Anthropic has started selling its shares to investors. On Tuesday 10 November the maker of Claude launched the roadshow for its US stock market debut, setting a price range of $70 to $78 a share and asking to list on the Nasdaq under the ticker “ANT”. At the top of that range the sale would raise about $64.4 billion, which would make it one of the largest share offerings ever and, Reuters says, possibly the first public listing by a frontier AI lab. It is the clearest sign yet of how much money is riding on the AI boom.

What Anthropic has announced

According to Anthropic’s own announcement, the offering covers 825 million shares of Class A common stock. Of those, 801,351,354 are being sold by the company and 23,648,646 by existing shareholders, and Anthropic gets no money from the second group. The banks running the sale, Goldman Sachs and Morgan Stanley, also have a 30-day option to buy up to another 123.75 million shares.

The sums are easy to work out. At $78 a share the 825 million shares come to about $64.35 billion, in line with the $64.4 billion Reuters reported. At the bottom of the range, $70, they would raise about $57.75 billion, which is our own arithmetic rather than a figure from the company. If the banks used their extra option at $78, the total would pass $74 billion.

A $26 billion private deal sits alongside it

The public sale is not the whole story. CNBC reports, from the amended filing, that Anthropic has agreed to sell $26 billion of a separate class of stock in a private placement at the same price as the IPO. The buyers are AMD, Broadcom, Google, Micron, Microsoft, Nvidia and TSMC. Most are chip and cloud companies, so the deal ties Anthropic more closely to the suppliers it relies on. CNBC says those shares will not be registered in the offering and will not be locked up.

How big is it really worth?

Here the headlines disagree, and it is worth knowing why. Reuters and The Decoder put the valuation at up to roughly $1.02 trillion, based on the number of shares in issue. CNBC says the company could be worth up to $1.405 trillion on a fully diluted basis, which also counts shares that could be created from options and other instruments. Both are described as the top end of the range, and both are above the $965 billion Anthropic was valued at in a private funding round in May.

Either way, the figure is lower than earlier hopes. CNBC reported last week that Anthropic had been aiming at about $2 trillion until recently, and that even $1.5 trillion was in doubt, with one investor calling it a “stretch”. The Decoder says backers pushed for a lower valuation.

The timetable and the demand

CNBC says, citing people familiar with the plans, that Anthropic intends to price the shares on 18 November and begin trading on 19 November. Those dates have not been confirmed in the filing, so treat them as reported rather than fixed. The same sources say the company is aiming to give about 20% of the shares to retail investors, far more than the 5% to 10% that is typical for a big IPO, and that the order book was already covered at the price range on an allocable basis.

Demand from big institutions, CNBC’s sources say, is stronger than it was for SpaceX at the same stage, in the record listing it completed earlier this year.

The numbers behind the pitch

The new filing adds a few fresh figures. CNBC reports that Anthropic’s annualised revenue was roughly $81 billion at the end of October, up about $10 billion from the month before. For the first nine months of the year, revenue was $33.24 billion, and the net loss was $104 billion, though most of that is an accounting charge linked to the changing value of convertible notes rather than cash spent. We covered the quarterly detail, including the slowdown in growth and the return to an operating loss, in our report on the amended filing.

The biggest risk remains the bill for computing power. The prospectus lists commitments of nearly $900 billion with partners including Amazon and Google, which is why the way the market values the shares matters so much. Our earlier look at the first leaked draft showed how fast that number has grown.

Why the timing is awkward

The roadshow arrives at a tense moment. Last week the Nasdaq fell 5.5% in a day, its sharpest one-day drop in 19 months according to The Decoder, as investors worried that the US midterm results could slow data-centre building; we reported that in our piece on the AI sell-off. CNBC also points to rising interest rates, which make the borrowing behind AI infrastructure more expensive, and to statewide ballot measures that could slow new data centres.

Anthropic has tried to build momentum. On Monday it released Claude Fable 5.2 and Mythos 5.2, its first new frontier models since its chief executive called for the industry to slow down in September, and on Tuesday it added Claude to Google Meet. We covered the models in our Fable 5.2 report. The Decoder, citing Reuters, says the terms were briefly sent to banks on Monday evening before being pulled back, a small stumble before the formal launch.

What it means for UK readers

This is a US listing, and the sources we checked do not say how ordinary UK investors might take part in the offering, so we would not assume they can. If you hold a pension or tracker fund, the effect may be indirect: a very large AI company could end up in the indexes many funds follow, and its share price is likely to become a daily barometer for AI spending, including the data-centre projects now being planned in Britain.

That is a reason to watch, not a reason to buy. The range can change and the sale can be delayed. Nothing here is investment advice.

What to watch next

Over the next week, the things to look for are whether the price range is raised or lowered, whether the 18 November pricing date holds, and how Anthropic’s shares behave on their first day. If the sale goes well, it could open the door for OpenAI and others. If it stumbles, expect more questions about whether the AI build-out has run ahead of the revenue.

Keep exploring