On the evening of October 26, a document of more than 300 pages went up on the website of the U.S. Securities and Exchange Commission. Most people will only read the first few pages, but those pages are enough to turn half a year of Silicon Valley speculation into something on paper.
The company behind Claude has come to knock. It plans to list on the Nasdaq Global Select Market under a three-letter ticker, ANT. In the prospectus, the number of shares to be sold and the price range are both still blank.
A Three-Hundred-Page Invitation
The blanks are not an oversight. Under the U.S. listing process, the registration statement is made public first, and the price range is filled in only before the roadshow. This document explains who the company is; for now it does not say what the shares will cost.
The company is called Anthropic. It was founded in 2021, several members of its founding team came out of OpenAI, and its chief executive is Dario Amodei. It has put more of its effort into enterprise customers.
The underwriter list sits near the back of the document, with Goldman Sachs and Morgan Stanley at the top and 24 banks in total sharing the deal. Twenty-four banks lifting one table is a formation rarely seen in recent years.
The prospectus says only that the company and some existing shareholders will both sell shares in the offering. The old shareholders cash out, the company raises money, and both sums sit in the same basket, a common arrangement when large tech companies go public.
A Curve That Starts at $41 Million
What makes people sit up is the revenue curve. In 2023, the company's full-year revenue was $41 million. In 2024 it was $386 million. In 2025, $4.59 billion.
In the first half of this year, that number became $16.286 billion, against just $1.16 billion a year earlier. More than tenfold in six months, a slope that would look steep in any industry.
In the second quarter alone, revenue was $11.56 billion. More than $10 billion in a single quarter is already the annual scale of a large listed company, and it took only three months.
More important is profit. In the same quarter, the company recorded $974 million in operating profit. It is the first time since its founding that it has made money at the operating level in a quarter, crossing that line earlier than its rival.
The First Profitable Quarter
But the updated filing of November 2 added the third quarter: an operating loss of $567 million for the period. The profit lasted a single quarter, and the break-even line was pushed back again.
Turn to the lower half of the income statement and another number appears: a net loss of $80.7 billion for the first half. A company with more than $16 billion in half-year revenue, losing $80 billion? At first glance it hardly looks real.
The prospectus explains that $72.4 billion of it came from changes in the fair value of convertible notes and similar instruments. As the company's own valuation rose, those instruments had to be revalued on the books, and the revaluation produced an accounting loss, not cash paid out.
In the same document, the operating loss for the half was $693 million, while operating cash flow was positive at $3.93 billion. Only with the three numbers together can you piece together the company's actual breathing rhythm.
Computing power is still the largest expense. In the first half, it spent nearly $9 billion on compute and infrastructure, equal to 55 percent of revenue for the period. That ratio is a ruler for whether this business can walk on its own.
As of June 30, cash and short-term investments totaled $131.7 billion. That number means that even if the offering goes poorly, it can keep going at the current pace for a long while.
The Rival Outside the Door
Pull the camera back and another company comes into view. According to China News Service, OpenAI lost $1.22 for every dollar of revenue in the first quarter this year; in the second quarter, the gap widened to more than $1.80.
The difference between the two companies is not the strength of their models but the shape of their books. One has just begun to post quarterly operating profit; the other's losses grow along with its revenue. Capital markets have always priced these two shapes differently.
According to The Wall Street Journal, Anthropic could raise $100 billion or more this time. The valuation range under market discussion is $1.5 trillion to $2 trillion. Neither figure comes from the prospectus.
If it prices at the top of that range, the company would surpass the $1.77 trillion record SpaceX set in June and become the largest listing in history. Note that this is an "if."
Even the bottom of the range is striking. At $1.5 trillion, it would be in the running for the global top ten by market value the moment it lists, ahead of many older companies that have been in business for decades.
The Price Is Not Written Yet
There is another page in the prospectus, and it is about power. A special class of stock held by the seven co-founders carries 50.1 percent of all shareholder votes. The public can buy shares, but it cannot elect a majority of the board.
There is also a public benefit fund holding 5 percent of the company, with no voting rights and no say in resolutions. The arrangement says it plainly: money may come in, but the steering wheel does not change hands.
The risk disclosures are candid too. The document says it cannot guarantee that revenue will keep growing at this pace. In the first half, two unnamed customers together accounted for about 23 percent of revenue.
Customer concentration is the kind of thing that gets bolded in any prospectus. The bigger the revenue, the more a single pillar matters if it loosens. That is a separate question from how much compute or how many models a company has.
As for when it lists, for now there is only the media account: mid-November. The company has not given a date, and neither has the exchange. Every "as soon as" needs the word "reportedly" in front of it.
As for why now, the document does not say. All that can be seen is the timetable: filing in June, going public in October, a possible listing in November, half a year for a road many companies take two years to walk.
Five years ago, this company was an office and a group of people in San Francisco. Five years later, it stands at Nasdaq's door with a document of more than 300 pages, and on either side of that door stand two very different sets of expectations.
Whether the door opens, when, and at what price, none of that has an answer yet. Only one thing is certain: this document has laid the books of a large model company out in front of ordinary people.
Disclaimer: This article is based on publicly available information and is for reference only. It does not constitute investment advice.