NextFin News - Aswath Damodaran, the NYU Stern finance professor known as the "Dean of Valuation," says Nvidia investors are being set up for a painful pattern: the company can beat its numbers and still see its stock fall, because management is deliberately lowering expectations for a "soft landing" while the share price still assumes a second, unproven market.
"Investors are also increasingly focused on the guidance that Nvidia is offering, and the negative reaction, I think, is coming from the sobering of that guidance. Nvidia, I believe, is trying to bring investors down for a soft landing, lowering expectations. However, it's going to be painful. Every earnings report, you're likely to see more of this phenomenon play out—where they beat the numbers but the stock price still goes down."
— Aswath Damodaran, NYU Stern professor of finance, in a recent television interview
The warning lands as Nvidia trades near a $5.20 trillion market capitalization with a forward price-to-earnings multiple of roughly 21 times — a level that looks modest next to the broader market but still embeds a revenue requirement that Damodaran's own reverse-engineered math puts at about $500 billion today, or closer to $600 billion if investors have to wait five years for the cash flows. The tension is the story: the multiple has compressed, but the bar has not come down with it.
The Expectations Trap: Why Beating Is No Longer Enough
The mechanism Damodaran is describing is not new, but it sharpens as valuations stretch. At ordinary multiples, a company beats estimates and the stock rises. At extreme multiples, the market has already priced the beat; the only surprise that moves the needle is a raise in what comes next. A company can report record revenue, record earnings, and a record stock decline — and leave observers asking what the market wanted.
Nvidia's own guidance illustrates how high that bar has become. For the second quarter of fiscal 2027, management guided revenue to approximately $91 billion, plus or minus 2 percent, with non-GAAP gross margin of 75 percent and non-GAAP operating expenses of about $8.3 billion. Wall Street's consensus sits just above that guided midpoint — roughly $91.9 billion to $92.2 billion in revenue and about $2.09 in earnings per share — which means the company must not merely meet its own already-aggressive forecast but exceed it, then guide beyond it, to satisfy a multiple that assumes continued hypergrowth.
Investors have seen this movie before. In late August 2024, Nvidia reported stratospheric revenue and earnings above expectations, and the stock still fell about 8 percent the next day, wiping more than $200 billion from its market value. Damodaran wrote about that episode at the time, using it to dissect the gap between what an earnings report says and what the market hears. The lesson was not that the business had weakened; it was that the expectation embedded in the price had outrun even a good report.
That is the practical meaning of Damodaran's "soft landing" framing. When management guides conservatively, it is not signaling weakness so much as trying to reset the bar at a reachable height. The problem is that a lower guide, even from an elevated base, reads to the market like deceleration — and deceleration is what a premium multiple cannot tolerate.
The Real Objection Is Not the Multiple — It Is the Second Market
It would be a mistake to read Damodaran's caution as a call that Nvidia's P/E is absurd. On the surface, the valuation metrics have been compressed by earnings growth that has far outpaced the share price. Trailing twelve-month revenue stands at $253.5 billion, net income at $159.6 billion, and earnings per share at $6.53. Gross margin is 74.2 percent, operating margin 64.0 percent, and profit margin 63.0 percent — a profitability profile that resembles a software monopoly more than a semiconductor company. Free cash flow over the same period was $119.1 billion, against $12.8 billion of debt and $53.2 billion of cash.
Against those numbers, a forward P/E near 21 times looks defensible, even cheap. Nvidia's forward multiple trades near its lowest levels in years and sits close to the S&P 500's forward multiple of about 23 times, despite Nvidia growing earnings many times faster than the index. On a purely mechanical reading, the stock is not priced for perfection — it is priced for strong growth that the company has repeatedly delivered.
So what is Damodaran's objection? It is not the current earnings. It is the bridge from here to the market capitalization. In his valuation framework, he works backward from a market cap to the revenues a company must eventually produce to justify it, given an assumed margin and cost of capital. Using Nvidia's roughly 53 percent net margin and an 8 percent cost of equity, he estimated that a $5 trillion company would need about $483 billion in annual revenue if those cash flows arrived today — and roughly $678 billion if investors must wait several years for the company to grow into them.
He summarized the same calculation in December 2025: "I use Nvidia, at the $5 trillion market cap, and assuming that the company can maintain its current nosebleed net margin and return on equity, to estimate breakeven revenues of about $500 billion (if delivered right now) or about $600 billion if you have to wait five years." He added that the breakeven figure is highly sensitive to margins: if profitability falls toward industry norms, the required revenue climbs toward $1 trillion or higher.
