Wall Street analysts are staying bullish on Nvidia. Traders aren’t

Perhaps the easiest way to understand Nvidia $NVDA’s latest quarter is to picture two rooms.

In one room: analysts, doing what they’ve done for years now — updating models, pushing price targets higher, and treating Nvidia’s numbers like a recurring proof that the AI boom is still real, still big, and still very hungry.

In the other room: traders, staring at the same print and asking a different question. Not “Did Nvidia beat?” Of course not; a beat is practically muscle memory at this point. The question is whether the whole AI economy — the hyperscaler spending spree, the venture-funded model factory, the “we’ll monetize later” product roadmap — is starting to look less like a growth story and more like a balance-sheet dare.

Nvidia handed Wall Street a record quarter anyway. 

Revenue for Q4 FY26 came in at $68.1 billion, with data center revenue of $62.3 billion and GAAP EPS of $1.76. The company forecast $78 billion in revenue for the current quarter, while explicitly not assuming any data center revenue from China in that outlook.

And then the stock fell. About 5% Thursday morning, a response that made (another) record quarter feel almost beside the point. The company did what it needed to do. The stock acted like it needed something else: a reason to believe the next phase of the AI buildout is going to turn into cash on a timetable that Wall Street can live with. 

So yes, Nvidia’s latest quarter looked superlative. And yes, the reaction still looked skeptical. The company’s earnings did what they’ve been trained to do in this era: dwarf the last “dwarf,” widen the gap between the company and the rest of the semiconductor universe — and still leave the stock trading like a referendum on the AI buildout. Because Nvidia has become the quarterly witness for an AI economy that is now being treated less like a market revolution and more like a capital allocation problem with a very expensive supporting cast.

And when Nvidia’s quarter is the clean part, the rest of the conversation gets loud fast — analysts, strategists, columnists, ETF shops, bears with a single scary datapoint, and long-term bulls who sound like they’re talking to the short-term tape from down the hall and through a closed door. So the post-Nvidia-earnings result is a split-screen: fundamentals that still look like a machine, and a stock that trades like the machine has to prove it can run without subsidies from investor patience.

Strong results, stubborn doubt

Investors have been trained to expect Nvidia to be spectacular. Now, they’re punishing it for being familiar.

The stock being down “underscores that investors’ long term growth concerns remain,” wrote Deepwater Asset Management managing partner Gene Munster, even as he conceded the obvious: “The fundamentals and outlook came in better than elevated whisper numbers.” That’s Nvidia’s new trap. When “better than whispers” is the baseline, the market starts grading the story around the quarter instead of grading just the quarter itself.

Kevin Cook, senior stock strategist at Zacks Investment Research, said the market didn’t even have the right setup coming into earnings. Nvidia shares “were NOT priced for perfection,” he said, “and yet the company delivered perfection and then some — which isn’t easy at their scale.” 

The market asked for an A-plus, the market got an A-plus, and the market still started asking for a higher grade.

Nvidia CEO Jensen Huang’s job, now, is to turn that surrounding story into something that feels inevitable rather than exciting. On the earnings call, he reached for the cleanest possible claim: that “the demand for tokens in the world has gone completely exponential.” He also offered the governing equation he wants investors to repeat to themselves when they start worrying about budgets: “In this new world of AI, compute equals revenues … I am certain at this point that we’ve reached the inflection point.”

The market has heard the speech before. This time, it treated it like testimony, not theater.

Richard Clode, a portfolio manager at Janus Henderson, wrote that the reason the reaction can look icy even when the numbers look molten is because “the debate has shifted away from near-term results and toward the sustainability of AI capex spending, amid concerns around its quantum, monetisation and potential cashflow degradation.” Joseph Moore, the managing director and head of U.S. semiconductors at Morgan Stanley $MS, said the numbers were at the high end of expectations, the stock reaction was muted, and the reason is that “the bigger debates holding the stock back are longer term in nature.”

Traders are no longer buying the latest quarter; they’re buying the durability of the spending cycle behind it.

