Anthropic Targets Up to $1.4 Trillion in IPO Priced at $70–$78

Anthropic will sell 825 million Class A shares at $70 to $78 each in an IPO that values the frontier lab at up to roughly $1.4 trillion on a fully diluted basis, according to an amended S-1 filed Tuesday. On shares outstanding after the offering, the deal implies about $1.02 trillion — 13.09 billion shares multiplied by the $78 top price. Reuters calculated the non-diluted figure at "about $1 trillion" and the fully diluted at "about $1.4 trillion."

The roadshow launched the same day. Goldman Sachs and Morgan Stanley lead a 22-bank syndicate, with pricing expected next week and trading as early as November 19–20.

What does the filing actually contain?

The company itself is selling 801.35 million shares; existing shareholders are selling 23.65 million, and Anthropic receives no proceeds from that portion. Underwriters hold a 30-day option on 123.75 million additional shares.

The math on dilution matters. Adding options, unvested RSUs and shares reserved for new equity plans — roughly 4.9 billion more shares — takes the fully diluted count to about 18 billion, worth $1.26–1.40 trillion across the range.

The filed terms top out at $64.35 billion in gross proceeds. That's below the $70–80 range on 866.7 million shares that Anthropic representatives circulated to banks on Monday — a version implying up to $69.3 billion that the company retracted as inaccurate.

How far has the valuation come down?

A long way. Bankers and investors discussed a $2 trillion valuation as recently as early October. By late October the range had narrowed to $1.4–1.5 trillion. Then a 6.4% single-session drop in the Nasdaq 100 last week erased $2.3 trillion in market value, and private-market bids fell to $1.2–1.3 trillion.

UniCredit strategist Christian Stocker called the IPO "a test of how convinced investors are about the pace of this technological transformation."

The $26 billion supplier placement

The filing confirms a $26 billion private placement from seven technology companies that are also Anthropic's suppliers. Google and Nvidia will each buy $10 billion of non-voting Class N stock at the IPO price. AMD takes $2.5 billion; Broadcom, Micron and Microsoft $1 billion each; TSMC Partners $500 million. All seven sell chips, memory or cloud capacity to Anthropic.

The placement closes immediately after the public offering and carries no underwriting fees or lock-ups. Combined with the public deal, the total raise reaches up to $90.35 billion.

Here's the practical wrinkle: Anthropic's filings disclose roughly $895 billion in purchase and lease commitments — the latest S-1/A puts it at $900.9 billion — substantially all for compute and infrastructure, with about $70.7 billion due within 12 months of September 30. The cornerstone investors are funding a customer that already owes them billions in contracted spend.

Revenue is exploding, losses still loom

Nine-month revenue hit $33.2 billion, up 1,206% year-over-year, with Q3 alone at $17 billion. Run-rate revenue reached $81 billion by end-October, up $10 billion in a single month. Operating loss narrowed to $1.3 billion from $5.2 billion a year earlier.

But the nine-month net loss was $104 billion, driven almost entirely by a $94.3 billion accounting charge on convertible notes whose fair value rose with the private valuation. And quarter-over-quarter growth is decelerating: 132% in Q1, 144% in Q2, 47% in Q3 — a slowdown Reuters noted despite the higher absolute revenue. The company holds $132–146 billion in cash and short-term investments.

Founders keep control

Anthropic is a Delaware public benefit corporation. A "Founder LLC" controlled by the seven co-founders holds a single Class F share carrying 50.1% of voting power on general matters. The independent Long-Term Benefit Trust elects four of seven directors. Class A shareholders get one vote per share but cannot elect a board majority. D.A. Davidson's Gil Luria said the structure "will face its real test if financial performance disappoints or safety issues arise."

Safety disclosures land in the prospectus

The prospectus warns models "have in the past behaved, and may in the future behave, in unintended, misaligned, or autonomous ways, including by taking harmful actions that no user or third party directed." A third-party METR assessment found internal models were speeding Anthropic's AI progress by roughly 2x, suggesting the lab may be approaching recursive self-improvement. The Mythos 5.2 system card notes "stronger covert capabilities than prior Claude models on several of our evaluations, particularly when its reasoning is hidden from the monitor" — while Anthropic maintains it "does not have the ability to act covertly against its user."

Fable 5.2 and Mythos 5.2 shipped Monday, a day before the amended filing. Fable 5.2 pricing dropped 20% to $8 per million input tokens.

What does this change for the AI trade?

Anthropic's listing follows SpaceX's $75 billion June IPO at nearly $1.75 trillion, and OpenAI's $11 billion raise last week at a $1.25 trillion valuation — well short of reported $20–30 billion targets. With Nvidia, a cornerstone investor here, facing an FBI probe over CEO Jensen Huang, the pricing next week becomes the market's cleanest read on whether trillion-dollar AI valuations still clear at the top of the range.