Nvidia can sell more chips and still disappoint investors

You can be right about the demand for AI computing and still be wrong about the return on Nvidia stock. The missing step is the share of that demand that becomes profit, followed by the price investors will pay for that profit.

Revenue is the number that makes the headline. Margins decide how much of it survives. For a company already earning unusually high margins, assuming those margins persist is a substantial part of the investment thesis.

Our position is that an NVDA valuation deserves a margin stress test even when the sales outlook looks strong. You do not need to predict an AI spending crash to ask whether future revenue will be equally profitable.

The Theory A chart puts the investment question next to the operating results: how much is the market paying for the earnings? Hover over reported periods and look for changes in the earnings trajectory. In the full Analyze page, the financial-flow visualization lets you follow revenue through costs and cash flow.

If your plan includes the Sales/Revenue comparison, switch between it and Earnings over the same range. Each view uses its own valuation multiple, so compare the underlying figures and their growth, not just the distance between the drawn lines. The reported margin figures below provide the accounting check.

The reported numbers give both sides something to work with

In fiscal 2026, Nvidia’s revenue rose to $215.938 billion from $130.497 billion. Annual GAAP gross margin moved from 75.0% to 71.1%. Sales growth and margin direction were different stories within the same year. The company’s results release reports both.

Stopping there would be selective. In the quarter ended July 26, 2026, Nvidia reported $96.221 billion in revenue and a 75.0% GAAP gross margin. The more recent quarterly result does not support a claim that margins are already collapsing.

The useful question is how durable that profitability will be. A quarterly margin, an annual margin and a normalized margin answer different questions. None should be quietly substituted for another.

What if revenue grows 40% and profits go nowhere?

Take a fictional business with 100 units of revenue and a 50% net margin. It earns 50. Next period, revenue rises to 140, but net margin falls to 35%. It earns 49. Sales grow 40%; net income falls 2%.

That example is deliberately blunt. It is not a forecast for Nvidia, and the net margins in it are not the gross margins in the table above. It isolates a relationship that a revenue-growth headline can hide.

Gross profit is what remains after cost of revenue. Net income also reflects operating expenses, financing, taxes and other items. To turn an Nvidia sales forecast into an earnings forecast, those steps need their own assumptions. To turn company earnings into EPS, the share count needs one too.

A lower margin does not automatically mean a bad outcome. Enough additional revenue can more than compensate. The point of the exercise is to find the combination your purchase price requires, rather than treating sales growth as a substitute for the calculation.

What could squeeze the profit on the next dollar?

Three mechanisms are worth modeling. Customers could gain bargaining power. A change in the mix of products and complete systems could change costs as well as sales. Competitors or customer-designed alternatives could become good enough for some workloads, putting pressure on the economics of incremental business.

These are scenarios to test, not established outcomes. A sensible stress case might combine slower growth with a moderate margin reduction. A harsher case would allow both margins and the valuation multiple to fall. They need not move independently: investors may pay less for each dollar of earnings if those earnings appear less durable.

Be equally careful with the word “temporary.” A launch cost may fade. A lasting change in product mix may not. Read the explanation for a margin movement before drawing a straight line through it.

The strongest case for margins staying high

Customers care about the economics of a working system, not just the invoice price of a chip. If Nvidia continues to lower a customer’s total cost of useful computing, it could retain strong pricing power while improving the customer’s economics. Faster hardware alone is only one possible source of that value.

That argument deserves a place in the model. It also needs evidence over time. Strong margins through product transitions, repeat orders that turn into cash and sustained customer demand at attractive economics would make a durable-margin case more credible.

The recent quarterly margin is evidence against assuming an immediate collapse. It is not a contract for the next decade. The same discipline applies to the bearish case: competition appearing in a slide deck does not prove Nvidia will lose its pricing power.

The number to watch beside revenue growth

Track how much incremental revenue becomes operating income, then reconcile operating income with cash generation. Compare like periods and consistent accounting measures. A change in non-GAAP definitions can make an apparent trend less informative than it first looks.

We would become more comfortable with a rich growth assumption if it repeatedly arrived with resilient margins and cash generation. We would become less comfortable if the sales story stayed strong while each extra dollar produced less profit and required more capital.

“Is Nvidia overvalued?” cannot be answered by pointing to a large market. The valuation has to work through to the earnings a share can claim. That is where the argument becomes testable.

For a visual explanation of the accounting, see gross, operating and net profit margins. For a related valuation mistake, read the cyclical P/E trap.