A record-sized listing with an unusual caveat
Anthropic, the San Francisco company behind the Claude AI assistant, has moved its blockbuster initial public offering into the roadshow phase, setting an expected price range of $70 to $78 a share for 825 million Class A shares . At the top of that range, the share sale would raise about $64.4 billion before any separate private placement, while Reuters reported that the deal could value Anthropic at up to roughly $1.4 trillion on a fully diluted basis . Scripps News framed the listing as a public-market debut without a clear precedent: a dedicated AI model developer, structured as a public benefit corporation, attempting to enter the market at a valuation more commonly associated with the world’s largest listed companies .
The important detail is that the $1.4 trillion figure is not a simple number printed in the prospectus as a declared valuation. Anthropic’s amended filing sets out the share count, the proposed $70-to-$78 price range, the planned Nasdaq listing under “ANT,” and the related private placement, but outside analyses derive the headline valuation by deciding which shares, options and restricted stock units to include . The Desk noted that the filing itself does not state either a $1.4 trillion valuation or a $90.4 billion combined raise; those numbers are calculations based on the filed terms and Reuters’ fully diluted interpretation . That distinction matters because public investors will be buying into a capital structure whose headline value can look very different depending on whether it is measured on an outstanding-share or fully diluted basis .
What Anthropic is actually selling
Anthropic says the proposed offering consists of 825 million Class A shares, including about 801.4 million new shares sold by the company and about 23.6 million shares sold by existing stockholders . The company would not receive proceeds from the stockholders’ portion of the sale . It also expects to grant underwriters a 30-day option to purchase up to another 123.75 million Class A shares . Goldman Sachs and Morgan Stanley are the lead underwriters, with a broad bank syndicate listed in the registration statement .
The filing and company announcement also describe a concurrent private placement: AMD, Broadcom, Google, Micron, Microsoft, Nvidia and TSMC Partners have agreed to buy an aggregate $26 billion of Class N common stock at the IPO price, contingent on the offering closing . Reuters reported that, combining the public offering and these cornerstone investments, Anthropic could raise up to about $90.4 billion, though some of the IPO proceeds would go to selling shareholders rather than the company itself . The Desk’s arithmetic shows why the figure is conditional: $78 multiplied by 825 million shares is about $64.4 billion, and adding the $26 billion private placement gets to about $90.4 billion before accounting for who receives the cash and whether underwriters exercise their option .
At the midpoint of $74 a share, Anthropic estimates net proceeds to the company from the IPO and concurrent private placement of about $84.6 billion after underwriting discounts, commissions and expenses, or about $93.7 billion if the underwriters’ option is exercised in full . That makes the transaction not just a liquidity event for the AI sector but also a financing mechanism for a company whose future depends on securing enough computing power to keep Claude at or near the frontier .
Growth that looks extraordinary, even by AI standards
The bull case for Anthropic starts with growth. The company’s prospectus says consumption-based revenue rose from $288 million in 2024 to $3.8 billion in 2025 and then to $26.3 billion for the nine months ended September 30, 2026 . Subscription-based revenue increased over the same periods from $98 million to $789 million and then to $6.9 billion . Scripps News reported that Anthropic generated $17 billion in sales in its most recent quarter and $33.2 billion in the first nine months of 2026, compared with $4.6 billion for all of 2025 .
Reuters reported that Anthropic’s run-rate revenue was about $81 billion as of October 31, up $10 billion from September 30, according to the IPO filing . That metric is an annualized snapshot, not a guarantee of future annual revenue. Anthropic’s own filing cautions that run-rate revenue may not be indicative of revenue for any future period and that growth rates are expected to decline as the market matures and the revenue base becomes larger .
The company’s revenue model is built mainly around “pay-for-intelligence,” in which customers pay for usage metered in tokens and priced by model . Anthropic says customers can also buy subscriptions, and some enterprise clients sign annual or multiyear contracts with minimum spend commitments . That structure helps explain why investors are focused not only on the size of Claude’s user base but also on the durability of enterprise usage, the reliability of large customers, and whether consumption can keep rising fast enough to absorb infrastructure costs .
The cost side: compute, leases and commitments
The bear case begins with the same fact pattern that powers the bull case: frontier AI is capital intensive. Scripps News reported that Anthropic is on the hook for nearly $900 billion over the next few years for computer power and data center infrastructure . Reuters similarly said the company has nearly $900 billion in computing commitments with partners including Amazon and Google . Anthropic’s amended filing gives investors a granular view of how large and layered those obligations have become .
