Nvidia can beat earnings and still lose the AI trade
Nvidia can validate AI demand with a strong earnings report without proving that every AI infrastructure stock deserves the same enthusiasm. The post-earnings test is whether spending is broadening into profitable networking, custom silicon, and foundry suppliers—or remaining concentrated in Nvidia’s ecosystem.
Nvidia does not need to miss earnings for the AI trade to lose momentum. A beat on Aug. 25 could confirm that hyperscaler demand remains real while still triggering a rotation away from the most crowded expression of that demand. The reason is simple: investors are no longer asking only whether AI spending exists; they are asking who captures the durable profits as the buildout moves beyond GPUs.
The setup itself shows how high the bar has become. Options were pricing a 5.4% Nvidia move after earnings, equivalent to roughly a $280 billion swing in market value. That is a large reaction, but it is below the 6.5% move implied ahead of the May report and below Nvidia’s 7.4% average post-earnings move over the last 12 quarters. The market is prepared for a headline number. It is less clear that a headline number alone will be enough to create fresh enthusiasm.
That distinction matters because a beat can validate the category while failing to validate the current positioning. Nvidia is already a $5.10 trillion company, with revenue growth of 65.5%, EPS growth of 66.0%, and a 63.0% net margin. Those are exceptional figures, but they also create a demanding comparison point for every incremental dollar of AI spending. If the report confirms that demand is strong but offers no evidence that the profit pool is expanding beyond Nvidia’s platform, investors can reasonably conclude that the AI trade is real while the Nvidia trade is too concentrated.
Nvidia bulls have a credible answer. The company is not merely selling accelerators; its ecosystem, product cadence, and networking exposure make it the organizing platform for the buildout. Its prior quarter delivered $46.7 billion of revenue, up 56% year over year, while Blackwell data-center revenue rose 17% sequentially and data-center networking revenue increased 79% year over year in the first half of fiscal 2026. A beat-and-raise could therefore pull the entire group higher if it shows that the spending cycle is still accelerating. But that argument answers whether Nvidia remains central, not whether it remains the only stock that deserves premium treatment.
The more useful comparison is with suppliers that monetize different bottlenecks. Broadcom’s revenue growth is slower than Nvidia’s at 23.9%, but its EPS growth is 286.6% and its net margin is 38.8%. That combination points to a different kind of AI exposure: custom silicon, connectivity, and operating leverage rather than direct dependence on one GPU platform. TSMC adds another layer. Its revenue growth is 33.0%, EPS growth is 44.3%, and net margin is 50.4%, with a P/E of 30.14 versus 36.05 for Nvidia. If AI demand is broadening into manufacturing capacity and custom architectures, these are not secondary details; they are evidence about where the economics of the cycle are spreading.
Arista Networks offers the clearest test of whether the buildout is moving into the network around the compute. Its second-quarter revenue reached $3.036 billion, up 37.7% year over year, and management reported its first quarter above $3 billion while guiding third-quarter revenue above consensus expectations on AI networking demand. Arista is not a cheap stock—the supplied market data puts its P/E at 56.63 and its PEG at 2.60—but its 46.6% YTD gain shows that the market is already willing to reward a credible infrastructure beneficiary. The question after Nvidia reports is not whether Arista can benefit. It is whether more of the AI complex can demonstrate the same combination of demand visibility and earnings conversion.
AMD shows why a strong operating print may still produce a weak stock reaction. The company forecast third-quarter revenue of about $13 billion, plus or minus $300 million, implying roughly 41% year-over-year growth, and reported data-center revenue of $6.72 billion after more than doubling. Yet investors wanted a larger AI payoff and shares fell. That is the exact risk Nvidia faces at a much higher level of scale: when expectations are built around explosive AI upside, solid growth can be interpreted as insufficient rather than impressive. AMD’s current valuation also underlines the dispersion inside the trade, with a P/E of 83.97 and a net margin of 15.6%. The market is not rewarding every AI-linked growth rate equally; it is judging the quality, durability, and profitability of the growth.
This is why the immediate post-earnings move should be read as a relative signal, not a referendum on AI demand. Nvidia can beat, guide higher, and still underperform Broadcom, Arista, or TSMC if investors decide that the next phase of spending will favor networking, custom silicon, and foundry capacity. Conversely, a broad rally across those names would strengthen the case that AI capex is becoming a durable infrastructure cycle rather than a one-platform rush. The late-cycle semiconductor pattern is useful here: leadership often starts with the obvious beneficiary, then the market asks whether the rest of the supply chain can convert demand into sustained profits.
That does not make the dot-com analogy a forecast. Today’s major AI suppliers are already highly profitable, and the central question is not simply whether the technology is real. It is who owns the margin pool as customers diversify their architectures and the buildout becomes more complex. Nvidia’s 63.0% net margin gives it a formidable advantage, but Broadcom’s earnings growth, TSMC’s manufacturing economics, and Arista’s networking execution show why a real AI boom can still produce a more selective stock market. The best evidence of a healthy cycle may ultimately be Nvidia’s ability to create profitable neighbors, not just another quarter of dominance for itself.
Our read is that Nvidia’s earnings should be judged on breadth as much as on the beat. A strong report that leaves Broadcom, Arista, and TSMC behind would confirm demand but keep the concentration risk intact; a strong report followed by improving participation across those suppliers would make the AI trade more durable, even if Nvidia’s own stock reaction disappoints.
We would change our view if Nvidia’s guidance showed that demand is accelerating across the broader infrastructure stack and that the company is expanding the addressable profit pool rather than simply absorbing more of it. Until then, the cleanest conclusion is also the most contrarian: Nvidia can win the earnings release and still lose the next phase of the AI trade.
Our take, not advice. This is opinion commentary — informational only, not personalized investment recommendations. Markets carry risk. Do your own research and consider your own situation before any trade.
▌The Daily Briefing · Free
A new stock idea, every evening.
One stock worth watching each weekday, plus the analysis behind it. Free, in your inbox.
