In the late summer of 2025, Nvidia once again outpaced Wall Street's expectations, reporting 56 percent revenue growth and projecting a future in which artificial intelligence infrastructure spending reaches four trillion dollars before the decade closes. Yet even this performance — the ninth consecutive quarter of growth exceeding fifty percent — was met with a moment of investor hesitation, as data center revenues fell short of estimates for the second straight quarter. The episode reveals something enduring about the human relationship with progress: we are rarely satisfied by the extraordi
Nvidia Beats Earnings Expectations, Signals 50%+ Growth Ahead Despite Data Center Slowdown
The slowest growth in nine quarters, yet still 56 percent.
Mark
Why did the stock fall if Nvidia beat expectations?
Mimi
Because data center revenue—the division that matters most—came in below estimates for the second time. Investors have gotten used to Nvidia exceeding on every metric. Missing on the one that drives the business is a warning sign, even if the overall picture is strong.
Mark
But the company is still growing at 56 percent year-over-year. That's extraordinary.
Mimi
It is. But it's also the slowest growth in nine quarters. When you've been doubling or tripling, 56 percent starts to feel like deceleration. The market is asking whether the AI buildout is slowing down, not whether Nvidia is failing.
Mark
What's the H20 situation really about?
Mimi
It's a chip designed for China that Nvidia can't sell there because of U.S. export restrictions. The company wrote off $4.5 billion because of it. Now, after Huang met with Trump, there's hope for licenses to ship it. That could add $2 billion to $5 billion in revenue this quarter alone.
Mark
So geopolitics is now part of Nvidia's earnings story.
Mimi
Completely. The company's growth is real and grounded in genuine demand. But how much of that demand they can actually capture depends on what Washington decides about China policy. That's not something Nvidia controls.
Mark
The $3 trillion to $4 trillion figure—is that credible?
Mimi
It's Nvidia's projection for AI infrastructure spending by 2030. It's not a guarantee, but it's the company's way of saying this isn't a bubble. The buildout is going to take a decade. If that's true, then even 56 percent growth is just the beginning.
Mark
What should investors watch for next?
Mimi
Whether data center revenue stabilizes or continues to miss. Whether the H20 actually gets licensed for China. And whether the company can maintain 50-plus percent growth as it gets larger. Those three things will determine whether this is a temporary slowdown or the start of a longer deceleration.
The Pulse
- Nvidia beat earnings expectations with $46.74 billion in revenue and a $54 billion Q3 forecast, yet its stock initially fell because the data center division — the heart of the entire growth story — missed estimates for the second quarter running.
- The company has now sustained over fifty percent year-over-year revenue growth for nine straight quarters, but this quarter marked the slowest pace in that streak, signaling that even the most powerful growth curves carry the weight of their own scale.
- A $4.5 billion writedown tied to H20 chip export restrictions to China removed billions in potential revenue, leaving a $2–5 billion opportunity this quarter dangling on the outcome of U.S. licensing decisions and diplomatic conditions.
- CEO Jensen Huang's direct engagement with President Trump signals Nvidia is actively lobbying to restore access to the Chinese market, treating geopolitics as a business variable it cannot afford to leave unmanaged.
- Finance chief Colette Kress anchored investor confidence by projecting $3–4 trillion in AI infrastructure spending by decade's end, framing the current buildout not as a speculative surge but as the foundation of a long structural transformation.
- Nvidia's networking revenue nearly doubled to $7.3 billion, suggesting customers are constructing entire AI ecosystems — not merely buying chips — which deepens the company's indispensability across the infrastructure stack.
In the late summer of 2025, Nvidia once again outpaced Wall Street's expectations, reporting 56 percent revenue growth and projecting a future in which artificial intelligence infrastructure spending reaches four trillion dollars before the decade closes. Yet even this performance — the ninth consecutive quarter of growth exceeding fifty percent — was met with a moment of investor hesitation, as data center revenues fell short of estimates for the second straight quarter. The episode reveals something enduring about the human relationship with progress: we are rarely satisfied by the extraordinary once we have come to expect it. Nvidia stands at the center of a genuine technological transformation, yet must also navigate the geopolitical currents that no balance sheet can fully contain.
Nvidia reported its second-quarter results on Wednesday, beating Wall Street's earnings and revenue estimates, yet its stock dipped in after-hours trading before recovering the following day. The culprit was familiar: data center revenue of $41.1 billion fell short of analyst expectations for the second consecutive quarter, a pattern that has become its own kind of story within the larger story of Nvidia's dominance.
The underlying numbers remained remarkable. Adjusted earnings per share came in at $1.05 against an expected $1.01, and total revenue of $46.74 billion exceeded the $46.06 billion consensus. The company projected third-quarter revenue of $54 billion, above analyst estimates of $53.1 billion. Overall revenue grew 56 percent year-over-year, extending a streak of nine consecutive quarters with growth above fifty percent — a run that began when generative AI first appeared in Nvidia's financials in mid-2023. Yet this quarter was the slowest of that entire stretch, a quiet reminder that exponential growth eventually meets the resistance of its own scale.
