RBB

By Brett Arends

The first draft of the AI company's prospectus is light on details, but some things immediately leap out.

Anthropic, which was founded by CEO Dario Amodei, is preparing to go public.

Anthropic, the artificial-intelligence company behind the Claude chatbot, is finally going public. The company released the first draft of the prospectus for its initial public offering on Monday, although crucial details - especially anything about the actual stock price or valuation - have yet to be disclosed.

But among what is in this document, two things immediately leap out.

First, the company is already on the hook for over $500 billion in payments over the next few years for hardware purchases and leases from the likes of Amazon (AMZN), Google (GOOG), Broadcom (AVGO), SpaceX (SPCX) and Advanced Micro Devices (AMD).

The second is that while the founders and executives say up front that they have "pledged" to give 80% of their wealth to philanthropic causes, it turns out this isn't a pledge at all.

These commitments are "voluntary" and "revocable," express nothing more than intent, and "do not constitute legally binding obligations," the prospectus admits on page 118.

In which case, they are not pledges, they are mere statements of intent. The repeated assertions at the top of the prospectus that "our co-founders have each pledged to dedicate 80% of their personal equity in Anthropic to philanthropic causes" are effectively meaningless.

This is their money, not mine, so it is their business if they give it away. They are under no obligation to do so, and they deserve full praise if they do.

But it's not a pledge if it's not a pledge. Some cynical people might even wonder whether this nice-sounding offer appears near the top of the prospectus, several times, as a bit of pre-IPO spin. And whether the get-out clause is buried on page 118 in the hope that no one will actually read that far.

They're probably right about that.

As for the expenses: We already knew that AI infrastructure spending is off the charts, but the details in the Anthropic prospectus make for some fascinating reading.

The company has signed commitments to pay at least $114.5 billion over the next seven years to Google, plus $110 billion to Amazon and $44.9 billion to Microsoft, all for use of their respective AI and cloud infrastructures. All payment commitments are "regardless of actual usage."

It has also committed to paying $161.2 billion to Broadcom for data-center and equipment leases.

And another $44.6 billion and $41.3 billion, respectively, to private AI infrastructure companies NScale Global and Lambda.

That's $516.5 billion right there.

There's also a commitment to spend an unspecified amount, expected to exceed $20 billion, with computer chip maker AMD.

And another unspecified amount, "up to $120.5 billion," with SpaceX.

You know what they say: A hundred billion dollars here and a hundred billion dollars there, and pretty soon you're starting to talk about serious money.

Meanwhile, Anthropic's operating cash flow in the first half of this year (see page 168) was $3.9 billion.

There is nothing untoward about these commitments, which are the cost of doing business, and they will work out fine so long as Anthropic's revenue booms, as the company hopes it will.

But they are effective debts that need to be factored in to any calculations of the company's financials or valuations. Those with long memories will remember that similar commitments and other financing arrangements turned out badly when the technology bubble popped at the end of the 1990s.

-Brett Arends

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

10-26-26 1650ET