Nvidia Earnings Expose the Cracks in the AI Trade
Nvidia's banner quarter reveals a tech trade splitting into winners and losers across every layer. Software recovers while chipmakers diverge, and Alphabet hemorrhages value — but the real risk is whether hyperscalers can spend a trillion dollars on AI.
All About Acceleration
Nvidia posted a banner quarter on Wednesday and the market responded the way it always does when the king of AI chips delivers: with relief, gratitude, and a sudden reminder that not everyone gets to share in the glory.
Shares in Marvell Technology, another major chip designer, plunged after its own strong earnings report. Two companies with similar results. Opposite fates.
This is the new reality of the AI trade — and it is reshaping how investors think about who wins and who pays the price.
The Infrastructure Haves and Have-Not
For years, the playbook was simple: buy AI exposure through Nvidia, or buy the Magnificent Seven and hope for the best. Both strategies worked because AI demand looked like a rising tide lifting every ship in the harbor.
That tide is gone.
Nvidia’s results are bifurcating the trade into clear layers — the infrastructure haves and the infrastructure have-nots. At the top sit the companies with guaranteed access to advanced chips and the capital to deploy them. Everyone else is now priced on suspicion.
Steve Koenig, Macquarie’s US head of software and services research, put it plainly: “This is all about acceleration.” Stocks that can demonstrate they are accelerating — whether through revenue growth, margin expansion, or tangible AI integration — are getting rewarded. The rest are being abandoned.
The Software Rebound
One of the less-discussed shifts from this earnings season is the software trade coming back from the dead.
Software stocks spent most of 2026 getting hammered by investors convinced that AI made large swaths of the industry redundant. That narrative has cracked. Over the past week, software stocks roared back as investors began pricing in a more nuanced view — one where AI doesn’t destroy software, but rather separates the accelerators from the stagnators.
The companies that can embed AI into their products and monetize it are recovering. The ones that can’t are still down. The sector is no longer a binary bet on disruption; it is a differential bet on execution.
The Lag Seven
Even among the biggest names in Big Tech, the divergence is stark. The Magnificent Seven is starting to look more like the Lag Seven to some strategists.
Amazon is up 15 percent over the past month. Nvidia is up 10 percent. Alphabet, meanwhile, has shed roughly $692 billion from its market value, falling 15 percent from its May all-time high.
Alphabet’s decline tells you everything about the new logic. The company is making massive infrastructure investments in AI while simultaneously losing top talent and facing growing concerns that it is falling behind. The market is not punishing Alphabet for spending — it is punishing it for spending without clear evidence of acceleration.
The difference between Amazon and Alphabet matters more than the direction of either stock. Amazon’s infrastructure bet is priced as a competitive advantage. Alphabet’s is being priced as a liability.
The Trillion-Dollar Question
The real risk in this trade may not be demand — AI demand is still hot. The risk is execution.
Hyperscaler capital expenditure is now expected to exceed $1 trillion in 2027. That is an enormous commitment. And it depends on a supply chain that is already straining under the weight of competing orders, labor shortages, permitting delays, and rising local opposition to data center construction.
Ellen Hazen, chief market strategist at F.L. Putnam Investment Management, flagged this as the central uncertainty: whether that trillion dollars can actually be spent. Some of the financing deals hyperscalers have signed are massive, but financing is not the same as deployment.
Chad Morganlander at Washington Crossing Advisors warned that for unprofitable projects, debtholders will eventually pay the price. The implication is clear: some of today’s AI infrastructure bets will look very different in hindsight, and not all of them will succeed.
What This Means for Valuations
The segmentation visible in Nvidia’s aftermath will not disappear. It is the rational response to a market where AI demand is real but unevenly distributed.
Chipmakers face a two-tier valuation: Nvidia commands a premium for its dominance, while other designers are judged on whether they have a credible path to securing capacity. Software is split between companies that can prove AI acceleration and those that cannot. Big Tech is divided between players deploying capital effectively and those burning it without visible returns.
Investors who treat AI stocks as a monolith are already behind. The trade has moved past broad conviction into granular discrimination — and the earnings season has only made that sharper.
Who Wins and Who Loses
The winners so far are the companies with guaranteed chip access and the balance sheets to back aggressive deployment. Amazon and Nvidia are leading the pack. Software companies with credible AI monetization stories are recovering from their drawdown.
The losers are companies making big bets without clear acceleration paths. Alphabet’s trajectory illustrates this. So do the chip designers with strong earnings reports that still get punished — Marvell included.
The next test will be whether hyperscaler capex translates into shipped product and revenue growth in 2027. If it does, the infrastructure haves widen their lead. If it doesn’t, the market will start questioning whether the trillion-dollar spend is a burden rather than an asset.
The trade is getting more segmented, not less. The question for investors is no longer whether AI is real — it is which part of the stack they are exposed to, and whether that exposure is accelerating fast enough to justify the price.