Nvidia’s next earnings may have the AI boom riding on them

The air around the tech market has gone strangely heavy — that charged quiet before a weather shift — and everyone seems to be waiting for Nvidia to break it. Over the past two weeks, stocks have sagged, nerves have frayed, and the AI boom that once felt invincible suddenly looks vulnerable. Nvidia’s earnings next week, for the third quarter of its 2026 fiscal year, have become the moment investors are treating like the clearing storm: either the sky opens back up, or the forecast gets a whole lot darker.

It’s a strange position for a company to occupy, especially one reporting numbers on a timeline as routine as the earnings calendar. But Nvidia hasn’t been a normal company in a very long time. It has become the market’s proxy for the entire AI buildout — a stand-in for hyperscaler capex, sovereign spending, enterprise ambition, and investor psychology. If the AI cycle is the narrative of the moment, Nvidia is the protagonist investors keep returning to, anxious to see whether the plot still moves.

That tension colors everything leading into Wednesday. Wall Street is expecting another towering quarter: Per Zacks, that means revenue around $54.6 billion, EPS roughly $1.24, with overall revenue still barreling ahead in the mid-50% range year over year, driven largely by data centers. Nvidia guided this quarter to around $54 billion months ago, excluding any contribution from China’s H20 chips, and no one has seriously floated the idea that Nvidia will miss. Instead, the Street is testing the edges of its own imagination — wondering whether numbers this big can stay this big, and what it means if they can’t.

What the Street expects

If the consensus looks oversized, the analyst models circling it look almost surreal. 

Oppenheimer just bumped its price target to $265 (from $225) and sketched out a quarter at $55 billion — the kind of glide path most companies would present as a decade-long ambition. Their note zeroes in on the drivers: the shift to GB300 Ultra, surging demand for the rack-scale NVL72 system, and what they describe as “insatiable AI appetite” across hyperscalers. They also pointed back to CEO Jensen Huang’s projection that Blackwell and Rubin could generate $500 billion in cumulative revenue by the end of 2026, serving a $4 trillion addressable market.

Wells Fargo matched that $265 target (from $220) but went further on the long-term math. They now model $209 billion in revenue for FY26, $301 billion for FY27, and nearly $383 billion for FY28 — growth curves that would have been dismissed as fantasy a year ago. Their logic is that hyperscalers haven’t slowed down, and until they do, Nvidia’s numbers are still too low.

Citi has taken a different angle, zooming out to the capex architecture. They now expect global AI infrastructure spending to exceed $2.8 trillion by 2029, with hyperscalers pouring nearly $490 billion into AI capex by the end of 2026. In their view, Nvidia’s out-year EPS has 2-8% upside baked into the cycle, and none of it requires a heroic assumption.

And then there’s Gene Munster, whose optimism comes with its own gravitational force. He argues the Street’s 2026 estimates remain “too conservative,” pointing out that Blackwell and Rubin revenue could easily run 10% ahead of Wall Street models, especially with sovereign AI programs stepping into the bidding.

Investors aren’t showing up next Wednesday to see whether Nvidia can deliver a blockbuster quarter. Wall Street is already modeling that. They’re tuning in to see whether the tone of the call — the nuance, the calibration, the underlined phrases — matches the audacity of those forecasts. Nvidia has made a habit of confirming its own mythology. This quarter carries the burden of proving that mythology can outlast a market wobble.

Still, the mood is different now than it was even a month ago. The Nasdaq has slipped, AI stocks have shed billions, and analysts have shifted from open-throated optimism to something closer to guarded confidence. The Street wants vision, but it also wants reassurance. And Nvidia sits in the middle of all of that contradiction — the company’s expected to lift the market back into conviction, even as the market has started whispering its doubts. That means Nvidia is being relied upon to post another monster quarter and narrate a world where that momentum can keep going.

The numbers that could move the stock

The trouble with a quarter this strong is that the headline results almost don’t matter. Nvidia can clear a $55 billion bar and still make investors nervous if the commentary hints at friction. And the friction is real.

Data-center operators have begun talking openly about power ceilings — not just in Asia or Europe, but in major U.S. regions where grid expansion can’t keep pace with demand. Bringing a gigawatt online is an infrastructure project, not an engineering one. Recent earnings suggest that Big Tech has become Big Power. Hyperscalers have already started staggering deployments to match availability. Goldman Sachs estimates that global power demand from data centers could jump by about 50% by 2027 and as much as 165% by the end of the decade, driven largely by AI workloads. McKinsey, looking specifically at AI-ready capacity, pegs the growth rate at roughly 33% a year through 2030.

That gap between how fast chips can ship and how slowly gigawatts come online is already showing up in earnings. Nvidia doesn’t control that dynamic, but it lives inside the revenue cadence that results from it.

Inside Nvidia’s own world, the questions are more specific. Blackwell is supposed to be the backbone of the next leg of growth, and the GB300 platform remains the centerpiece of almost every bullish model on the Street. Stifel’s Ruben Roy has tried to keep the focus on the order book — his supply-chain checks still point to GB300 orders ramping into year-end “even as sustained GB200 demand continues” — but investors want to hear from Nvidia whether “ramping” still means accelerating, or simply holding a very high level.

China sits slightly off to the side of these operational issues but adds its own volatility to the story.  Washington briefly reopened the door for H20 exports in September, then turned around and reportedly moved to block sales of B30A — Nvidia’s Blackwell-based, export-compliant chip designed specifically for China — before it really ramped. Beijing, meanwhile, has told state-funded data centers to stop using foreign AI chips altogether and to strip them out of new projects that are less than 30% complete.

When Nvidia guided this quarter to that $54 billion in revenue, it told investors to treat any Chinese H20 sales as upside to that forecast. And as much as Nvidia wants to find its way back inside China’s market, the company’s business in the country right now, Huang said, is down to zero. Trump has said the company can’t sell its best AI chips to Beijing. For the Street, that means China functions less as a growth catalyst and more as a stress test. Any signal that demand there is stabilizing — or that B30A is gaining traction without inviting fresh regulatory trouble — will be read as free optionality.

And all of these threads tie back to the question that animates every AI debate on Wall Street: Will hyperscalers keep spending at vertical velocity? Amazon, Microsoft, Meta, Alphabet, and Oracle have turned AI infrastructure into the largest capex wave in modern corporate history, spending staggering amounts of money — hundreds and hundreds of billions — without slowing. There have been some hiccups, but the Street still largely believes in that wave. What it wants from Nvidia is confirmation that the water level isn’t dropping.

Nvidia won’t offer capex forecasts, but the way it talks about customer demand, buildouts, and future capacity will tell analysts whether their long-term models are still too conservative or suddenly too bold.

That’s why Wednesday feels heavier than it should. Nvidia’s earnings call will take on a dual role — a financial update and a sentiment reset — and the market is already treating it that way. A clean quarter will be treated as an exhale moment. A cautious guide — even a nuanced one — could be treated as the first signal that the supercycle is shifting its pace. Investors don’t want perfection; they need continuity — proof that the AI cycle still has altitude, that the clouds overhead are temporary, that the weather is shifting in their favor.

Nvidia’s earnings next week won’t settle the AI debate. But they will reset the atmosphere around it. This is the company that has shaped the past two years of market psychology more than any other. On Wednesday, Nvidia’s job next week is to show that the story is still expanding — not just in the numbers, but in the world those numbers describe.