When the CEO of a company building advanced AI starts warning about reckless spending in the AI industry, it's worth paying attention. Dario Amodei, who runs Anthropic, just told the New York Times DealBook Summit that he's "very concerned" about what the industry is calling "YOLO" spending—and he thinks it could end badly for some players.

The Timing Gamble

Here's the problem Amodei laid out on Wednesday: While the technology itself looks promising, the economics are getting sketchy. Companies are pouring massive amounts of money into AI infrastructure right now, but nobody really knows when—or if—those investments will generate proportional returns. It's all about timing, and timing is incredibly hard to predict.

"I think there may be players in the ecosystem who, if they just make a timing error, they just get it off by a little bit, bad things can happen," Amodei said. He called it a "core dilemma"—the uncertainty around how quickly AI will generate real economic value versus how long it takes to build all those data centers powering the technology.

The industry has started calling this investment surge "YOLO" (You Only Live Once), which tells you everything about the mentality. Amodei noted how difficult forecasting future compute requirements really is, and warned that both under-investing and over-investing carry serious risks. "I think there are some players who are YOLO, and I'm very concerned," he added.

Beyond the spending concerns, Amodei emphasized that companies and governments need to start preparing for large-scale workforce retraining as AI rapidly evolves and takes over more jobs.

The Math Problem

Other industry leaders share similar worries. IBM (IBM) CEO Arvind Krishna recently pointed out that the numbers behind the AI supercycle don't really work. With an estimated $8 trillion in capital spending on the table, companies would need roughly $800 billion in profit just to service the interest on that investment.

Yet AI spending has become crucial to the broader economy. A recent analysis by The Kobeissi Letter found that without the massive capital flowing into AI infrastructure, the U.S. would currently be in recession. So we're in this strange situation where AI spending might be simultaneously too much for individual companies and essential for economic growth.

Anthropic's Big Plans

Interestingly, Anthropic itself is gearing up for what could be a blockbuster public offering. The company, which counts Google (GOOG) (GOOGL) and Amazon.com (AMZN) among its backers, has hired Silicon Valley law firm Wilson Sonsini for early IPO preparation work. They're currently negotiating a funding round that could push the company's valuation above $300 billion—setting the stage for a potential 2026 public debut that would rank among the largest tech IPOs ever.

The company's financial trajectory looks impressive. Anthropic expects to hit a $9 billion annual revenue run rate by the end of 2025, with projections of $20-26 billion in annualized revenue for the following year, driven by strong enterprise demand. The current run rate is approaching $7 billion, up from over $5 billion just back in August.

So Amodei finds himself in an interesting position: warning about excessive industry spending while simultaneously building a company that's raising money at astronomical valuations. Perhaps that insider perspective is exactly what makes his concerns worth taking seriously.