NEW YORK — For three days this week, Nvidia looked like it could do no wrong. Then Kevin Warsh walked to a podium in Wyoming.
NVDA shares fell roughly 3.4% on Friday to close near $220, retreating from the post-earnings high of $227.98 reached Thursday — when the company’s fiscal second-quarter report sent the stock up nearly 8.7% in a single session. The driver of Friday’s reversal was not Nvidia’s business, which by every measure continues to defy the expectations of analysts who specialize in setting very high expectations. The driver was the Federal Reserve.
Fed Chair Warsh used his Jackson Hole address to signal that the central bank’s appetite for rate cuts remains limited, citing persistent services inflation and a labor market he described as still “meaningfully tight.” Rate-sensitive technology stocks sold off broadly. Nvidia, which had absorbed the most optimism of any chip company this earnings season, gave back more than its peers.
The broader market context mattered: the S&P 500 fell on Friday as Warsh’s comments reset rate-cut expectations across Wall Street.
But the week’s arc should not be mistaken for a repudiation. Nvidia’s second-quarter results — reported after Wednesday’s close — were, by the standards of a company that regularly exceeds forecasts, extraordinary.
Revenue came in at $96.22 billion for the period ending July 27, a 106% increase over the same quarter a year earlier and above the $92.17 billion consensus. Adjusted earnings per share reached $2.46, a 128% gain from a year ago, clearing the $2.10 estimate that itself had been revised sharply upward in recent weeks. Data center revenue, the engine of Nvidia’s transformation from a gaming chip maker into an AI infrastructure company, accounted for roughly 88% of total sales.

| Security | Close | Change | % Change |
|---|---|---|---|
| Nvidia (NVDA) | ~$220 | ▼ | -3.4% |
| SOX (Semiconductor Index) | — | ▼ | -2.7% |
| NASDAQ 100 (NDX) | 19,840 | ▼ | -2.3% |
| S&P 500 | 5,580 | ▼ | -1.9% |
The guidance was where the real conversation started. Chief Financial Officer Colette Kress projected fiscal third-quarter revenue of approximately $108 billion at the midpoint — a figure that would represent another sequential step up and a continued pace that has made Nvidia’s trajectory nearly impossible to model. But it was Kress’s longer-term signal that produced the most discussion among analysts: she projected fiscal 2028 revenue growth of roughly 70%, against a Street consensus that had been closer to 45%.
“Demand is growing beyond 70% but is constrained by supply,” Chief Executive Jensen Huang told CNBC’s Jim Cramer on Wednesday evening, in an interview that aired on Mad Money. Huang said the company is working to close that supply-demand gap.
The supply answer, at least in part, came from a separate disclosure. Nvidia announced a $500 billion AI infrastructure financing arrangement with a consortium that includes Apollo Global Management, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs, and KKR. The deal would fund new data center construction, with Nvidia’s hardware at the center of those facilities. The sheer scale of the commitment signals that major financial institutions are no longer treating AI infrastructure as a speculative bet — they are treating it as an asset class.
Amazon Web Services separately announced it would purchase two million Nvidia GPUs and deploy the company’s new Vera CPU. AWS confirmed it would use the chips to power next-generation AI inference workloads across its global cloud regions — a contract that speaks to both the demand environment and Nvidia’s current absence of credible competition at the high end of AI accelerator performance.
For shareholders watching the buyback math, Huang offered a clear message: the company returned $26 billion to shareholders last quarter and still has $99 billion remaining on its repurchase authorization. He said Nvidia plans to increase buyback activity with next year’s free cash flow, which at the current revenue trajectory would represent a substantial capital return program.
The question Warsh’s remarks reopen is one that has trailed AI-infrastructure stocks for months: at what point does a higher-for-longer rate environment begin to affect capital budgets at hyperscalers? The top five hyperscalers — Amazon, Microsoft, Google, Meta, and Oracle — are expected to collectively spend $1.3 trillion on capital expenditures next year, up from roughly $800 billion in 2026. Most of that spending flows directly or indirectly to Nvidia. A sustained rate shock that slows corporate investment could compress that number.
Nothing in Friday’s action suggests that reckoning has arrived. The pullback was measured — semiconductor stocks as a group fell more than Nvidia, which held up better than the PHLX Semiconductor Index on a relative basis. What the day did accomplish was a reminder that Nvidia’s stock price, even at these levels, is not immune to macroeconomic signals that have nothing to do with Jensen Huang’s roadmap.
The week ultimately delivered what few companies can: a quarter that genuinely surprised a market priced for perfection, guidance that moved the long-term growth narrative forward rather than sideways, and a capital commitment from the financial industry that frames AI infrastructure as infrastructure — not as a technology cycle with a known expiration.
The rate debate will return. It has a way of doing that. But Nvidia’s problem is not demand. Its problem, as Huang said, is that demand is already outrunning what it can build.
That is not the kind of problem that a speech in Wyoming resolves.

