Anthropic Updates Prospectus: First Three Quarters Revenue $33.242 Billion, Swings from Profit to Loss, Valuation Expectations Lowered

On November 2 local time, Anthropic's latest revised IPO prospectus disclosed that the company's revenue for the first three quarters reached $33.242 billion, a year-over-year increase of 1206%.

However, the quarter-over-quarter revenue growth rate in the third quarter fell from 144% in the second quarter to 47%, and operating results under GAAP (US Generally Accepted Accounting Principles) also swung from profit to loss.

At the same time, the company disclosed approximately $895.4 billion in computing power and infrastructure procurement and leasing commitments, and for the first time explicitly warned in its IPO filing: internal estimates show that future models may soon approach the risk threshold of "AI significantly accelerating progress" in its Responsible Scaling Policy (RSP).

Behind the growth are high computing power costs. Data shows that Anthropic's computing and infrastructure spending in the first three quarters of this year reached $18.808 billion, a year-over-year increase of 329%, accounting for 57% of revenue. Among this, computing power and infrastructure spending in the third quarter alone was approximately $9.863 billion, a year-over-year increase of about 382%, accounting for 58% of revenue in that quarter. In the third quarter, the company had an operating loss of $567 million, whereas in the second quarter it had achieved an operating profit of $974 million.

Anthropic explained that the third quarter included $5 billion in additional short-term computing power costs, as well as approximately $2 billion in non-cash expenses related to a donation matching program. The operating loss for the first three quarters was $1.26 billion, but after excluding stock-based compensation, non-cash charitable donations and settlement expenses, adjusted operating profit was $4.1 billion.

The total computing power and infrastructure commitments disclosed by the company this time amount to approximately $895.4 billion, of which about $447.5 billion is expected to occur before the end of 2030, and about $69.8 billion matures in the next 12 months (from October 1, 2026 to September 30, 2027).

In addition, approximately $213.8 billion of the total corresponds to commitments that carry termination rights or preconditions before the relevant arrangements take effect. These conditions include partners obtaining financing, facilities being ready, and regulatory approvals. Anthropic warned that some partners rely on third-party financing to build computing power, and if financing tightens, the relevant capacity may be delayed, reduced, or become more expensive.

Page 29 of the revised filing also adds a noteworthy risk disclosure: based on internal estimates and expectations for future models, Anthropic estimates that it may soon reach the threshold defined by the RSP as "significant accelerated progress brought about by AI."

This standard refers to AI driving progress at a rate where one year is equivalent to the baseline progress of two years from 2018 to 2024. The company emphasized that it has not yet determined that any existing model has reached this threshold and is still evaluating the relevant data. This means that AI-assisted R&D may accelerate model iteration, and may also cause safety, alignment, and evaluation capabilities to lag behind technological progress.

According to market sources, some existing and potential investors are currently considering a valuation for Anthropic of approximately $1.5 trillion, lower than the previous expectation of approximately $2 trillion.

In addition, the company's leading competitor OpenAI is also facing similar pressure. According to market sources, OpenAI's new round of financing target has been reduced from $30 billion to $20 billion, and the target valuation has been reduced from $1.4 trillion to $1.25 trillion. The pullback in tech stocks, rising financing costs for AI infrastructure, and investors' scrutiny of the payback period for massive computing power spending are jointly changing the capital market's pricing of AI companies.

The adjustment in the valuation expectations of the two companies does not mean that AI demand has peaked. Anthropic's revenue in the first three quarters still grew more than 12-fold, and enterprise customers continue to expand. But the market has begun to shift from chasing revenue growth to asking whether profits can be sustained, whether computing power commitments can be fulfilled, and whether AI can truly improve R&D efficiency.