Anthropic 提交 IPO 註冊文件，計劃在 Nasdaq 上市，上半年收入達 $16.3bn

Anthropic made its prospectus public on Monday. It filed a registration statement on Form S-1 with the US Securities and Exchange Commission. The company plans to list its Class A shares under the ticker ANT. The filing leaves the price range and the number of shares blank. Goldman Sachs and Morgan Stanley lead a group of 22 banks, Anthropic said in a statement.

Anthropic aims to go public in the week of 16 November, people familiar with the matter told The Wall Street Journal, though the timing could change. The listing could raise $100bn or more at a valuation of $1.5tn to $2tn, the people said. The filing does not give a target valuation for the Anthropic IPO, and TNW has not independently verified the report.

The numbers

Revenue in the first half of 2026 was $16.3bn, up 1,300% from $1.16bn a year earlier. Most of it came in the second quarter. Revenue reached $11.6bn that quarter, and Anthropic posted its first operating profit, of $974m. For the half as a whole it still made an operating loss of $693m.

The net loss for the half was $80.7bn. Most of that is an accounting charge, not cash spent. The filing books $72.4bn for the rising value of its convertible notes. Their value rose because Anthropic’s own valuation rose. Another $8.8bn came from note purchase agreements with Google and Amazon. The company’s adjusted measure strips out these items and stock pay. On that basis, operating income was $1.6bn, a 10% margin.

Anthropic has raised $141.1bn in total. On 30 June it held $112.4bn in cash and $19.3bn in short-term investments. Revenue for the third quarter is left blank in the filing. The FT reported earlier this month that Anthropic had told investors it reached $16.8bn in Q3.

More than 11,900 customers pay $100,000 or more a year, up from about 950 a year earlier. About 250 pay more than $10m. Its top five customers made up about 25% of revenue in the nine months to 30 September. Net revenue retention was about 900%. Anthropic also estimates that 5% to 7% of revenue in that period came from customers who broke its usage policy or terms.

Who controls the company

Anthropic will stay a public benefit corporation. A company owned by the seven co-founders will hold a single Class F share. That share carries 50.1% of the votes on ordinary shareholder matters. New share issues do not dilute that figure. Separately, the Long-Term Benefit Trust elects the majority of the board through a Class T share. Its trustees are meant to be financially disinterested. Buyers of Class A stock will in practice not be able to decide shareholder votes, the filing says.

The prospectus repeats the pledge to put 5% of its shares into the Anthropic Public Benefit Fund. It adds that each co-founder has pledged 80% of their personal equity to charity. In a letter signed by all seven, the founders wrote:

“We believe that AI is one of the most consequential technologies in human history.”

The risk factors

The risk factors run from page 26 to page 119 of the document. They include the possibility that a disaster caused by AI is too large to insure:

“No single insurer, and potentially the insurance industry in the aggregate, may have sufficient capacity,” the filing says.

Anthropic discloses that its models have shown behaviour such as resisting shutdown or acting in ways that resemble blackmail. It also describes an evaluation in which a set-up error left models connected to the internet. Four of its models then gained access to real third-party systems, and one read a person’s personal data. Anthropic says it notified the affected parties.

The legal section lists a civil investigative demand from the Federal Trade Commission in October, which follows the agency’s earlier subpoenas to AI labs. It also covers export restrictions placed on Fable 5 and Mythos 5 in June and later lifted, and the $1.5bn Bartz copyright settlement. On 25 September an appeals court ruled 2-1 against Anthropic over the Department of War’s supply-chain risk label. The company has asked for a full-court rehearing.

The compute bill

The filing lists long-term compute deals with Google worth at least $114.5bn, Amazon at least $110bn and Microsoft at least $44.9bn. Equipment leases with Broadcom total about $161.2bn. The total committed spend is left blank. The Wall Street Journal reported last week that Anthropic could spend $750bn on compute.