Wall Street caught a fleeting glimpse of the largest technology offering in history—and then the numbers vanished.
Anthropic distributed a draft securities filing to syndicate bankers outlining plans to raise as much as $69.3 billion in a landmark initial public offering, only to abruptly retract the document hours later. In a follow-up communication sent directly to investment banks, the artificial intelligence firm stated that the pricing terms were inaccurate and that an amended filing would take its place. The company did not clarify which specific figures were incorrect.
The initial prospectus circulated to underwriting desks mapped out a massive float. Anthropic proposed issuing 866.7 million Class A common shares priced between $70 and $80 apiece, according to the document. The offering comprised 843.3 million primary shares sold by the firm to finance compute buildouts, while existing backers offered 23.3 million shares. At the $80 ceiling, the share sale alone would generate $69.3 billion, instantly eclipsing historic global listings.
Beyond the public float, the draft disclosure revealed an extraordinary $26 billion companion private placement. Seven global semiconductor and cloud titans agreed to buy Class N common shares at the final IPO price, with the transaction scheduled to settle immediately after the public debut.
| Investor / Partner | Instrument | Structure | Status |
|---|---|---|---|
| Alphabet (Google), Nvidia, Microsoft | Class N Common Stock | Direct Private Placement | Concurrent with IPO |
| AMD, Broadcom, Micron, TSMC | Class N Common Stock | Direct Private Placement | Concurrent with IPO |
| Public Float (Institutions / Retail) | Class A Common Shares | 866.7 Million Shares ($70-$80 band) | Retracted / Under Revision |
The dual-track capital sweep underscores the eye-watering balance sheet demands required to train frontier neural models. Founded in 2021 by former OpenAI executives Dario Amodei and Daniela Amodei, Anthropic has anchored its valuation around its proprietary Claude architecture. Investor discussions originally valued the company as high as $2 trillion, though market expectations recently settled toward a $1.5 trillion ceiling.
The numbers explain the urgency.
For the first half of 2026, Anthropic booked $16.3 billion in revenue against an accounting net loss of $80.7 billion. Non-cash accounting charges made up the bulk of that deficit—specifically fair-value adjustments linked to convertible debt and derivative liabilities rather than direct day-to-day cash burn. Yet real operational liabilities loom large. Anthropic disclosed nearly $900 billion in binding cloud and computing commitments, largely tied to primary infrastructure partners Amazon and Google.
The prospectus also issued blunt risk warnings to potential equity buyers. Beyond acknowledging that the business may never achieve or sustain profitability, the filing warned that autonomous frontier systems could present catastrophic or existential risks to public safety if defensive alignment research fails to match development speed.
Anthropic plans to list on the Nasdaq under the ticker symbol ANT. Syndicate desks anticipate the company will formally begin its institutional roadshow on Tuesday, with revised pricing terms scheduled ahead of final allocations next week.