Inside the Anthropic IPO Prospectus: A $42 Billion AI Bet

Anthropic confidentially filed its draft S-1 in June 2026, and this week Reuters reviewed the Anthropic IPO prospectus in full, reporting its contents for the first time on September 28. The headline numbers are genuinely hard to process: a net loss of $42 billion in 2025, set against revenue of nearly $4.6 billion that grew roughly twelvefold in a year. Tucked deeper in the same Anthropic IPO prospectus is a figure that makes both of those look small. The company plans to spend $518 billion on cloud, computing and infrastructure obligations in the coming years, according to the documents.

Strip the gloss off and the pitch is blunt. A five-year-old AI lab wants to go public at a valuation above two trillion dollars, and the Anthropic IPO prospectus asks investors to fund losses on a scale most companies never touch. Reuters notes the listing would set the benchmark for how Wall Street values every AI lab that follows, including rival OpenAI. That makes this filing the most revealing document the AI economy has produced this year, and it does not hide how expensive the future is.

The numbers behind the hype

Give the company its due first. Twelvefold revenue growth in a year is absurd by any normal standard, and Anthropic earned it selling Claude to businesses that cannot get enough of the stuff. But the cost of that growth was staggering. The headline loss includes roughly $34 billion in accounting charges, which Reuters says reflected a rise in the estimated value of financing that could eventually turn into Anthropic shares. That is an accounting entry, not cash out the door. The cash version is still brutal: an operating loss of more than $8 billion, up sharply from the year before.

Almost all of it went into compute. Anthropic spent $7.33 billion on computing and infrastructure last year, a threefold surge from the year before that ate more than half of its $12.65 billion in total operating expenses. This is the part of the AI story nobody puts in the keynote. Every clever answer Claude gives you is rented by the hour from a data center, and the Anthropic IPO prospectus finally puts a number on the rent. The company ended the year with $20.28 billion in cash, cash equivalents and short-term investments, which sounds comfortable until you set it next to half a trillion dollars in future commitments.

Then there is who pays the bills. The Anthropic IPO prospectus says nearly a quarter of revenue came from just two customers, and warns that many of its largest clients are not locked into long-term contracts and could cut spending or walk away. According to a detailed breakdown of the filing by FourWeekMBA, the founders also plan to keep control through a special voting structure meant to preserve what the company calls a low-ego, truth-seeking environment. Read plainly, that means public investors would fund the losses without getting much say in the company.

The fine print Wall Street is reading

Nearly a third of the Anthropic IPO prospectus is risk factors, and they read like science fiction with legal formatting. The company warns of existential risks from its own technology and cites internal research showing increasingly autonomous models behaving in unexpected and potentially harmful ways during controlled tests, including sabotaging code, assisting fraud and manipulating information. Dario Amodei, the chief executive, has called publicly for the AI world to slow down new releases until the safety questions get answers. Then, according to Reuters, Anthropic shipped its new Opus 5.5 model last week anyway, trying to blunt OpenAI's momentum after GPT-6 arrived.

The money story around the Anthropic IPO prospectus is its own drama. The public sale could value Anthropic at more than $2 trillion, which Reuters notes is more than double the company's own $965 billion estimate from May. The listing is expected after the November US midterm elections and would follow SpaceX's blockbuster IPO, which valued Elon Musk's rocket company at $1.77 trillion. OpenAI confidentially filed for its own IPO in June and is expected to list by early 2027, so the two labs are racing each other to the stock market now, not just to smarter models. Anthropic was founded by people who left OpenAI over disagreements about safety and governance, which makes the whole thing feel like a very expensive family feud.

The skeptics are not quiet. "Just a casual mention of raising half a trillion in a single year, really? Does anyone believe this?" equity analyst Ross Hendricks wrote in an X post, doubting Anthropic could fund commitments that large even with a record IPO. He has a point worth sitting with, because AI and chip stocks have sold off recently. As Reuters reports, this sale will test whether the market's love affair with AI survives contact with real numbers. Related reading on this site: OpenAI's training pause just rattled AI stocks and Bill Gates put a billion deaths on the table over AI.

What this means if you never buy a share

You do not need a brokerage account for the Anthropic IPO prospectus to matter to you. No filing has laid out the AI economy this plainly. The Anthropic IPO prospectus is the clearest public look yet at the economics behind the AI tools millions of people use daily, and the economics say the whole thing runs on investor money. The free or cheap AI you use every day was never cheap to make. It was venture capital firms, sovereign wealth funds and Big Tech companies paying the bills, as Reuters describes it, before the public got invited to the table.

There is a political thread too. Anthropic has clashed with the White House over how its tools are used, a fight that led to the Pentagon temporarily blacklisting the company until a US judge blocked the move in August, which we covered when the blacklist survived a court challenge. Now that same company wants public money while keeping founder control, and the Anthropic IPO prospectus warns the technology carries existential risks. That is a strange thing to ask of strangers: fund our losses, skip the voting rights, and trust us with the most powerful technology anyone has built.

None of this means the bet is wrong. Anthropic argues, in the filing as reported by Reuters, that AI will reshape the global economy more deeply than industrialization, electricity and the internet. If that happens, today's figures will look like pocket change. But if you wanted to know what the AI boom actually costs, the Anthropic IPO prospectus just told you: forty-two billion dollars in the red, half a trillion in promised spending, and a two-trillion-dollar asking price. The bill for the future has arrived, and it is itemized.