A smaller implied move doesn’t mean investors expected a small quarter. It means investors expected a big quarter — and were already looking for the next doubt to pop up. And that’s how Nvidia ends up in an absurd posture: delivering superlatives while still being treated as a proxy for everyone else’s profitability.

Analysts are underwriting Nvidia; traders are pricing the whole AI economy

On the analyst side, the posture is still broadly constructive. The notes argue about the reasons, but they rhyme on the conclusion: Demand is still there, the economics still hold, and Nvidia still looks positioned to keep taking the largest bite out of the AI infrastructure cycle.

BNP Paribas’ David O’Connor kept an Outperform rating, and bumped his target to $270. The results were “a positive,” he wrote, yet “SBC adjustments, a one-time impact,” slightly “t[ook] the edge off a stellar quarter.” In this environment, “stellar” comes with footnotes. William Blair framed the quarter as operational proof that the cycle continues. “Nvidia reported a robust finish,” analysts wrote, and the results “largely affirm our view that the company remains the major beneficiary of the ongoing AI investment cycles.” That angle is about continuity. The buildout is still building, and the beneficiary still looks like the beneficiary.

Wedbush, as usual, went maximalist; its note arrived in inboxes as: “Nvidia Delivers Blockbuster Quarter/Guide; Like Watching Jordan With the Bulls.” Dan Ives wrote, “To put it bluntly...this quarter, guide, and conference call gave tech investors everything they wanted in a present with a red bow,” and he treated bubble talk as a category error. “Anyone that calls this an AI Bubble is missing the pure numbers and capex dollars being put towards the datacenter, GPU, use cases, and infrastructure piece of this AI Revolution,” he said.

The quarter’s results were “superlative,” said Asit Sharma, a senior investment analyst and lead advisor at The Motley Fool, who then pointed to a detail that suggests customers are buying deeper into Nvidia’s stack: The scaling of the networking business ($11 billion quarterly run rate, 3.5x growth year over year) is a “clear sign that companies understand the value proposition of buying as much of the data center (factory) as possible from Nvidia.” That’s a lock-in story told through line items, not adjectives.

Sharma says the stack is turning into a cost argument, not just a performance argument: “the confluence of networking, rack scale systems, CPUs (Vera) that interconnect with GPUs (Rubin), and CUDA are providing a lower total cost of ownership for inference versus increasingly formidable competition.” He also folded in some of Nvidia’s favorite phrasing — “higher performance per watt” and “lower cost per token” — because that’s the language the company uses when it wants the market to see capex as rational rather than reckless.

Another analyst made the same point with a slightly different emphasis. “Nvidia’s answer came through integration — silicon, NVLink, networking, and full systems working together toward an even more robust platform offering,” said Thomas Monteiro, a senior analyst at Investing.com, who flagged the datapoint that sits at the center of most bull cases: “The roughly 75% gross margin remains arguably the most important datapoint.” When investors worry about commoditization, margins become the lie detector.

There’s also a strategic reason Nvidia and its boosters keep dragging the conversation toward inference. It’s maybe where the company can most credibly connect compute demand to business value, and it’s where the hyperscalers can most plausibly argue they’re building revenue engines rather than monuments. Sid Sheth, the CEO of d-Matrix, said that “the most important takeaway from Nvidia’s earnings call was ‘inference equals revenue,’” which is “another strong signal that the center of gravity is shifting from training models to running them.”

In 2026, “analysts staying bullish” is less about celebrating a quarter and more about defending an ecosystem thesis: the spend remains large, the spend remains necessary, and Nvidia remains central to the spend.

Traders, meanwhile, are doing a different job. They aren’t underwriting. They’re stress-testing.

What turned skepticism into selling

Nvidia earnings aren’t just Nvidia earnings anymore. They are “the quarterly reaffirmation of the health of the AI trade,” wrote Jake Behan, Direxion’s head of capital markets, with “extensive ripple effects” that have an impact on “hyperscalers, semiconductor companies, struggling software names, the NDX, SPX, and the list goes on.” Nvidia is being asked to certify the business model of everyone buying its products.

