Special Report: What Investors Are Buying in Anthropic’s IPO

SPECIAL REPORT

Anthropic PBC is about to ask public investors for more money than almost any company has ever raised in a stock offering, and it is doing so with its terms still in flux. The maker of the Claude AI models was expected to begin formally marketing its initial public offering as soon as Tuesday, a day after a draft filing that briefly circulated among its banks showed a target of up to $69.3 billion and was then pulled back as inaccurate.

What buyers get, if the deal goes ahead as planned before Thanksgiving, is a stake in one of the fastest-growing businesses on record, attached to nearly $900 billion of computing commitments, a governance structure that keeps public shareholders far from control, and a shareholder list that doubles as a roster of its biggest suppliers. This report walks through what Anthropic has filed, what has only been reported, and what investors will be watching as the roadshow begins.

What the draft showed, and what it didn’t

Late Monday, a representative of Anthropic emailed banks a copy of a securities filing that showed the company selling about 866.7 million Class A shares at $70 to $80 each. About 843.3 million of those were new shares from the company and about 23.3 million came from existing holders. A follow-up email said the pricing details were inaccurate and that the filing would be updated. It didn’t say which figures were wrong, and Anthropic declined to comment. MERGER MONDAY reported the details of that draft on Tuesday morning.

Nothing in the draft is on the public record. As of early Tuesday, the latest version of Anthropic’s registration statement on the Securities and Exchange Commission’s database was still the amendment filed Nov. 2, which leaves the price range and share count blank. A further amendment with those terms typically appears when formal marketing starts.

The draft was smaller than the plan described just a day earlier. People familiar with the matter had said the company aimed to raise $80 billion to $100 billion at a valuation of $1.25 trillion to $1.75 trillion. Last week, after a steep selloff in technology stocks, some prospective buyers were discussing $1.1 trillion to $1.3 trillion, down from talk of $2 trillion in early October.

The draft’s most revealing detail may have been the side deal. Alphabet’s Google, Nvidia, Advanced Micro Devices, Broadcom, Micron Technology, Microsoft and Taiwan Semiconductor Manufacturing had agreed to buy $26 billion of Anthropic’s Class N common stock at the IPO price, in a private placement set to close right after the listing. Added to the top of the draft range, that would bring total new money to roughly $95 billion.

The growth story

The pitch rests on revenue growth with few precedents at this scale. Anthropic’s filings show revenue of $16.29 billion in the first half of 2026, up from $1.16 billion a year earlier, and about $17 billion in the third quarter alone, roughly twelvefold the year-earlier quarter. Revenue for the first nine months totaled $33.24 billion.

The company’s annualized revenue run rate has since reached about $81 billion, up from about $65 billion at the end of July, according to people familiar with the figures. That number is not in the filing.

The growth is slowing in percentage terms, however. Third-quarter revenue rose about 47% from the second quarter, after sequential gains of 144% in the second quarter and 132% in the first. Investors setting a price will have to decide how much of that deceleration is the natural result of a much larger base and how much signals a ceiling.

The customer base is broadening. As of Sept. 30, Anthropic counted more than 11,900 customers generating at least $100,000 of trailing 12-month revenue, more than 1,900 above $1 million and about 250 above $10 million, according to its original registration statement.

The compute bill

The other side of the ledger is the cost of the computing power that trains and runs Claude. Compute and infrastructure expenses reached $9.9 billion in the third quarter and $14.4 billion for the first nine months, up 329% from a year earlier. Anthropic swung to a $567 million operating loss in the third quarter, after $974 million of operating income in the second, mainly because it bought about $5 billion of short-term computing capacity at prices well above its multiyear contracts.

The headline losses look far larger, but most of them aren’t cash. The company reported a net loss of $80.75 billion for the first half, of which about $72.4 billion reflected changes in the fair value of convertible notes, derivatives and similar instruments whose recorded value rises along with the company’s valuation. The third-quarter net loss was $23.29 billion, again driven mostly by those marks. Some of those notes convert to stock at the IPO, which should take part of that charge out of future results.

The commitments are the bigger number. The amended filing tallies about $895.4 billion of purchase commitments and leases, up from about $417 billion in the original filing, stretching as far as 2045. Nearly half falls due by the end of 2030, including about $69.8 billion in the 12 months after Sept. 30. Anthropic has agreed to spend at least $110 billion with Amazon Web Services through 2036 and more than $110 billion with Google Cloud through 2033, largely regardless of how much capacity it actually uses.

Lenders are carrying part of the load. On the day Anthropic filed publicly, Bank of America, Blackstone, Citigroup and Morgan Stanley committed $63 billion to a special-purpose vehicle that will buy Broadcom chips and networking gear and lease them to Anthropic. Against those obligations, Anthropic had about $146.6 billion of cash and other liquid assets at the end of September.

Who owns Anthropic, and who controls it

Public investors will own economics, not control. Anthropic is a public benefit corporation, which requires its board to balance shareholders’ financial interests with its stated purpose of developing advanced AI responsibly. Its Long-Term Benefit Trust, a body of financially disinterested trustees, will hold Class T stock that generally lets it elect a majority of the board.

