Anthropic's IPO prospectus leaks: sales up 1,088% in 2025, but losses of $41.97 billion

According to leaked information about Anthropic's IPO prospectus — which the company itself has not made public officially — the firm behind Claude saw its revenue soar in 2025 to $4.59 billion, growth of 1,088% over the previous year. Reuters had access to the document, and a subsequent report, cited by Yahoo Finance and attributed to the Financial Times, added further details about the figures and the company's plans.

The flip side of that explosive growth is equally striking: Anthropic is said to have lost $41.97 billion in GAAP accounting terms during 2025, far above the $8.31 billion lost in 2024. In other words, losses multiplied fivefold while revenue multiplied elevenfold, a pattern that reflects the pace of investment and cash burn characteristic of the race for artificial intelligence infrastructure, rather than a problem of commercial demand.

The prospectus itself reportedly includes, according to the cited sources, a warning aimed at future investors: the company's technology could pose "existential risks to humanity." That is unusual language for an IPO document, but consistent with the public stance Anthropic and its founder have maintained on the potential dangers of the most advanced AI models, including Claude itself.

One of the most eye-catching data points in the analysis is the volume of commitments Anthropic has taken on in cloud, computing and the infrastructure needed to train and run its models: some $518 billion. It is a figure that, if confirmed, illustrates the extent to which Anthropic has tied its future — and that of its suppliers — to the execution of an infrastructure plan of enormous scale. The reasoning is straightforward: if the company fails to meet the growth and revenue expectations that justify that spending, the risk would not stay inside Anthropic alone but would spread to the chain of cloud and computing providers with which it has signed those commitments, with implications for both their shares and their debt.

In parallel, the company is said to be aiming for a $2 trillion valuation in its market debut, more than double the $965 billion at which it was valued in its last private funding round, closed in May. That valuation leap, if it materializes, would make the deal one of the most ambitious technology IPOs seen to date, and would subject the real accounts of one of the most decisive companies in the current generative AI wave to public scrutiny for the first time.

An indirect beneficiary of that potential valuation leap would be Salesforce, according to the article. The enterprise software company initially invested $50 million in Anthropic in early 2023 and took part in subsequent rounds; its stake was valued at $5 billion in the latest funding round. If the IPO valuation approaches the $2 trillion being floated, that stake could appreciate significantly, making Salesforce a relatively low-profile player but one heavily exposed to Anthropic's stock market success.

It is worth stressing that all of this comes from leaks and from sources with knowledge of the prospectus, not from a document filed publicly with regulators or officially confirmed by Anthropic. The figures for revenue, losses, infrastructure commitments and target valuation could differ from what is ultimately included in a formal filing, should one take place. Even with that caveat, the episode foreshadows the kind of scrutiny that will surround Anthropic if it goes public: a company with extraordinary revenue growth, equally extraordinary losses, and a level of financial commitment to AI infrastructure that ties it directly to the health of the entire ecosystem — cloud providers, chipmakers and other big tech companies — that underpins the current artificial intelligence boom.