Depending on which figure you stare at, Nvidia is either suspiciously cheap or a flashing warning light. It trades at about 29X earnings, a discount to the Nasdaq-100’s 34, which for a company growing revenue 56% year-on-year looks almost reasonable. Then you notice the price-to-sales ratio sitting north of 30, a level that has historically preceded some very ugly corrections, and the reasonable feeling evaporates.
Both readings are true at once, which is what makes this the most interesting argument in the market. The bull case rests on real profits ($26.4bn net income last quarter, 53% margins) and a $2 trillion cloud backlog. The bear case rests on circular financing, peak-cycle risk, and the awkward question of whether the hyperscalers pouring $800bn into buildout will ever see returns that justify it. We lay out both sides and let you draw your own conclusion. (Not investment advice; we are not your financial adviser, but feel free to buy a course if you wanna support the work we do.)
The bull case: the numbers are not made up
Start with the thing bears often skate past: Nvidia is wildly, unambiguously profitable. Last quarter it posted $26.4bn in net income on 53% net margins, with revenue up 56% year on year. This is not a 1999-style story of valuation floating free of any earnings; the earnings are real, enormous, and still growing fast. That is why the price-to-earnings multiple of roughly 29 sits below the Nasdaq-100’s 34. On the metric that actually accounts for profit, Nvidia looks cheaper than the index it is supposedly bubbling.
The forward story is bigger still. Nvidia points to a cloud-computing backlog north of $2 trillion, and expects the combined capital spending of the top five US hyperscalers to climb from around $800bn in 2026 to $1.3 trillion in 2027. Wall Street has mostly bought it: consensus price targets cluster around $306, with the bulls reaching toward $343, and the analyst tally runs heavily to “buy”. As the optimists frame it, you are not paying up for hope; you are paying a below-market multiple for the one company selling the shovels in a gold rush that is still accelerating.
The bear case: the shape of the risk
Now the other side, because it is more than a vibe. The first worry is price-to-sales above 30. That is not a made-up red line; historically, valuations at that level across large-cap tech have often preceded sharp corrections. Profitability softens the blow, but it does not repeal the pattern.
The sharper worry is circular financing. Nvidia has commitments entangled with the very customers buying its chips, including reported arrangements around OpenAI and Anthropic, which raises the uncomfortable question of how much of the demand is genuinely independent. It is worth deferring to someone who has actually run the numbers here. Yik Ban Chong of Phillip Securities, who flags the concern, also puts it in proportion: he estimates these circular deals represent “less than 10%” of Nvidia’s roughly $380bn of revenue visibility through the end of 2026. Real risk, in other words, but not the whole edifice.
Then there is peak-cycle risk, the oldest trap in semiconductors. The entire bull case assumes hyperscaler capex keeps climbing. But those same hyperscalers are under mounting shareholder pressure to show that all this AI spending produces actual returns. If the applications do not start paying for the infrastructure, the buildout could slow fast, and Nvidia sits at the exact point in the chain where a spending pause hits first and hardest. Add geopolitical and export-control risk on top, and you have a company whose fortunes depend on a lot of other people’s optimism holding steady.
So which is it?
Honestly, both cases are internally coherent, which is precisely why smart investors are on opposite sides of this trade. The bulls are right that the profits are real and the multiple is not insane. The bears are right that the demand is concentrated, partly circular, and cyclical in a way the market may be underpricing. The disagreement is not really about Nvidia’s present; everyone agrees the present is spectacular. It is about whether 2027’s numbers arrive as promised or as a warning.
What we would gently push back on is the certainty at either extreme. Anyone telling you Nvidia is an obvious buy, or an obvious short, is skipping the part where two contradictory readings are both supported by the same balance sheet. The useful posture is to hold both in your head at once and size your conviction accordingly. (Once more, for the lawyers: this is information, not advice, and we are not your financial adviser.)
Related on Top Tool Stack: the market rally as bond yields cooled.