The data center division, which generated $41.1 billion, remains the company's core engine. Within it, compute revenue of $33.8 billion declined one percent from the prior quarter, a drop that finance chief Colette Kress attributed to $4 billion less in H20 chip sales. The H20, a processor designed specifically for the Chinese market, has become a geopolitical pressure point. Nvidia wrote down $4.5 billion related to the chip after export restrictions halted sales to China, and the company estimates the restrictions cost it roughly $8 billion in potential revenue during the quarter.
Following CEO Jensen Huang's meeting with President Trump, Nvidia is now signaling it expects to receive U.S. licenses to resume H20 shipments to China. Kress indicated the company could generate between $2 billion and $5 billion in H20 revenue this quarter if conditions allow — a significant caveat that places a meaningful portion of near-term growth in the hands of diplomacy rather than demand.
Beyond the geopolitical uncertainty, the broader picture Nvidia painted was one of durable, structural growth. Kress told analysts the company expects $3 trillion to $4 trillion in AI infrastructure spending by the end of the decade — a projection that frames the current buildout as the beginning of a long transformation rather than a speculative peak. Networking revenue, which covers the components that connect GPUs into large-scale AI systems, nearly doubled year-over-year to $7.3 billion, suggesting customers are investing in entire ecosystems. Net income rose 59 percent to $26.42 billion.
What the quarter ultimately illustrated is the peculiar position Nvidia now occupies: a company so central to a generational shift in computing that even its shortfalls are measured against expectations no other firm could approach. The question going forward is not whether demand is real — Nvidia's results suggest it is — but whether geopolitical forces will allow the company to fully capture the opportunity in front of it.
Nvidia delivered earnings that beat Wall Street's expectations on Wednesday, yet the stock stumbled in after-hours trading anyway. The reason was familiar: data center revenue, the engine of the company's growth, came in below estimates for the second consecutive quarter. By Thursday, shares had recovered some ground, but the tension remained—a company growing faster than almost any other on earth, yet somehow still disappointing investors who have come to expect the impossible.
The numbers themselves told a story of sustained momentum. Nvidia reported adjusted earnings per share of $1.05 against expectations of $1.01, and revenue of $46.74 billion versus the anticipated $46.06 billion. More importantly, the company projected revenue of $54 billion for the current quarter, plus or minus 2 percent, which exceeded analyst consensus of $53.1 billion. The stock, up 35 percent so far this year after nearly tripling in 2024, had already priced in extraordinary performance. What Nvidia needed to signal was not just that it had met the moment, but that the moment itself was not ending.
It did. Finance chief Colette Kress told analysts that the company expects between $3 trillion and $4 trillion in artificial intelligence infrastructure spending by the end of the decade. That projection anchors everything else: it means the buildout is not a bubble, not a temporary surge, but the foundation of a decade-long transformation in computing. Nvidia's overall revenue rose 56 percent in the quarter from $30.04 billion a year earlier. More striking still, the company has now posted year-over-year revenue growth exceeding 50 percent for nine consecutive quarters, dating back to mid-2023 when generative AI began showing up in its results. Yet this quarter marked the slowest growth in that entire stretch—a reminder that even exponential curves eventually flatten.
The data center division, which generated $41.1 billion in revenue, remains the company's lifeblood. That figure fell short of the StreetAccount estimate of $41.34 billion, continuing a pattern of missing on the metric that matters most. Within that division, $33.8 billion came from compute—Nvidia's graphics processors, the chips that do the actual work of training and running AI models. That number declined 1 percent from the previous quarter, a decline Kress attributed to $4 billion less in H20 sales. The H20 is a processor custom-built for the Chinese market, and it has become a geopolitical flashpoint.
Nvidia had written down $4.5 billion related to the H20 after export restrictions limited its ability to sell to China. The company previously estimated that if the chip had been commercially available during the quarter, it could have added $8 billion in sales. During the period, Nvidia sold no H20 chips to China, though it did release $180 million worth of H20 inventory to a customer outside China. Now, following CEO Jensen Huang's meeting with President Donald Trump, Nvidia is signaling that it expects to receive U.S. licenses to ship the H20 to China. Kress said the company could generate between $2 billion and $5 billion in H20 revenue this quarter if geopolitical conditions permit.
That caveat—if geopolitical conditions permit—hangs over everything. Nvidia's growth story is real, grounded in genuine demand for the infrastructure that powers artificial intelligence. But it is also hostage to forces beyond the company's control. The networking business, which sells the components needed to connect GPUs into the massive systems that train large language models, nearly doubled year-over-year to $7.3 billion. That growth suggests customers are not just buying chips; they are building out entire ecosystems. Net income increased 59 percent to $26.42 billion, or $1.08 per share, from $16.6 billion a year earlier.
What emerges from these numbers is a company that has achieved something rare: it has become indispensable to a transformation that is still in its early stages. The slowdown in data center revenue growth, while real, appears to reflect not waning demand but the mathematics of scale—it is harder to grow 56 percent when you are already the dominant supplier. The question now is whether Nvidia can sustain that growth, whether the $3 trillion to $4 trillion in AI infrastructure spending actually materializes, and whether geopolitical tensions will allow the company to capture the full market opportunity in front of it. For now, Wall Street is betting yes.
Notable Quotes
Nvidia expects between $3 trillion and $4 trillion in AI infrastructure spending by the end of the decade.— Colette Kress, Nvidia finance chief
Want the full story?
Read the original at CNBC ↗
Economy & Finance