Anthropic is selling into a market that has gotten colder, and it’s doing it anyway. That tells you how much cash it thinks it needs. The company’s nine-month net loss is $104 billion, and its compute and infrastructure commitments sit at $900.9 billion. Those numbers explain the timing better than any market-sentiment argument.
In an updated filing on Tuesday, Anthropic set its terms: 825 million Class A shares at $70 to $78 each. That would raise $57.75 billion to $64.35 billion. At the top of the range, the fully diluted valuation hits $1.404 trillion. That’s well above the $965 billion from its May fundraise. It’s also below the $2 trillion the company was reportedly aiming for a couple of months ago, so the market has already trimmed the ambition.
The more interesting piece is the private placement. Anthropic has agreed to sell $26 billion of non-voting Class N stock at the IPO price to AMD, Broadcom, Google, Micron, Microsoft, Nvidia and TSMC. Google and Nvidia are putting in $10 billion each. Look at that list. It’s chip designers, a memory maker, a foundry and cloud providers. These are the companies Anthropic pays for compute, and now they’re also owners. Together with the public offering, Anthropic estimates net proceeds of $84.6 billion to $93.7 billion. Underwriters including Goldman Sachs and Morgan Stanley can also buy up to 123.75 million more shares at the IPO price.
That circularity is worth watching. Supplier money coming back as customer spending is how much of the AI buildout has been financed, and critics have flagged it for a while. It props up demand figures and ties everyone’s fortunes together. If the revenue keeps climbing, it looks like smart alignment. If it stalls, it looks like a closed loop.
On revenue, the story is strong. Third-quarter revenue was $17 billion, and nine-month revenue was $33.2 billion. Anthropic holds nearly $132 billion in cash and short-term investments. But the gap between $33.2 billion in revenue and a $104 billion loss is huge, and public investors will be asked to underwrite it for years.
They’ll also get little control. As a public benefit corporation, Anthropic gives its founders a Class F share carrying 50.1% of the voting power. The Long-Term Benefit Trust elects a board majority. Class A buyers are essentially buying economics, not influence. That’s a harder sell than the Google or Meta dual-class setups, because the trust adds a mission layer that can override pure shareholder returns.
The comparison with OpenAI is unavoidable. Its biggest rival was reportedly valued at $1.25 trillion last week after raising $11 billion, nearly half the $20 billion it first sought. So both labs are finding that investor appetite has limits. Still, Anthropic is going to public markets while OpenAI stays private, and that gives it a transparency and liquidity edge, along with quarterly scrutiny OpenAI can avoid for now.
Trading is expected to begin close to November 19 on the Nasdaq under the ticker ANT, per CNBC. For developers and founders building on Claude, the practical takeaway is simple. A well-funded Anthropic can keep buying compute and pricing aggressively against OpenAI and Google. A badly received IPO would do the opposite. This one is not noise. It’s the first real public-market test of whether frontier AI economics hold up.
