Investor Can Take From This
None of this means the AI buildout is ending. Nvidia’s customers are still buying what it can make, and the company raised its outlook for next year. What has changed is the question the market is asking. Not long ago it was how much demand there is. Now it is who is paying, on what terms, and with whose credit.
That shift matters for anyone with AI exposure, whether through Nvidia directly or through broad index funds where it ranks among the largest holdings. The growth figures in an earnings release tell you how fast a company is selling. The guarantees, the receivables, and the purchase commitments tell you how much risk it took on to get there. Those details tend to matter most when sentiment turns, and sentiment around chip stocks has been turning for weeks.
There is a valuation wrinkle worth noticing as well. Commentary on 24/7 Wall St. described the shares as trading at roughly 12 to 13 times forward earnings after the selloff. A low multiple can mean the market is missing a bargain. It can also mean investors doubt that the forecast earnings will arrive on schedule, or at the quality the headline suggests. Which reading is right depends on how the financing structure behind the AI boom holds up, and one quarter cannot settle that.
So is Nvidia a company with a financing problem, or a company that simply sits at the center of an industry that has started to finance itself in more complicated ways? Reasonable people can disagree. What is harder to argue with is that the footnotes now move the stock as much as the headline. Investors who read them will at least know what they own.
This article is written for educational and informational purposes only and does not constitute financial or legal advice. The views and analytical frameworks presented draw on publicly available information and reported commentary from industry participants. Readers are encouraged to consult primary sources and form their own informed views on these complex topics.





