Nvidia’s Earnings Could Unwind the AI Trade—And Take the Market With It

The $91 Billion Test

Nvidia’s fiscal second-quarter results, due Wednesday after close, aren’t just another earnings report. They’re the pivot point for a multi-year capital-spending cycle that has lifted index funds, buoyed tech valuations, and rewritten the rules of semiconductor investment.

Seema Shah, chief global strategist at Principal Asset Management, put it bluntly in a CNBC interview Monday: Nvidia sits at the center of this week’s market setup. The question isn’t whether the company will beat estimates—prediction markets already price that at roughly 96 percent—but whether its guidance confirms that the trillions of dollars of AI infrastructure spending are actually generating returns.

“Both Nvidia earnings and what we’re likely to hear from Chair Powell at Jackson Hole are really key to how the market is going to be moving,” Shah said. But Powell’s forward guidance will be thin. That leaves Nvidia’s outlook as the single biggest catalyst for equity markets this week.

The bar is staggering. Management has guided for revenue of $91 billion, plus or minus 2 percent, and non-GAAP gross margin of 75 percent, plus or minus 50 basis points. That guidance explicitly excludes any China data-center compute revenue—a notable gap given the ongoing export restrictions. For context, first-quarter revenue came in at $82 billion, up 85 percent year-over-year, with data-center sales alone at $75 billion.

CEO Jensen Huang has framed the argument clearly: “Compute is revenues. Compute is profit.” He insists that “tokens are now profitable,” positioning Nvidia’s platform as the monetization proof point that investors have been demanding.

If Nvidia clears that bar, the AI infrastructure narrative stays intact. If it stumbles—even by a fraction—the psychological foundation of the trade could crack.

Who’s Betting on a Beat

Prediction markets are pricing in high confidence. Polymarket assigns a 93.5 percent probability that non-GAAP gross margin lands between 74 percent and 76 percent, and a 78.5 percent chance that data-center revenue exceeds $85 billion. The average analyst price target is $304.73, implying roughly 45 percent upside from recent levels around $210.

Yet the stock has slipped 6.94 percent over the past week, trading below $210, even as it remains up 12.28 percent year-to-date. That divergence suggests the market is front-running the possibility of a guidance miss. Investors are asking for evidence that the AI factory buildout is throwing off returns commensurate with the capital commitments made by hyperscalers like Microsoft, Google, Amazon, and Meta.

The underlying tension is one of timing. Shah described the AI cycle as entering a “prove-it” phase. Investors now know the capital-expenditure plans; what they’re waiting for is the monetization timeline. “We want to know that that’s actually going to be monetized, that there’s going to be some kind of return on investment,” Shah said.

That demand isn’t limited to Nvidia. It’s rippling through the entire semiconductor supply chain—luminous, Coherent, Applied Materials—and into cloud services, where margins are increasingly tied to AI utilization rates.

The Proving Grounds

The math behind the anxiety is straightforward. The $1 trillion revenue target Shah referenced—spanning Blackwell and Rubin chips from 2025 through calendar 2027—assumes that every dollar spent on AI infrastructure eventually translates into profitable workloads. If that link breaks, the entire capex cycle risks becoming a wealth-destructive loop.

Rising Treasury yields haven’t deterred big tech from aggressive spending plans. The 10-year yield stood at 4.69 percent on August 20, 2026, near the period high of 4.75 percent hit in late July. Higher borrowing costs normally discipline capex, but AI has become an exception—a strategic necessity rather than a discretionary upgrade.

That exception only holds if the returns materialize. Shah noted that the AI trade has broadened geographically: U.S. tech, Asian chipmakers, Latin American resources, European end-use adoption. The surface area has widened, which means a wobble in Nvidia doesn’t just affect one sector—it recalibrates valuations across multiple regions.

Who Wins, Who Loses

A clean beat, especially on gross margin and forward guidance, would reward Nvidia shareholders and reinforce the thesis that AI capex is a durable, high-return investment. It would also bolster related names: TSMC, ASML, AMD, and the cloud providers that depend on Nvidia hardware. The broader market, particularly growth-heavy indices like the Nasdaq 100, would likely absorb the positivity and move higher.

A miss—or worse, cautious guidance—would punish those same positions. Semiconductor ETFs could drop 5 to 10 percent in a single session. Cloud stocks with heavy AI exposure might see their multiples compress as investors reassess the payback period on infrastructure spend. In a worst-case scenario, the sell-off could spill into defensive sectors as portfolio managers rebalance away from growth altogether.

Shah warned plainly: “If we were to see any wobbles there, then unfortunately that could be bad news for the broader market.” That’s not hyperbole. The AI trade has become a core positioning for institutional portfolios; unwinding it quickly is costly and disruptive.

What Happens Next

Beyond Wednesday, the implications stretch into autumn and beyond. If Nvidia confirms monetization is on track, the capex cycle likely continues through 2027 with renewed vigor. If doubts emerge, hyperscalers may slow procurement, chip makers could face inventory corrections, and the narrative driving tech valuations for the past three years would lose its anchor.

The Jackson Hole symposium later in the week adds another variable. While Shah expects little forward guidance from Federal Reserve Chair Jerome Powell, any shift in tone regarding interest rates could amplify Nvidia’s market impact. A dovish surprise might cushion a miss; a hawkish tilt could worsen it.

For investors, the lesson is clear: Nvidia’s earnings are no longer a company-specific event. They’re a macro signal. The $91 billion revenue test, the gross margin durability, the China commentary, the Rubin launch timeline—all of these are being read as proxies for the health of the entire AI infrastructure economy.

The market has priced in a beat. The real question is whether the beat comes with enough conviction to satisfy a new, more demanding phase of the cycle. If it does, the AI trade breathes easy. If it doesn’t, the unwind begins—and it will touch far more than just Nvidia.