Over the past few weeks, the AI industry has teetered closer to financial calamity than ever. Investors have begun asking more questions about tremendous and circular spending that is funding the AI build-out. Major AI data-center players are loading up on more and more debt, some of it in the form of junk bonds. In the midterm elections, Americans overwhelmingly voted for referendums and candidates that would slow data-center development. And then, as if that weren’t enough uncertainty, Jensen Huang admitted last week to having an affair with an alleged Chinese spy and stepped down as CEO of Nvidia—the AI-chipmaking giant that has been the financial backstop for the AI boom. Even before the uncertainty produced by Huang’s resignation, Nvidia was reportedly considering curbing some of its spending commitments.
In other words, if this is a bubble, about now would be the time for it to pop.
[Read: ]A $5.7 Trillion Face-Plant
And yet, tomorrow, amid this tremendous uncertainty, Anthropic will forge ahead with its immense public offering—beginning to trade on stock exchanges and aiming to raise more than $90 billion in the process, which would be the most of any IPO ever. The timing could either be calamitous or turn out to be a coup. Demand is red-hot: The company reportedly received nearly seven times more orders than there are shares available. Anthropic may emerge not just as one of the world’s largest companies but as the AI trade’s latest savior and lynchpin. That’s not, however, necessarily good news for the rest of our economic well-being.
Tomorrow’s public offering will cap off an astounding 12 months for the company. For years, Anthropic was known as OpenAI’s safety-forward younger sibling. Formed by Dario Amodei and six other former OpenAI employees in 2021, the company consistently operated in the shadow of its older sibling, which became the face of the AI industry. That changed last November, when Anthropic released a new model, Claude Opus 4.5. The model was widely considered to represent a step change in AI capabilities, helped usher in the era of agentic AI, and pushed Anthropic to the front of the pack. Many of the company’s core, early bets—its emphasis on AI coding and other workplace tools, its positioning as being safety- and responsibility-first—paid off. While OpenAI focused on consumer products such as a chatbot and a TikTok-style AI-video app, Anthropic attracted big enterprise customers. Anthropic’s quarterly revenues, according to its IPO filings, have grown more than eightfold since the end of 2025. Just from July to September of this year, the company made more than $16 billion. If, even a year ago, you’d told any analyst that Anthropic would go public before OpenAI, they’d be shocked. Now it seems obvious. (Anthropic did not respond to a request for comment.)
If Anthropic’s public offering goes well, it will fuel renewed confidence and investment in the industry as a whole and, perhaps, fix its wobble. Anthropic’s listing on stock markets (“ANT,” if you’re curious) can be taken as a sign of maturation—evidence that Anthropic and AI, as a major force in the economy, are here to stay. Not only is the company’s revenue growth meteoric; it expects its spending—money that will flow throughout the data-center and AI ecosystem—to grow in lockstep. Anthropic is committed to spending more than $900 billion on data centers and related infrastructure, and said in its IPO paperwork that it expects its expenditures to only “increase significantly.” Being bullish on Anthropic means, by necessity, being bullish on AI; it means believing that, at some point, all of this spending pays off, yielding, as Anthropic states in its founders’ letter, cures for cancer and Alzheimer’s disease, alongside tens of trillions of dollars in global economic growth. Any number of investors looking to sell shares of Nvidia could switch that money over to Anthropic, and Amodei could assume the role of Jensen Huang: the steadfast and responsible shepherd of the industry.
Of course, the opposite could happen. Should Anthropic’s offering flop, it will almost certainly kick off an extended period of pain for Silicon Valley. Anthropic’s demand for more and more computing power—to train more powerful models and serve more customers—is perhaps the single most important indicator of the strength of the data-center boom. If investors prove nervous about even this company, which is cleanly winning the AI-capabilities race and growing its revenues at a historic pace, then the entire edifice is suspect. And a successful IPO, in and of itself, would reflect good vibes but not directly change the underlying mismatch between revenues and planned spending across the AI industry, Anthropic included. Another $900 billion in promised AI and data-center expenditures may not make this infrastructure boom more sustainable—but it could add to losses if and when the bubble does pop.
Perhaps the most likely course is somewhere in between total AI dominance and collapse. Anthropic goes public and sees sustained, even spectacular, success as a business and a stock. The AI-investment flywheel, flush with new cash, spins ever faster. Fortune 500 companies that deploy Claude agents see their profit margins rise. Meanwhile: All of the AI money flooding markets, alongside spending on data-center construction, keeps inflation high. All of those agents mean that the labor force remains stagnant, or shrinks.
If you listen to Anthropic’s communications, both the IPO and the company’s immediate profits are secondary concerns. The company’s prospectus warns investors, again and again, that it might make decisions that do not maximize profits. And that significant increase in spending, Anthropic notes, “may increase regardless of whether our financial performance improves.” That cavalier attitude toward making money is in line with Anthropic’s origins as a safety-oriented AI lab (and it was, at the start, genuinely a lab). It’s also in line with Anthropic leadership’s zealotry about the inevitability and imminence of superintelligence, which stands out even among their AI-executive peers. Going public, and all of the capital that allows Anthropic to access, is a sign that the firm is intent in securing its role in dictating the direction of such all-powerful machines.
This IPO lands at a time of anxiety about AI as the technology of the future. Rogue-agent swarms and a deluge of AI-powered cyberattacks have left many wondering whether the industry should slow down and led many to suggest that it do so—including Amodei himself. And yet, Anthropic is forging ahead, shutting down fear and criticism just as all of its peers have done at every turn; the AI industry’s self-styled superego may turn out to be the id.
Anthropic was founded, above all, on faith in Amodei’s clairvoyance and judgment. Seemingly the Anthropic CEO’s most candid answer to why he’d left OpenAI was in a Bloomberg interview this past summer in which he said, “Why argue with someone when you don’t have the same vision and you don’t trust them? The way to resolve it is: You go off and do your thing; they go off and do their thing.” Going public is entirely consistent with such messianism. What sets Anthropic apart is that it is a company of true believers—in the almost transcendental power of AI and, as important, in themselves.
Last year, at Anthropic’s headquarters, I had a telling conversation with two of its researchers, Trenton Bricken and Sholto Douglas. Bricken was telling me that in an ideal world, he’d want AI development to proceed far more slowly and carefully. Douglas disagreed: Any delay “comes at the cost of millions of lives,” he told me. At the time, I wrote that the impulses to speed and slow AI development seemed to be at war within the company. Now one side has clearly won out.
Leave a Reply