Anthropic to Open Its Books to Investors Ahead of a Potential $2 Trillion IPO

The AI lab will share previously undisclosed quarterly financials this week, just as a messy revenue disclosure from rival OpenAI rattled markets and exposed how little outsiders actually know about these companies' books.

Anthropic is set to disclose its third-quarter financial results for the first time on Wednesday, sharing the numbers directly with a group of institutional investors as the company lays the groundwork for what could become the largest initial public offering ever attempted. Two people familiar with the matter, who spoke on condition of anonymity because the details are confidential, described the meeting to CNBC. Bloomberg first reported that it was taking place. Anthropic declined to comment.

The timing is not incidental. Private AI companies have never been required to disclose their finances the way public ones are, but their revenue figures have started moving markets anyway, as investors try to price in just how much money these labs are actually making versus how much they're burning. Anthropic's decision to share Q3 numbers with potential investors comes right as that dynamic blew up in spectacular fashion with its biggest rival.

OpenAI's numbers just spooked the whole market

Last week, OpenAI gave current investors an updated revenue figure as part of an ongoing fundraising round: $50 billion. That was a jolt, because the number widely reported before that had been $70 billion. The gap wasn't really a sign that OpenAI's business had suddenly shrunk. According to a person familiar with the matter, the lower figure reflected a change in accounting that stripped out revenue tied to cloud partners, not a drop in actual demand for OpenAI's products.

Nuance aside, the headline discrepancy did real damage. Confusion over the numbers dragged down major semiconductor stocks and the broader Nasdaq, a reminder that even sophisticated markets can react first and parse the fine print later. It also raised an uncomfortable question for the entire sector: if a company as closely watched as OpenAI can produce two wildly different revenue figures depending on how it counts partner revenue, how much can anyone trust the numbers coming out of any private AI lab?

That's the backdrop against which Anthropic is choosing to disclose its own figures. It's not yet clear whether Anthropic will account for its revenue the same way OpenAI did, including or excluding partner contributions, but the stakes of that choice just got a lot more visible. Whatever Anthropic shows investors this week will be read as a test of how much confidence the market should place in private AI valuations generally, not just Anthropic's.

For Anthropic itself, the meeting serves two purposes. It gives investors a clearer picture of the company's actual growth and profitability heading into a public listing, something that's been hard to come by since the company has kept its financials close. And it gives Anthropic's executives and the bankers running its IPO process a read on how the market reacts, both to the numbers themselves and to the accounting choices behind them, before the company has to make any of this official in a public filing.

The scale of the bet

The numbers under discussion are enormous. Anthropic is pushing ahead with plans for a listing that could value the company at around $2 trillion, based on its latest reported revenue figures, according to people familiar with the plans. Those same people said Anthropic's revenue could top $100 billion by the end of this year. If the $2 trillion valuation holds, it would put Anthropic's debut ahead of SpaceX's, which stands as one of the most consequential private-to-public transitions in recent memory.

Anthropic filed confidentially with the Securities and Exchange Commission earlier this year, a common move that lets companies start the regulatory process without making their financials public right away. The company's S-1, the document that would eventually spell out its official financials for public investors, hasn't been released. Reuters has reported that Anthropic plans to list in November, though people familiar with the company's plans told CNBC they couldn't confirm an exact date.

Rates are the wild card

Even if the numbers come in strong, Anthropic would be going public into a market that's gotten noticeably more skeptical of AI spending. Climbing interest rates are putting pressure on parts of the broader AI trade, and that anxiety got a very public airing this week. Broadcom CEO Hock Tan told a conference on Monday that as rates rise, access to capital could become the single biggest constraint on AI companies, not demand, not technology, but simply the cost and availability of money to fund the buildout.

Tan's comments landed hard. Chip stocks were lower Tuesday, extending a sell-off that began overnight on worries about whether the financing underpinning the AI boom is as durable as it looks. That selloff and the OpenAI revenue confusion are really two symptoms of the same underlying nervousness: a market that has poured enormous sums into AI infrastructure and companies and is now double-checking the math.

For Anthropic's bankers, that makes gauging investor appetite for growth and AI stocks right now one of the most important variables in the entire IPO process, arguably more important than the revenue figure itself. A company can have great numbers and still face a cool reception if the broader market has decided it's done taking AI valuations on faith.

Wednesday's meeting won't settle any of that on its own. But it will be the first real data point the market gets on how Anthropic's business actually looks, at a moment when every number out of this industry is being read as evidence for or against the idea that AI's financial foundations are as solid as its valuations suggest.