Can Anthropic’s IPO live up to the hype? What finance pros think

This whole AI boom has been one wild ride. In a couple short years, LLMs, once a way for college students to avoid writing their own essays, has been touted as the most groundbreaking invention since the Internet, electricity, or construction. AI was going to cure cancer, put us all out of work, or launch Skynet—pick your poison. Then some started warning of a potential stock market bubble that could put tulipomania to shame. More recently, some CFOs have pumped the brakes on tokenmaxxing after receiving some eyewatering AI bills.

Then, last week, Anthropic filed an S-1 with the SEC, and the rhetoric ramped up once more. Possibly the biggest IPO in history! A potential $1 trillion to $2 trillion valuation! 1,300% revenue growth! A $40 trillion total addressable market!

But how much of the hype is, well, hype? How sustainable is that level of growth? Are there hidden red or yellow flags in the S-1 that could spell trouble down the road? And what’s the deal with that change in auditors? To find out, we asked the kinds of people for whom a nearly 400-page S-1 constitutes a beach read: accountants, CFOs, CEOs, and IPO specialists.

So, how’s Anthropic really doing? Skeptics and bubble-watchers may not enjoy hearing this, but many finance gurus say the AI lab’s got a good thing going. “They have a great business,” Jim Neesen, managing partner and founding executive at accounting advisory firm Connor Group, which helped tech companies like Tesla, Uber, and Spotify to go public, told CFO Brew. “I think it surprised people how strong financially they were.”

Anthropic brought in $16 billion in the first half of this year, Neesen noted. “It has operating profit when it’s adjusted. It has 45% margins. They have a cash war chest of $130 billion. Those were numbers that people weren’t necessarily expecting.”

“The zeitgeist on the internet is that these numbers are surprisingly good,” Siqi Chen, the CFO-turned CEO of CFO.ai, told CFO Brew. He added, “Generally, you don’t go IPO if you don’t think your next two quarters are locked.”

Some observers raised concerns that too much of Anthropic’s business depended on hyperscalers. The S-1 showed healthy diversification, according to Neesen. “They had nearly 12,000 customers spending over $100,000 a year, and 2,000 customers spending over a million a year,” he pointed out.

Meteoric, but for how long? But to what extent can that growth continue–especially given that many companies scaled back their AI spend this year? “We’re starting to see some of the pullback with our customers who were huge Anthropic enterprise buyers because they chewed through so many tokens,” Justin Etkin, co-founder and CEO of spend management platform Tropic, observed. “If we play the story forward a quarter from now,” he added, the “hyperbolic curve that we’re seeing is 100% going to start leveling off.”

In fact, Anthropic’s Q3 2026 financials suggest this may be happening already: Its revenue grew 144% in Q2 but only 47% in Q3, CNBC reported.

Even Anthropic warned in its S-1 that it “does not expect” the same level of growth going forward, per Bloomberg, which wrote that changing demand could “pose a significant threat to the business given Anthropic’s substantial commitments on data centers and chips.” It pledged to spend around $475 billion on compute in the next seven to ten years, and many of its agreements with hyperscalers are non-refundable.

Competition is also ramping up. The performance gap between frontier models and cheaper open-source models, Chen said, is getting narrower, and smaller AI companies are increasingly able to replicate the kind of work the major labs perform, “particularly around post-training and fine-tuning.”

“Where is the long-term value if we believe the gap will continue to close?” he asked.

Striking while the iron’s hot. Neesen, though, feels Claude is sticky enough to give Anthropic an edge. One disclosure from the S-1, he observed, is that “when customers sign up, they spend about 900% more the next year.” As he noted, “We have Claude, and once it gets hooked into your business, you’re using it a lot. It becomes part of your daily lifestyle.”

Etkin believes that Anthropic may simply be seizing the high-revenue moment. “Anthropic is going public at the perfect time,” he said. “In advance of an IPO, you want to run up your growth as much as possible, optimize your metrics, and they are right in the sweet spot.”

Auditor switch. Leave it to former PCAOB board member Christina Ho to observe the fact that Anthropic changed auditors. The company disclosed in its S-1 that it had switched from Grant Thornton, its auditor from 2023 to 2025, to EY, she told CFO Brew.

That move isn’t unusual, said Ho, who is now chief assurance officer at AI auditing platform and CPA firm Oath Verified, as it’s common for companies to move to the Big Four for auditing once they go public. But, she pointed out, EY, which Bloomberg Law calls the “go-to auditor” for AI companies, also audits Amazon and Alphabet, which have multiple overlapping relationships with Anthropic.

“When you think about the relationship between Anthropic, Amazon, and Alphabet, they’re investors…they are customers, they are suppliers” for one another, Ho said. That’s not necessarily a threat to independence, she said, nor is it a violation of current auditing rules, but it does create a novel situation worth watching. “I don’t think in the past that there’s [been] something like that at this scale,” Ho said.

IPO rush incoming? Anthropic’s S-1, Neesen said, could mean the AI IPO rush is on. If the company and its rival OpenAI perform well after going public, “it can actually propel more of these AI companies that are going to look to go [public] in ’27 and ’28,” he said. He estimates that 50 to 100 AI companies, five or 10 of them substantial, could file for IPOs in coming years.