Anthropic IPO filing shows $17 billion quarter, $23 billion loss

Anthropic lost $23.3 billion in the third quarter, according to an amended initial public offering (IPO) filing the artificial intelligence (AI) company made with the Securities and Exchange Commission (SEC) on Monday. Most of that, $21.9 billion, was a non-cash charge from the rising value of its convertible notes and related instruments, not money spent running the business.

The operating picture is much closer to even. The filing shows an operating loss of $567 million for the three months ended Sept. 30, after $974 million of operating income in the second quarter. Revenue reached $16.96 billion, up from $1.38 billion a year earlier, an increase of 1,127% by the filing’s own count, and about 47% above the second quarter’s $11.56 billion. The figures are preliminary and unaudited: the prospectus says its information “is not complete and may be changed,” and it describes the quarterly figures as “unaudited quarterly consolidated results of operations.”

What’s driving it

The headline loss comes from accounting, not operations. The filing records a $21.94 billion “change in fair value of convertible notes, derivative liabilities, and SAFEs” for the quarter, plus $1.64 billion of other expense. The fair-value changes, the company says, “were primarily driven by an increase in the fair value of our outstanding convertible promissory notes resulting from an increase in our enterprise valuation.” In plain terms, as investors valued Anthropic higher, the debt that converts into its stock became more costly on paper. The remaining convertible notes are “all of which are held by Amazon,” the filing says.

Spending is rising fast, too. Compute and infrastructure expense was $9.86 billion in the quarter. Anthropic says revenue grew 1,127% year over year while compute and infrastructure expenses grew 382%, which it presents as proof of compute efficiency. About 44% of revenue for the first nine months came through cloud service provider (CSP) marketplaces. Operating cash flow was $6.6 billion for the nine months, against cash used of $2.0 billion a year earlier, and the company held $95.6 billion in cash and cash equivalents plus $36.3 billion in short-term investments at Sept. 30, the filing says.

The larger number sits in the commitments. Anthropic says it has agreed to “an aggregate committed spend of approximately $895.4 billion, substantially all of which is related to compute and infrastructure.” About $447.5 billion is expected before the end of 2030, and “approximately $69.8 billion” is due in the 12 months after Sept. 30.

Our read

The $23.3 billion loss will draw the headlines, but it mostly measures how much more the company is worth on paper, not how much cash it burned. On the operating line Anthropic is roughly at break-even, and it generated cash over nine months. That is an unusual profile for a company at this stage of a build-out.

The figure that matters more is $895.4 billion. Against a quarter of just under $17 billion in revenue, those commitments are a bet that sales keep compounding for years. The $69.8 billion due in the next year alone is larger than the company’s revenue for the first nine months. Anthropic has the cash to cover the near term, according to its own balance sheet, but the later years depend on growth it cannot yet show.

The risk factors add something rare for a prospectus. A company is telling would-be shareholders that its own safety rules may soon require it to change how it builds models. Investors are being asked to price a business whose own safety policy could constrain how it builds products. All of these numbers can still change before the SEC declares the registration effective.

Who has a stake

The filing names Amazon, Google and Microsoft as the cloud providers Anthropic relies on for compute, and says the three “have made significant equity investments in our company” through their affiliates, while also backing companies whose models compete with Anthropic’s. Anthropic is a public benefit corporation, required to weigh its stated purpose alongside shareholder returns. Its Long-Term Benefit Trust, which the filing describes as “a financially disinterested trust,” “generally has the power to elect and remove a majority of the authorized members of our board of directors.” Dario Amodei, the chief executive and a co-founder, is among seven co-founders the filing describes as stewards of the mission.

In its risk factors, Anthropic warns that “our internal estimates and our expectations regarding future models, rather than our current models, indicate that we may imminently meet our Responsible Scaling Policy’s definition of ‘dramatically accelerated progress attributable to AI.’” The disclosure concerns future models. The filing states: “We have not designated any of our current models as exceeding this threshold, and we continue to assess the underlying data.” The company says it is evaluating what meeting that research-and-development threshold would mean and will publish further reports under its Responsible Scaling Policy (RSP).

What’s next

Anthropic has applied to list its Class A common stock on the Nasdaq Global Select Market under the symbol “ANT,” the filing says. The price range and share count remain blank, and the filing gives no date. The shares “may not be sold until the registration statement filed with the Securities and Exchange Commission is effective.” A later amendment would be expected to fill in pricing, and any update to the RSP would show how the company plans to handle the threshold it describes.

Sources: Anthropic, PBC Form S-1/A, filed with the SEC Nov. 2, 2026