Late last week, as I predicted, Anthropic’s Dario Amodei decided to stick by his two thin red lines on fully autonomous weaponry and domestic surveillance, incurring the wrath of both President Trump and Secretary of War Pete Hegseth. In a post, Trump called Anthropic a “radical left, woke company” and instructed all federal agencies to “immediately cease” using its technology. Hegseth followed up by labeling Anthropic a “supply chain risk to national security”—an unprecedented act that Dean Ball, who worked with the Trump administration on its A.I. Action Plan, described as an attempted “corporate murder.” A few notable subplots here are worth teasing out:

And now for the main event…

Last week, Nvidia reported another blowout quarter: $68.1 billion in revenue, a 73 percent surge over the same period from the previous year and about $2 billion higher than what Wall Street was expecting. Guidance for the current quarter also came in above expectations, at $78 billion in revenue, as the company—already the world’s first $5 trillion market cap firm—continues to chart new territory. And yet the stock fell 5 percent after the earnings call, a slide that has continued into this week. Trading at around $180 per share, it’s now well off the highs of $212 per share that the company notched as recently as October.

Nvidia, of course, has inserted itself into virtually every layer of the A.I. stack—from supplier to both early-stage and pre-I.P.O. investor. Last year, it invested $17.5 billion in private companies, including “A.I. model makers that purchase our products,” according to the company’s Q4 report. That’s not to mention another $3.5 billion in “land, power, and shell guarantees” to support the all-important data center build-out, as well as $27 billion in cloud service agreements. More recently, Nvidia has committed to investing up to $10 billion in Anthropic, and just invested $30 billion in OpenAI.

This positioning explains its fortress balance sheet and too-big-to-fail status, but it has also invited skepticism about the company’s size and reach if and when the industry hits an air pocket. Nvidia’s most recent 10-K filing readily acknowledged that a “significant” amount of its revenue comes from a “limited” number of customers. For fiscal year 2024, the company noted, one customer accounted for 13 percent of total revenue; in fiscal year 2025, three customers represented 23 percent; and last year, two customers represented 36 percent. More broadly, hyperscalers contributed more than half of Nvidia’s data center revenue. Yes, Microsoft, Meta, Google, and Amazon alone plan to spend about $700 billion combined on data center infrastructure this year. But what happens when they move past this phase of their A.I. build-out? Or if the expansion stalls out entirely? These concentration risks extend to the neoclouds as well: CoreWeave, for instance—which is relying on billions in debt to build out its infrastructure—gets the majority of its revenue from Microsoft. (Naturally, Nvidia has invested $2 billion in CoreWeave, too.)

This year, at least so far, the Magnificent Seven seem to have lost some of their sheen. Shares of Microsoft are down 14 percent for the year; Amazon is down 8.4 percent; Nvidia is down 4.5 percent; Alphabet is down 4.5 percent; Apple is down 3 percent; and Meta is just about flat. These six companies alone represent more than a quarter of the S&P by weight. And yet, the broader index is down only slightly on the year.

Jeff Currie, an investment analyst at Carlyle, has described this shift as a rotation out of tech. Since October, he wrote last week, tech (“new economy”) stocks have been under pressure, while energy, metal, and mining (“old economy”) stocks have soared. When a similar phenomenon happened during the dot-com collapse in 2001, it was called the “revenge of the old economy.” These early indications of broadening—which investors have been clamoring for—follow several years of extreme tech concentration. “The rotation out of Big Tech has underpinned sizable capital flows into value-oriented stocks, small caps, and international equities,” LPL Financial’s chief technical strategist, Adam Turnquist, wrote the other week. “This shift has created a ‘drinking water from a fire hose’ effect for many smaller segments of the market absorbing these outsize inflows.”

Tejas Dessai, the head of thematic research at Global X, told me that the shift underway is a simple one: The hyperscalers “are becoming the utilities of the intelligence age.” As Currie noted, this means a shift from asset-light to asset-heavy businesses; according to Dessai, it also means that “they will hold all this risk” themselves, and the market, at least in part, is beginning to reprice these companies accordingly. Considering also that hyperscaler capex intensity has surged well above expectations every six months for the past few years, and you get an environment where, according to Dessai, “Nvidia will continue to dominate. Whether it’s fairly priced or not remains to be seen.”

He continued: “You have to think about how much capex we need to spend to build a sector from the ground up. When you think about it from that perspective, we have an estimate for about $6 trillion to $7 trillion that will be spent on A.I. infrastructure between 2025 and 2030—that doesn’t sound so ridiculous when you think about running the global G.D.P. and fusing A.I. across the board.” He doesn’t think things will slow down until all that necessary capex has been spent and the infrastructure to support this “new sector” has been built.

But are we actually building a new sector, and do we need new-age utilities to support it? The trouble is that the A.I. pipeline is long and somewhat obscured, the investments are enormous and circular, the return on those investments remains unclear, and, perhaps most critically, the companies for whom the infrastructure is being built cannot today survive on their own. If the monetization narrative shifts and the funding faucet is subsequently turned off, the economics of this new sector would look dramatically different. (The current utility sector, meanwhile, is… profitable.) Even so, Dessai doesn’t think we’re near the top of the market. “I don’t think we’re at that peak of the scaling-up period,” he said. “I think we’re still in the early stages.” Maybe it’s time to buy the Nvidia dip after all. (This is not investment advice.)

That’s all for today. I’ll see you on Thursday.

Ian

Need help? Review our FAQ page or contact us for assistance. For brand partnerships, email ads@puck.news.

You received this email because you signed up to receive emails from Puck, or as part of your Puck account associated with {{customer.email}}. To stop receiving this newsletter and/or manage all your email preferences, click here.

Puck is published by Heat Media LLC. 107 Greenwich St., New York, NY 10006