Anthropic Lists at $83 a Share, Raising $81.3 Billion

Anthropic became a public company on Thursday, November 19, 2026, after its initial public offering priced at $83 per share, above the $74 midpoint of its original range. TechCrunch reported that the offering raised $81.3 billion in total, including $1.95 billion for the company's seven founders.
Each founder sold about 3.4 million shares. At $83, that comes to roughly $280 million per founder before underwriting fees and taxes. At the $74 midpoint, the planned sales would have yielded about $250 million each. Each founder has pledged 80% of personal equity in the company to philanthropy. Anthropic states that these commitments are voluntary, revocable expressions of intent and do not constitute legally binding obligations.
Anthropic itself raised $104.8 billion, according to the same report, with a possible $117 billion if the underwriters buy their full allotment of shares, the extra block they can take to meet demand. The offering included $26 billion of special non-voting Class N stock alongside the Class A common stock. Nvidia bought $10 billion of the Class N shares and Google bought $10 billion. AMD, Broadcom, Micron, Microsoft and TSMC divided the remainder.
After the IPO, Google owns 13.4% of Anthropic and Amazon owns 14.6%, a combined 28%. Amazon obtained its stake by converting a small percentage of promissory notes, essentially IOUs, from its earlier startup investment. Had it converted all of them, the report says, Amazon could have owned nearly 24%.
On the spending side, Anthropic disclosed about $70.7 billion in compute and infrastructure commitments due in the next 12 months, as of the quarter ended September 30, 2026. Compute refers to the processing capacity that trains and runs AI models, supplied largely by data centers full of chips. Think of it as reserving seats years ahead for a venue that does not yet exist. The company also disclosed $451.6 billion in commitments due before the end of 2030, bringing the total to a ballpark of $1.05 trillion to $1.08 trillion. After the offering, the company had nearly $230 billion of working capital, including capital it already held.
The filing and pricing timeline ran over several weeks. Anthropic filed its registration statement on Form S-1 with the U.S. Securities and Exchange Commission on October 26, 2026. It launched the roadshow for the Class A offering on November 10 and priced the offering on November 18. On October 18, the company had announced it would commit a portion of its shares to the Anthropic Public Benefit Fund, a new entity whose structure increases the stake as the company grows.
The product calendar ran in parallel. On September 22, Anthropic introduced Claude Opus 5.5 for agentic coding and knowledge work, at 40% lower running cost than Opus 5 on typical workloads. Claude Haiku 5.5 followed on October 7 as its fastest small model, built for summarization, subagents and browser use. On November 9, Anthropic introduced Claude Fable 5.2 and Claude Mythos 5.2 as its most capable models for technical work in coding, finance and the sciences.
In my view, the Class N buyer list is the detail worth reading most closely. Nvidia, AMD, Broadcom, Micron and TSMC design, fabricate or supply the chips and memory that run large models, while Google, Amazon and Microsoft operate the clouds where those models are served. Each of those parties has a direct commercial interest in how much capacity Anthropic buys, and that is the context to weigh against the commitment figures above.
The broader context here is the scale of the balance sheet. Multi-year commitments of this size are only feasible when a company can finance them, and the offering gives Anthropic public capital and a cash position that its compute roadmap now depends on. The same arithmetic cuts both ways. A $1 trillion-plus infrastructure obligation ties future revenue, public-market expectations and hardware delivery schedules together, and any slippage in one of them will be visible to shareholders quickly.
The philanthropic pledges and the Public Benefit Fund are the other part of the story. Anthropic describes the founder commitments as revocable and non-binding, so their practical weight will depend on what the founders do with their equity over time. The fund's structure, which grows with the company, is a more durable mechanism on paper. For practitioners, the more immediate consequence is that a frontier model developer now operates as a listed company with quarterly disclosure obligations covering its compute commitments. That disclosure is what made the $1 trillion-plus figure visible at all, and it gives customers and competitors a clearer view of the capacity behind the model roadmap.


