The $1.5 Trillion Question Nvidia’s Earnings Can’t Answer

Good morning, I’m filling in for Spencer Jakab. Futures point to a weak opening for tech stocks ahead of Nvidia earnings later today. Oil prices continue to fall, while long-term Treasury yields are catching their breath after a sharp drop yesterday. The release of the Fed’s preferred inflation gauge for July will provide another data point in the tug of war between U.S. Treasury Scott Bessent and bond investors.

​📈 Follow our live markets data and coverage.

When tech companies started amping up spending on AI about three years ago, David Cahn, a venture capitalist at Sequoia Capital, asked a simple question: What was the gap between AI investment and the revenue generated from AI?

Back then, he called it AI’s $200 billion question. That, he estimated, was roughly the total revenue needed for investments to make financial sense. It seemed like a lot. In 2024, it became a $600 billion question. In 2025, $840 billion. This year, it’s $1.5 trillion.

These days, fewer people seem to care about the answer. It certainly won’t weigh on investors this afternoon, when Nvidia—the main chip supplier to the AI boom—reports quarterly results.

Why not? Well, Nvidia’s immediate demand signals have been strong. Customers continue to buy huge numbers of its AI chips. Tech giants signaled that capital spending on AI was going to stay high when they reported earnings this summer.

This will likely lead to another Nvidia report highlighting fat margins and an eye-popping backlog. Meanwhile, its price-to-forward-earnings ratio is around 19 times—not high for a fast-growing tech company.

The longer-term question is just how durable its results will prove.

Ultimately, Nvidia and other AI chip makers are living on borrowed time. At some point, big spenders will reach a breaking point where their cash piles are smaller, and they’re unable or unwilling to raise more money from debt or equity investors. If AI turns out to be worth less than it costs, that is inevitable.

Sure, a new crop of AI-chip customers could emerge, aided by financial backstops Nvidia has started providing. Even then, the amount of spending needed to keep Nvidia’s revenue skyrocketing is questionable.

Look at the company’s sales outlook: Analysts surveyed by FactSet expect nearly $400 billion in its current fiscal year, which ends in January, rising to $570 billion the next and around $700 billion the year after.

To calculate the revenue-investment gap on AI, Cahn estimates the total cost of an AI data center is about twice the cost of the chips. Then he builds in a margin for cloud-service providers and the AI product company that serves the end user. In his formula, Nvidia’s $700 billion of revenue in two years would require about $2.1 trillion of end demand.

True, companies have started generating significant revenue from AI, which is giving some investors comfort. OpenAI was on pace for about $27 billion of annual revenue in the second quarter. Anthropic’s annual revenue run rate was at more than $46 billion for the period. Google, Amazon and Microsoft are also bringing in revenue.

But it’s still nowhere near enough to answer Cahn’s question. Until that happens, the AI boom remains a build-it-and-they-will-come story.

↘️ Intuit: The company, known for its tax-preparation and accounting software, forecast slower sales growth in the year ahead as it navigates declines in its desktop business. Shares dropped more than 11% premarket.

↘️ Zoom Communications: The videoconferencing company’s second-quarter revenue rose as its enterprise business recorded its strongest growth in years. Still, shares dipped more than 5% premarket after its guidance for the current quarter fell short of Wall Street expectations.

↗️ Meta Platforms: The tech giant and state attorneys general have discussed a possible mid-trial settlement of a case accusing Meta of making Facebook and Instagram addictive for young users, Bloomberg reported. Shares inched up premarket.

🔎 J.M. Smucker will report quarterly results this morning. Salesforce and CrowdStrike will report after the closing bell.

Cloud computing, meet cloudy numbers. After years of asking investors to take its AI strategy on faith, Microsoft stands out among tech giants for the opacity of its financial reporting.

Across three critical drivers of its AI future—cloud computing, capital expenditures, and its relationship with OpenAI—the company’s disclosures fall well short of what investors need. While Microsoft’s lack of transparency has long drawn complaints, its rapid shift to a capital-intensive business model gives the issue newfound weight.

📰 On this day in 1919, the Coca-Cola Co. successfully sold shares to outsiders for the first time. A syndicate of banks and brokers from around the country bought 417,000 shares to resell to retail investors at an initial offering price of $40. (An earlier attempt at a stock offering in 1892 had failed miserably.)

WSJ | Buy Side: This unique makeup bag has sides that fold down, plus smart pockets for keeping cosmetics and other toiletries secure while traveling.

Asa Fitch is a writer for The Wall Street Journal’s Heard on the Street column, covering technology, chips and AI.