A limited liability company owned by the seven co-founders, including Chief Executive Dario Amodei and President Daniela Amodei, will hold a single Class F share carrying 50.1% of the voting power on most other matters. Employees other than the founders will receive Class E stock with narrow voting rights. Class A shares sold in the IPO carry one vote each, and Class N shares held mostly by strategic partners carry none. The structure makes Anthropic a controlled company under Nasdaq rules, which exempts it from some board-independence requirements.

The founders also have large economic stakes. The filing shows each of the seven holding roughly 232 million to 244 million shares, after a 10-for-1 stock split completed in September.

Big Tech’s double role

Amazon and Google are Anthropic’s largest outside backers, and the filing says each will hold more than 5% of its outstanding stock after the IPO, in nonvoting Class N shares. Amazon has said in its own filings that it has invested $18 billion in Anthropic since 2023 and has agreed to provide up to $15 billion more. Much of Amazon’s position sits in convertible notes that turn into stock at the listing, and the gains on those holdings have already flowed through Amazon’s reported earnings.

Google invested about $3 billion in earlier rounds and said in April it could put in as much as $40 billion more if Anthropic hits certain milestones. It also holds an option to buy up to $10 billion of Class N stock at the IPO price. Microsoft and Nvidia agreed in November 2025 to invest up to $5 billion and $10 billion, respectively, as Anthropic committed to buy $30 billion of computing capacity on Microsoft’s Azure cloud.

That overlap is the heart of the bull and bear cases alike. The same companies that would profit from a higher share price also sell Anthropic the chips and cloud capacity it needs, and several of them, Google in particular, build competing AI models. A higher IPO valuation lifts the carrying value of their stakes; Anthropic’s ability to meet its compute commitments supports their cloud and chip revenue. Critics of the AI boom call such arrangements circular. Supporters say they lock in supply for a company whose growth is limited mainly by how much computing power it can get.

How it stacks up

At the top of the retracted range, Anthropic’s offering would rank as the second-largest IPO on record. SpaceX raised $75 billion in June, or about $86.2 billion including the overallotment option. Before that, the record belonged to Saudi Aramco, which raised $25.6 billion in 2019, growing to about $29.4 billion with its greenshoe, followed by Alibaba’s $21.8 billion New York listing in 2014, which grew to about $25 billion. Including the $26 billion placement, Anthropic’s total raise could exceed SpaceX’s base deal.

The closest private-market comparison is its chief rival. OpenAI closed $11 billion of new funding at a $1.25 trillion valuation on Friday, the first piece of a round that could reach about $20 billion, below the $30 billion and $1.5 trillion it first sought. OpenAI hasn’t set a timetable for its own listing. Anthropic was valued at $965 billion in a private round in May. A public valuation of $1.1 trillion to $1.3 trillion would put the two companies roughly level, but Anthropic’s price will be set daily by the market rather than negotiated with a handful of investors.

The risks

Compute costs. The largest cloud contracts, with Amazon, Google and Microsoft, generally can’t be canceled outside of an uncured material breach and must be paid whether or not the capacity is used. If revenue growth slows faster than expected, those fixed obligations would weigh heavily on results.

Decelerating growth. Quarterly growth fell from 144% to 47% in a single quarter. Anthropic’s valuation assumes that run-rate gains of the kind reported since July continue for years.

Competition. Anthropic competes with OpenAI, with Google’s own AI lab and with other well-funded developers, including some of its own suppliers and shareholders.

Supplier concentration. A few cloud and chip providers account for most of its computing capacity, and the largest of them are also rivals.

Regulation and politics. Investors are weighing the prospect of tougher oversight of AI after Democrats won control of Congress in last week’s midterm elections.

Governance. Public shareholders won’t be able to elect a majority of the board or outvote the founders, and the public-benefit mandate allows directors to weigh interests other than the share price.

Market conditions. The Nasdaq 100 fell 6.4% last Thursday, one of its worst sessions on record, as investors questioned the cost of the AI data-center buildout. A deal this large depends on that mood stabilizing.

What it means for deal flow

A deal of this size is a test of how much AI risk public markets will absorb at once. Bankers have positioned the offering to reach beyond institutions: Anthropic has weighed setting aside about 15% to 20% of the shares for individual investors, with JPMorgan Chase running U.S. retail sales and Citigroup handling retail demand in markets including the U.K., Canada, Japan and Australia, according to people familiar with the matter.

The outcome will set a reference price for every other AI developer considering a listing, OpenAI included, and for the private rounds being priced in the meantime. It also tests the financing model behind the AI buildout: supplier equity, chip-backed lease vehicles and multiyear cloud contracts. A strong debut would encourage more of all three. A weak one would make the next round of data-center debt and AI equity harder to place.

Until Anthropic files its range, every number above about the offering’s size and price is a target rather than a term. MERGER MONDAY will update this report when the amended registration statement is filed.

Filed under: Spartanburg SC Business News