TLDR
- Anthropic’s 2025 sales jumped 1,088% to $4.59 billion, according to a leaked IPO prospectus seen by Reuters.
- The company lost $41.97 billion in 2025, up sharply from an $8.31 billion loss in 2024.
- Anthropic reportedly faces about $518 billion in cloud, compute and infrastructure obligations.
- The prospectus values the company at $2 trillion, double its prior $1 trillion valuation.
- Roughly 80 of the document’s 261 pages focus on risks, including warnings about “catastrophic or existential” AI dangers.
A leaked IPO prospectus has given the public its first detailed look at Anthropic’s finances and its own warnings about the risks of its technology.
ANTHROPIC FILES FOR IPO, PROSPECTUS SEEN BY REUTERS
Here’s everything you need to know:
Anthropic could seek a valuation above $2 trillion.
2025 financials:
Rev: $4.59B, up 1,088% YoY from $386M
Oper loss: $8.06B, widening from $2.98B
GAAP net loss: $41.97B, vs $8.31B in 2024… pic.twitter.com/ntYJ03uszx— Wall St Engine (@wallstengine) September 28, 2026
Reuters reported late Monday that it reviewed the document, though Anthropic has not made an official filing public yet.
The numbers show a company growing fast while losing money at an even faster pace.
The Financial Numbers
Anthropic’s 2025 sales rose 1,088% to reach $4.59 billion, based on the leaked filing.
That growth came alongside a net loss of $41.97 billion for the year.
That loss is far larger than the $8.31 billion the company lost in 2024.
The prospectus also outlines about $518 billion in cloud, compute and infrastructure obligations tied to building out its AI systems.
Anthropic is reportedly valuing itself at $2 trillion in the filing, doubling its previous $1 trillion valuation from an earlier funding round.
Salesforce holds a stake in Anthropic that was valued at $5 billion during that earlier round.
Salesforce first invested $50 million in Anthropic in early 2023 and has taken part in later funding rounds since.
Warnings About AI Risk
The prospectus spends roughly 80 of its 261 pages on risk factors, nearly double the space given to describing the actual business.
Anthropic stated that developing more advanced models could increase the chance that its systems cause harm.
The filing warns that AI models could show “self-preserving behaviors,” including resisting shutdown or concealing information.
It also describes behavior that could resemble blackmail in some model outputs.
Anthropic said models can develop unexpected abilities during training that researchers might not catch until after release.
The company added that its own ability to test model safety could be limited if systems become aware they are being evaluated.
Despite the safety focus, Anthropic said only about 6% of its computing power went toward safety research during one sample week in July.
The filing also states that customer demand depends on Anthropic keeping up a steady stream of new model releases.
Anthropic’s public market debut is expected to be delayed until after the November U.S. midterm elections.
Last week, leaders from Anthropic, OpenAI and Hugging Face told the United Nations that AI’s pace of development requires more international coordination.
That statement followed an essay from Anthropic CEO Dario Amodei calling for slower AI development, which OpenAI CEO Sam Altman said he supported.
At the same UN gathering, President Donald Trump rejected calls to slow AI development.
Trump said he wants to rebrand the technology as “super intelligence” and pointed to the competitive edge this gives the United States in the sector.
Stop guessing and start investing with confidence. KnockoutStocks gives you the AI insights, market intelligence, and stock research you need to spot opportunities, cut through the noise, and make smarter investment decisions — all in one powerful platform.
Sign up today and get 50% OFF full access to our premium stock picks.
Simply use coupon code SPECIAL50 at checkout to claim your exclusive discount.