As of September 30, 2026, Anthropic listed $319.99 billion in fixed contractual obligations, including $313.82 billion of purchase commitments that primarily relate to cloud service provider hosting and other third-party infrastructure agreements . The same section says the table excludes approximately $451.2 billion in leases executed but not yet commenced, as well as about $105.7 billion of commitments entered after September 30, 2026, primarily for data center leases and hosting or infrastructure services . It also excludes certain short-term commitments, including about $17.4 billion under agreements cancellable within one year and another $6.6 billion of short-term agreements entered after September 30, 2026 .
This is why the IPO is more than a valuation event. It is also a test of whether public investors believe the AI infrastructure cycle can convert enormous upfront commitments into durable revenue and eventually profits. Anthropic had $146.6 billion of liquidity as of September 30, before the IPO and private placement proceeds, including cash, short-term investments and credit facility availability . Still, the scale of the compute obligations means the company will be judged on capital discipline as much as on model performance.
Profitability remains uneven
Anthropic’s financials show a company that is scaling at unusual speed but has not yet delivered consistent operating profit. Scripps News reported that the five-year-old company has had only one quarter of operating profit . The prospectus shows operating income of $974 million for the quarter ended June 30, 2026, followed by an operating loss of $567 million for the quarter ended September 30, 2026 . Anthropic attributed the September operating loss partly to $5 billion of compute under short-term commitments and $2 billion of non-cash charges tied to its donation matching program .
The net-loss picture is larger because of financing and accounting items. The filing shows a net loss of $41.97 billion for 2025 and $104.04 billion for the nine months ended September 30, 2026 . Some of that reflects changes in the fair value of convertible notes, derivative liabilities, SAFEs and partner financing arrangements, but public investors will still weigh the losses against the proposed valuation .
The company is asking markets to value it not on today’s profitability but on the possibility that Claude and future models become a core layer of business software. That is not unusual for high-growth technology offerings, but the numbers are far larger than in a typical software IPO. A valuation near $1.4 trillion implies that investors must believe Anthropic can turn current demand into a defensible, high-margin business after paying for the infrastructure required to serve it.
Governance and investor control
Anthropic’s governance structure adds another layer of complexity. The company is a Delaware public benefit corporation and says its purpose is the responsible development and maintenance of advanced AI for the long-term benefit of humanity . Its prospectus describes a Long-Term Benefit Trust intended to help ensure the company balances financial goals with public-benefit commitments . It also describes special share classes, including Class T and Class F shares, that shape board control and founder voting power .
Immediately before completion of the offering, a Class F share is expected to be issued to an entity controlled by the co-founders, and that share will carry 50.1% of total voting power on matters submitted to stockholders generally, including election of directors other than Class T directors . Scripps News reported that critics point to concentration of power in the ownership structure, while supporters view the design as part of Anthropic’s mission-driven approach to AI safety .
The lock-up schedule is also unusual. Scripps News reported that some non-founder insiders may be able to sell 10% of their holdings on Day One, followed by a staggered series of dates over the first year when non-founder employees may sell more shares . For public investors, that means the deal is not only about whether Anthropic’s valuation is justified; it is also about how much insider liquidity arrives after trading begins.
Why this IPO matters beyond Anthropic
The proposed Anthropic listing will be read as a referendum on the public market’s appetite for frontier AI. Reuters described the offering as a test of Wall Street’s faith in the AI boom after a period of AI-linked stock volatility and growing questions about whether infrastructure spending will generate adequate returns . The answer will influence not only Anthropic’s own stock but also how investors price other private AI labs, cloud infrastructure suppliers and companies tied to data center buildouts.
If the IPO prices strongly and trades well, it could validate the idea that public investors are willing to fund very large AI balance sheets before profits stabilize. If demand is weaker, it could force a reassessment of private AI valuations, especially for companies with heavy compute commitments and concentrated customer bases. Anthropic’s filing makes clear that the company’s opportunity is enormous, but so is the execution burden .
The headline is therefore accurate but incomplete: Anthropic is seeking a valuation that could reach $1.4 trillion, but the market is really being asked to price a bundle of assumptions about growth, compute economics, governance, safety and investor patience. The IPO’s success or failure will help determine whether the next phase of the AI boom is financed with public-market enthusiasm or with much tougher questions from shareholders.