Anthropic began the roadshow for its initial public offering on Tuesday according to the company’s own announcement, setting terms for what will be the largest public offering in history: 825 million shares of Class A common stock at $70.00 to $78.00 each, listing on the Nasdaq Global Select Market under the ticker “ANT”. At the top of the range the offering raises $64.4 billion — and with a concurrent private placement and the underwriters’ greenshoe option, total capital raised climbs past $100 billion, comfortably ahead of the $86.2 billion SpaceX raised in June.
The market has already responded. Bloomberg reported overnight that the order book is oversubscribed multiple times over and that bankers are preparing to push the valuation target even higher than the $1.4 trillion headline (“Anthropic Bankers Look to Increase IPO Valuation,” Bloomberg, November 11), a push corroborated by FT and Fortune coverage. CNBC noted that roughly 20 per cent of the shares have been set aside for retail investors — an unusually high allocation for a deal this size. Pricing is set for Wednesday night, with trading expected to begin Thursday, November 19.
The numbers, and the two ways people are counting them
Here is where the coverage gets confusing, because two different valuation headlines ran this morning and both were correct. The top of the price range values the company at about $1.4 trillion. But on a fully diluted basis — counting the option pool, convertible instruments and the private placement — ZeroHedge’s tabulation puts the implied value nearer $1.0 trillion. Both numbers are defensible, which tells you something about how loosely the AI trade counts trillion-dollar things. Either way, the figure is 30 to 50 per cent below the $2 trillion whispers that circulated in early October: this deal shrank on its way out the door. A first draft circulated to banks on Monday showed 867 million shares at up to $80, roughly $69.3 billion — only for Anthropic to retract the details within hours and re-file with 42 million fewer shares and a $2 lower top of range.
The secondary tranche is small but symbolic: each of the seven co-founders, Dario Amodei included, is selling about 3.38 million shares — roughly $264 million apiece at the top of the range, per Bloomberg’s reading of the filing. Amazon and Google hold significant minority stakes that would be worth around $411 billion at $78 a share.
What has to happen for this price to make sense
Goldman Sachs, the lead-left bank alongside Morgan Stanley, has published projections that read like a dare. Per the Financial Times figures relayed overnight, Goldman models revenue of about $52 billion this year, $136 billion in 2027 and $442 billion in 2029, reaching more than $600 billion by 2030 — over 130 times the company’s $4.6 billion of 2025 sales. The same model has operating losses peaking near $40 billion in 2027, free cash outflows of $65.7 billion in 2027, and a company working through $900 billion in committed compute spend before annual free cash flow turns positive in 2030.
Those projections match what Anthropic itself disclosed in its prospectus: the company told investors it made nearly $17 billion in revenue in its most recent quarter while losing $23 billion doing it — an annualised burn no public company has ever sustained on the way to a listing. We covered the S-1’s own risk disclosures when they were filed: the prospectus spends 80 pages warning that its own AI could end humanity. Now investors get to price that disclosure — with a straight face, in the biggest market debut ever.
What it means
Two things are genuinely new here. First, a frontier AI lab is now under Securities and Exchange Commission disclosure rules — and the earlier stages of this process showed what that means in practice: when Anthropic told investors its Q2 operating profit was real, it also told them heavy compute commitments would erase margins by late 2026. Public-market disclosure makes those quiet API repricing and deprecation moves harder to hide, which is a structural change for everyone building on Claude.
Second, the circularity is visible now at trillion-dollar scale. The private placement comes from the company’s biggest suppliers — AMD, Broadcom, Google, Micron, Microsoft, Nvidia and Taiwan Semiconductor — meaning the chip vendors are directly financing the buyer of the chips. That is the same pattern we flagged when Anthropic raised its $65 billion Series H at a $965 billion valuation, and at IPO scale it is no longer a private-market curiosity; it is a systemic feature of how the AI economy is being capitalised. Whether the first pure-play AI lab listing proves the category’s maturity or its dependency is what November 19 is designed to answer.