He added that “perhaps no quarter has had more riding on it than this one.” The bellwether burden is heavy enough that the stock ends up carrying questions it can’t fully answer. And that leads to a money-now versus money-later split. “The AI trade needed some positive news and Nvidia’s earnings report brought plenty of it,” he said. “The problem is that this remains a money-later trade and market participants are very demanding of the money-now trade at the moment.” 

But “money-now” doesn’t only mean “prove payback.” It also means “prove discipline.” Yvette Schmitter, the CEO of Fusion Collective, called the quarter exceptional, then said that the “networking explosion” suggests Nvidia is tightening its grip on the stack. She also framed capital returns as a signal in a market demanding nearer-term payoff: “NVIDIA generated $35 billion in cash during the last quarter, but returned only $4 billion to shareholders.” Even investors who would maybe disagree with her conclusion can recognize the question behind it: What happens to the cash when growth is assumed?

That’s the tension that traders are pricing. Investors are worrying that Nvidia will keep prioritizing reinvestment in the AI ecosystem over returning cash to shareholders. Nvidia gets paid now. The broader AI economy is still making the case that, for the rest of them, “now” will arrive “soon.”

There’s also a broader spillover effect that matters for why Nvidia’s quarter is being treated as an ecosystem exam. Some of the market’s anxiety is about the software layer, where “AI revenue” can grow quickly without lifting total growth in a way that feels durable. AI can be real and still show up as substitution and cannibalization before it shows up as net-new acceleration. Michael Burry wrote in his “Cassandra Unchained” Substack that Nvidia’s purchase obligations reflect a shift toward “non-cancellable purchase orders well before demand is known.” The plumbing of the buildout is changing, and that plumbing can make the next downcycle uglier if buying slows.

The market’s skepticism isn’t necessarily irrational. Nvidia has a convincing argument that inference demand is early and expanding, but the short-term investor fears stability. Markets love durable trends. Markets hate lumpy ones.

That fear — deceleration, not a collapse — is why the stock can trade down even as analysts revise models upward. It’s why even bullish notes keep slipping in acknowledgments about long-term debates and cash-flow scrutiny. Nvidia can deliver a quarter that looks like a spreadsheet hallucination, and the market can still ask, essentially, what the next chapter is beyond the same story with even bigger numbers. 

You don’t have to fetishize charts to acknowledge that once a stock breaks below certain lines, a different set of traders shows up — the ones who don’t care about your bull case, because they’re trading a signal. Nvidia’s drop dragged peers lower. Key technical levels broke. The broader market didn’t catch a bid. Those mechanics matter because they explain how skepticism becomes a decisive move rather than a shrug.

Capex questions are behind a lot of the selling — can hyperscalers keep spending at this pace without investors revolting about margins and cash flow, and can they show enough monetization progress to keep the whole buildout from turning into a patience contest? That’s where the expectations treadmill meets the capex road. Traders are staring at the hyperscalers’ spending plans and asking: What does a “cool off” look like — and when will it arrive?

Traders aren’t grading Nvidia’s quarter against last quarter. They’re grading Nvidia’s quarter against the payback story of the customers who fund it — and against a market mood that has started asking for near-term proof from long-term buildouts.

So the split-screen holds.

Analysts are staying bullish because the numbers still support the core thesis: Demand hasn’t cracked, margins still point to pricing power, and the stack looks stickier over time. Traders aren’t because the trade has matured into scrutiny, and scrutiny brings deadlines. The market wants to see the money now, wants to see the capex cycle look sustainable under cash-flow pressure, and wants to see the rest of the AI economy act less like a sci-fi promise and more like a set of businesses that can fund their own ambitions without turning the whole market into a patience contest.

Nvidia delivered a monster quarter. The market is still discounting the economy that’s supposed to make the quarter repeatable — smoothly, for years, and with cash showing up sooner than later.

Nvidia delivered a monster quarter. The market is discounting whether the ecosystem behind it can deliver one just as smooth — and just as profitable — without relying on patience as its primary fuel.