Anthropic's IPO Filing Reads Like a Warning Label
Anthropic's IPO Filing Reads Like a Warning Label
Anthropic just opened its books for the first time. The numbers are staggering, the risks are stated in plain language, and the founders are keeping control anyway.
On Monday, October 26, Anthropic publicly filed its S-1 — the document a company files with the US Securities and Exchange Commission before it can sell shares to the public — to list on Nasdaq under the ticker ANT. The headline figures are wild: $16.3 billion in revenue for the first half of 2026, against an $80.7 billion net loss. And the two numbers that every investor actually wants — how many shares, and at what price — are still blank.
That blank is the first thing to understand. According to Decrypt's account of the filing, Goldman Sachs and Morgan Stanley lead a group of underwriters, but the share count and price range have not been set. Fast Company reports an investor roadshow — the tour where a company pitches its shares to big investors — could begin as soon as November 9, with a targeted valuation of around $1.5 trillion. Earlier reports had floated $2 trillion. Nothing is settled until that price is printed.
How do you lose $80 billion and still make money?
Most of that loss never involved cash leaving the building. About $72.4 billion of it is a non-cash accounting charge tied to convertible notes and similar financing instruments — promises that let early backers convert their money into shares later.
Strip that charge out and the picture changes completely. The filing shows an operating loss of just $693 million for the half-year — and, crucially, a $974 million operating profit in the second quarter alone, the company's first profitable quarter ever. Operating profit here just means profit from running the business, before financing charges and taxes. Operating cash flow for the half was positive too, at $3.93 billion.
The growth behind those numbers is genuinely hard to overstate. Revenue went from $41 million in 2023 to $4.59 billion in 2025, then to $16.3 billion in just six months of 2026, up from $1.2 billion in the same period a year earlier. Anthropic has separately told investors that third-quarter revenue reached $16.8 billion — but that figure is investor communication, not part of the filed financials, which run through June 30.
The customers are real, and they keep spending more
Around 250 customers now pay more than $10 million a year, and more than 10,000 pay at least $100,000, according to Fast Company's reading of the filing. Net revenue retention — how much existing customers increase their spending year over year — sits at around 900%.
That customer base is also a concentration risk. Roughly 20% of first-half revenue came from just two customers. If either of those relationships wobbles, a meaningful slice of the fastest growth story in tech wobbles with it.
There's a second dependency that's arguably more important. Almost half of Anthropic's first-half revenue came through cloud marketplaces run by Amazon, Google, and Microsoft. Those three companies are simultaneously Anthropic's computing suppliers, its business partners, and its competitors, and the filing warns access to their capacity could be curtailed.
The warning label, in its own words
Anthropic's risk section runs to roughly 80 pages, and it doesn't pull punches. According to Reuters' coverage of the filing, via Investor's Business Daily, it warns that advanced AI "could pose catastrophic or existential risks to society if not developed and deployed responsibly."
The filing also flags a quieter, more immediate problem. Between 5 and 7 cents of every dollar Anthropic receives comes from users it doesn't want on its platform — people who violate its terms of service or usage policies and whom it presumably intends to remove. That's revenue the company itself is telling you may not stick around.
Then there's the compute bill. Coverage of the filing puts non-cancelable purchase commitments — money Anthropic has already promised to spend on computing capacity — at about $265.6 billion, up from $54.6 billion at the end of 2025. Against that sits a cash pile of $112.4 billion plus $19.3 billion in short-term investments, after raising $141.1 billion in total.
Who actually runs this company after the IPO?
Here's the part that should matter to anyone thinking of buying in. A trust will elect a majority of Anthropic's board, while an entity owned by its seven co-founders will hold 50.1% of shareholder voting power. Public shareholders get the economics, but not control.
So what should you actually take from all this? If you use Claude, or build on it, the IPO means its finances will now be public every quarter — the growth, the compute spending, and the customer concentration will all be visible in the open. If you're watching the AI boom from the outside, this filing is the moment the boom has to show its working. Anthropic has now told the world, in audited detail, that it can grow faster than almost any company in history, that its biggest costs are still climbing, and that its own technology carries risks it can't fully insure against. Whether that's a buy, a pass, or a wait-and-see depends on which of those sentences you believe most.
Sources: Fast Company, Decrypt, Investor's Business Daily, Toolpilot.