Measured against Nvidia's current $253.5 billion revenue run rate, that breakeven target requires roughly doubling the business from here — while holding margins at levels no semiconductor company has sustained at scale. This is where Damodaran's "second market" point bites. In the same interview, he put it directly: "I think at this price, if you're buying, you're expecting the product and service market to be much bigger than people anticipate... Even if you assume Nvidia's dominance in the AI chip market — which I believe they currently have — and their ability to maintain these sky-high margins, you still can't justify the valuation without something additional happening. It's almost as if you're betting on Nvidia finding and dominating another market, and that's a tough call."
He then applies what he calls the three-P test to any breakeven revenue estimate: is it possible, is it plausible, and is it probable? On Nvidia, his judgment is that the breakeven revenue passes the first two tests but fails the third. That is a narrower claim than "Nvidia is a bubble." It is a claim that the most likely outcome falls short of what the price requires — and that when the gap between probable and required becomes visible, the stock reprices even on good news.
Cyclical or Structural: The Question That Determines the Answer
Every valuation argument about Nvidia collapses into one judgment: is the AI infrastructure buildout a cyclical capex wave that will overshoot and correct, or a structural shift in how the economy computes? Getting this wrong flips the conclusion, because a cyclical peak deserves a cyclical multiple while a structural platform deserves a premium that compounds.
The cyclical case has history on its side. Semiconductors have been a cyclical industry for four decades, with revenue growth that averaged single digits outside of boom periods. Capex waves in chips overshoot, inventories build, and prices fall. Nvidia's own past carries a cautionary marker: the 2018 crypto-mining hangover sent its stock down more than 50 percent before the AI era rescued it. The current cycle has its own concentration risk. Large cloud service providers represented approximately 50 percent of Nvidia's data center revenue each quarter through fiscal 2026, and those customers are developing their own AI accelerators and collaborating with competitors. Google's TPU, Amazon's Trainium, and Microsoft's Maia are not theoretical threats; Broadcom has announced a custom chip for OpenAI. If even a meaningful slice of training and inference workloads migrates to in-house silicon, the volume assumption behind the breakeven revenue weakens.
There is a counterweight inside Nvidia's own numbers. The company has said roughly 40 percent of its revenue now comes from customers outside the large cloud providers — enterprise, industrial, sovereign AI, and infrastructure partners — which broadens the demand base beyond a handful of hyperscalers. That diversification is one pillar of the structural argument.
The structural case is equally serious. AI datacenter spending is infrastructure, not a consumer gadget cycle: models grow larger, inference demand compounds with usage, and every enterprise workflow is a candidate for automation. Nvidia's gross margins of 73 to 75 percent at a $250 billion revenue scale are unprecedented for a chip company and look more like a platform with pricing power than a cyclical supplier. The CUDA software ecosystem creates switching costs that pure silicon competitors have struggled to breach. Management has laid out a product roadmap extending through Rubin Ultra in 2027 and Feynman in 2028, and Chief Financial Officer Colette Kress has pointed to annual AI infrastructure spending reaching $3 trillion to $4 trillion before 2030. If that buildout materializes, Nvidia's current revenue is the beginning, not the peak.
The cleaner way to separate the two forces is to assign each to a different variable. The demand leg — the total compute being installed — is probably structural. The multiple leg — the price investors will pay for each dollar of Nvidia's earnings — is cyclical, because it is set by sentiment, liquidity, and the perceived certainty of growth. The risk is not that AI demand disappears. It is that the market pays a structural multiple for a company whose growth rate is reverting toward cyclical norms, and that the two repricing events do not arrive at the same time.
Damodaran's own behavior reflects that split. He has owned Nvidia since 2018, well before the AI boom, and has said the stock's price has often "run ahead of its value," particularly in the last three years. He completed his exit from the position at the end of last year, selling in stages over four years, while retaining Microsoft — a company he describes as needing less to happen to justify its valuation because its cloud business resembles a utility. That is not a repudiation of Nvidia's business. It is a judgment about the margin of safety at the price.
