Anthropic heads for IPO at a valuation of up to $1.404 trillion, with revenue run rate at $81 billion

If you are building a product on Claude, Anthropic will soon become a public company reporting its numbers to investors every quarter. Until now, what we knew about its finances came from leaks and press releases. Now there is an updated IPO filing, packed with details that matter to more than just stock traders.

What exactly is going public

Anthropic is offering 825 million Class A shares at $70–78 each. That amounts to $57.75–64.35 billion. If the shares open at the top of the range, the fully diluted valuation will be $1.404 trillion. In May, at its last funding round, the company was valued at $965 billion. Yet just a couple of months ago, it was reportedly targeting $2 trillion, so the market has clearly tempered its ambitions.

Separately, the company is selling $26 billion of non-voting Class N shares at the IPO price. The buyers are AMD, Broadcom, Google, Micron, Microsoft, Nvidia and TSMC, with Google and Nvidia investing $10 billion each. Anthropic estimates net proceeds at $84.6–93.7 billion. The upper figure includes an additional 123.75 million shares that the underwriters—Goldman Sachs, Morgan Stanley and others—can purchase. The company wants to allocate up to 20% of the offering to retail investors.

Look at the buyers: they represent almost the entire computing supply chain, from chips and memory to manufacturing and cloud services. They are investing in a customer that buys their hardware and computing capacity.

The numbers: rapid growth and enormous commitments

The revenue run rate reached $81 billion at the end of October, roughly $10 billion higher than at the end of September. This is an annualized extrapolation of current sales, not revenue actually earned. Actual revenue was $17 billion in the third quarter and $33.2 billion over nine months.

Alongside those figures are some others. The nine-month net loss was $104 billion, while computing and infrastructure commitments totaled $900.9 billion. The company holds almost $132 billion in cash and short-term investments. The business is growing very quickly, but spending is growing even faster, and it has locked itself into long-term capacity contracts worth many times its entire annual revenue.

For comparison, OpenAI was reportedly valued at $1.25 trillion last week. It raised $11 billion, against a planned $20 billion. Both industry leaders face the same reality: investors are willing to pay, but no longer at any price.

Control and politics

Anthropic is a public benefit corporation. The founders hold a Class F share carrying 50.1% of voting rights, while the Long-Term Benefit Trust elects a majority of the board. Class A shareholders get a stake in the profits but have almost no influence over decisions. The seven co-founders—Dario and Daniela Amodei, Tom Brown, Jack Clark, Jared Kaplan, Sam McCandlish and Chris Olah—are each selling approximately 3.4 million shares. At the midpoint price of $74, that is about $250 million each.

The document also contains a political detail. In many places, the company uses “super intelligence” and “SI” as synonyms for “AI,” citing Trump's September executive order directing government bodies to use the new terminology. A small detail, but one that shows how Anthropic is managing its relationship with the administration.

What this means in practice

First, the supplier you rely on will start publishing financial reports. Quarterly results will show how margins change and how dependent the company is on large customers. That is a useful signal for anyone building Claude into their architecture and budget.

Second, pricing pressure. With compute commitments and losses on this scale, the company will have to demonstrate a path to profitability. That could mean higher prices or tighter usage limits. It makes sense to avoid depending on a single model and keep a fallback option.

Third, the caveats. The price range is preliminary, trading will determine the final valuation, and the listing date currently comes from CNBC reports. Run rate is not the same as revenue, and the $104 billion loss shows that the industry's economics have yet to stabilize. For founders, the takeaway is straightforward: if even a market leader is raising money on more modest terms than it wanted, it is better to replace assumptions of “endless AI funding” with unit economics calculations.