NVIDIA heads into its fiscal third-quarter report with a business that keeps getting bigger and a stock that keeps getting cheaper. Three months ago, management delivered its most confident call yet. It guided revenue sharply higher, introduced an explicit framework of roughly 70% growth for fiscal 2028, and said demand was accelerating even at the company's enormous scale. The market has since marked the shares down nearly 20%, trailing the S&P 500 by about 18 percentage points. The reason is the one blemish on that call: the first reset of gross-margin expectations in more than a year, driven by memory costs that rose faster than NVIDIA anticipated. This report, due after the close on November 17, will show whether investors overreacted to that margin story or correctly sensed that the cost of growth is rising.
Wall Street expects earnings of $2.49 per share on revenue of about $109 billion. That would be a 91% revenue increase from the $57 billion reported a year ago, with earnings nearly doubling from $1.30. The revenue estimate sits just above the $108 billion midpoint of management's $105.8 billion to $110.2 billion range. The profit estimate is more demanding. At $2.49, consensus is already above the top of NVIDIA's $2.32 to $2.47 earnings guidance, and the Earnings Whisper of $2.60 sets an even higher bar. In short, the Street is not just expecting NVIDIA to hit its plan. It expects the company to beat its own profit guidance comfortably, even as margins are supposed to be compressing.
That tension makes gross margin the most important figure in the release. Management guided to roughly 74% this quarter, down from 75%, with a trough of 71% to 72% in the fourth quarter before settling at 72% to 73% next year. A result at or above 74%, paired with reassurance that the fourth-quarter trough has not deepened, would suggest memory inflation is a manageable cost of doing business. Any sign that memory pricing has pushed the trough lower, or extended it further into fiscal 2028, would challenge the idea that the margin hit is a temporary side effect of a demand surge.
On the growth side, the evidence to look for is execution on Vera Rubin. Management said the platform entered full production in August and should contribute about a fifth of data center revenue this quarter, with orders from every major buyer. Data center sales have climbed from $51 billion to $89 billion over the past year, so another solid step-up toward the high end of guidance, along with confirmation that Rubin hit its share, would support the claim of the fastest product ramp in the company's history. Updates on the expanded AWS deployment, standalone Vera CPU shipments and early volumes of the Grok 3 LPX system would show whether last quarter's new catalysts are turning into revenue on schedule.
The balance sheet deserves equal attention. Inventory rose to $32 billion ahead of the Rubin launch, receivables stretched to 60 days on extended terms, and roughly $50 billion of investments in frontier AI labs have fueled concerns about circular financing. Inventory that falls as Rubin ships, and stable collection times, would ease those concerns. Further growth in either, alongside more balance-sheet support for customers, would add to them.
Sentiment has moved from modestly bullish before last quarter to modestly bearish now, and the stock sits well below its 200-day moving average of about $205, only about 5% above its post-earnings low of $170. That setup lowers the emotional bar while the numerical bar stays high. The central question is whether NVIDIA can show that its demand acceleration is still outrunning its rising costs. If Rubin lands as promised and the margin trough holds, the narrative from last quarter remains intact. If memory costs keep moving the goalposts, the market's skepticism will look well founded.