The Strongest Counter-Thesis, and What Would Break It
The bull case against Damodaran's caution is financially coherent, and it deserves weight. Nvidia's forward P/E of about 21 times sits near the S&P 500's forward multiple despite earnings growth that has raced far ahead of the index. Earnings have grown far faster than the share price, producing a PEG ratio well below 1 — the classic signature of growth at a reasonable price rather than speculation. Consensus expects fiscal 2027 revenue of roughly $393 billion, up from about $216 billion in fiscal 2026, which implies growth of more than 80 percent still being priced into the stock. Return on equity is 114 percent, return on invested capital 105 percent, and the company sits on $40.4 billion of net cash. This is not a company surviving on narrative; it is converting AI demand into cash at a rate the market has rarely seen.
Nor is Damodaran's skepticism a blanket rejection of AI. He distinguishes Nvidia from the purest AI-hype names and has kept Microsoft in his portfolio. His framework explicitly leaves room for disagreement: he does not claim a monopoly on the truth, and he argues investors should be willing to change their minds in the face of data. The bull case, in his own terms, requires only that Nvidia's breakeven revenue prove probable rather than merely plausible — that margins hold, that custom silicon stays contained, and that the AI factory buildout reaches the trillions.
But that bull case carries a specific vulnerability: it requires execution without a single meaningful miss. A multiple that assumes continued 80 percent growth and 73 to 75 percent gross margins has no tolerance for the normal friction of a maturing business — a guide that merely meets, a margin that dips on product mix, a hyperscaler that shifts more workloads in-house. The counter-thesis is strongest on the path and weakest on the destination: the nearer the company gets to the breakeven revenue, the more the stock should rise; the further the growth rate falls from the assumed path, the more violent the repricing, because the market will have learned that "probable" was priced as "certain."
That leads to a falsifiable signal. If Nvidia reports two consecutive quarters of datacenter revenue growth below 40 percent year over year while gross margin falls below 70 percent, the structural-growth-at-a-premium thesis breaks, and Damodaran's "improbable" call is vindicated. The opposite signal is equally specific: if gross margin holds above 73 percent and datacenter revenue growth stays above 60 percent for four consecutive quarters, the breakeven revenue moves from improbable to probable, and the current multiple will look like a bargain in retrospect.
What Comes Next: Three Horizons, Three Scenarios
The near-term trigger is the fiscal second-quarter report due after the market close on August 26, 2026. The number that will matter most is not whether Nvidia beats the roughly $92 billion consensus — it is whether management raises the bar beyond its own $91 billion guide and whether gross margin guidance holds at 75 percent rather than drifting toward the 73.5 percent non-GAAP level the company flagged for the following quarter. Guidance, not the print, is the variable Damodaran flagged, and it is the variable the market will punish or reward.
Beyond the earnings print, three horizons matter:
- Short term (sentiment and guidance): volatility around each quarterly report, with the stock reacting to the raise-or-not decision more than the headline beat. A "soft landing" guide that lands too softly will be read as deceleration.
- Medium term (fundamentals): the fiscal 2027 revenue path toward the roughly $393 billion consensus, and whether gross margin compresses toward 70 percent as product mix shifts and competition intensifies, or holds in the mid-70s.
- Long term (structural): the breakeven test — whether Nvidia can grow revenue toward $500 billion to $600 billion while sustaining margins that no semiconductor company has held at that scale.
Three scenarios follow from those horizons. In the base case, AI spending continues to grow but decelerates toward 30 to 40 percent annually, margins compress modestly, and the forward multiple holds in the low 20s; the stock delivers mid-single-digit to low-double-digit annual returns as earnings grow into the price. In the upside case, the AI factory buildout approaches the $3 trillion to $4 trillion annual spending that management has cited for 2030, custom silicon stays contained, margins hold, and the multiple re-expands as the breakeven revenue becomes probable. In the downside case, hyperscaler capital expenditure slows, in-house silicon captures a larger share of workloads, gross margin falls below 70 percent, and the forward multiple compresses toward 15 times — a drawdown of a third or more even if revenue keeps rising.
The beneficiaries and the exposed are not symmetric. If the multiple holds, the reward is participation in one of the most profitable growth stories in market history. If it compresses, the pain is concentrated in Nvidia holders and in the funds that have become the default way to own the AI theme. The asymmetry cuts the other way for investors who have already taken profits: Damodaran's staged exit over four years is itself a lesson in not requiring perfection from a position that has already paid.
The market is no longer asking whether Nvidia will grow. It is asking whether growth alone can justify a half-trillion-dollar revenue requirement — and whether "possible and plausible" is enough when the price has already been paid.
Explore more exclusive insights at nextfin.ai.
