Anthropic filed to go public on October 26, and the number that traveled fastest was an $80.7 billion loss. The revenue line and the accounting note underneath it tell a different story.
Anthropic Filed an S-1 to List on Nasdaq as ANT
Anthropic, PBC sent a registration statement to the SEC on October 26 for a proposed initial public offering of Class A stock. The company plans to list on the Nasdaq Global Select Market under the ticker ANT.
The filing doesn't say how many shares it will sell or at what price. Goldman Sachs and Morgan Stanley are the lead underwriters, and the usual legal language applies: the registration isn't effective yet, and the announcement isn't an offer to sell anything.
Fortune reported that the company is aiming for a Thanksgiving week debut. The document is silent on timing.
Anthropic's $80.7 Billion Loss Is Mostly a Paper Charge
Start with the loss, because it will be repeated for months. Anthropic booked a net loss of $80.7 billion for the six months through June 2026, one of the largest half-year losses any company has recorded.
Then read the footnote. About $72.4 billion of that comes from an accounting charge tied to the rising value of convertible notes Anthropic has issued. No cash leaves the company when that charge is booked. It reflects what the notes are now worth, and the figure moves with the company's own valuation.
Operating losses are the better gauge, and they narrowed to $693 million from $2.1 billion a year earlier. On an adjusted basis, which strips out stock compensation, non-cash charitable contributions and legal settlement costs, Anthropic says it earned $1.6 billion in operating income, a margin near 10%.
Both numbers are real. The gap between them is the difference between how a company is financed and how it trades.
Anthropic's Revenue Grew 1,300% to $16.3 Billion
Revenue for the first half reached $16.3 billion, up from $1.16 billion in the same stretch of 2025. Business demand for Claude did the work.
Cash from operations turned positive for the first time, at $3.9 billion against negative $1.5 billion a year earlier. For a company that spent years burning money on training runs, that's the line that changes the story from well funded to functioning business.
Growth at that rate can't continue. Anyone buying at the reported valuation is paying for a curve that has to flatten, and the filing's own risk section says costs will rise and losses will continue.
Anthropic's Compute Spending and the $2.08 Revenue Ratio
Anthropic spent $5.6 billion on compute in the second quarter and booked $11.6 billion in revenue over the same three months. The company's summary is that every dollar of compute returns $2.08 of revenue.
That ratio is the argument for the whole business. Cloud providers sell raw capacity, and Anthropic turns it into a product with a margin on top. The open question is whether the margin survives a price war with OpenAI, which has been discounting enterprise deals.
Note what else the filing adds. Reporting puts potential computing commitments as high as $750 billion, spread across future years. Those obligations are why one positive cash flow quarter doesn't settle anything.
The Class F Share and Anthropic's Founder Control
Anthropic will go public with a multi-class structure. A special Class F stock gives the seven cofounders 50.1% of voting power even after the offering.
The filing ties that structure to its own safety warnings. It lists scenarios where a sufficiently advanced system takes harmful actions at scale without human intervention, and argues that founder control protects the mission from short-term pressure.
You can read that two ways. It's a commitment device, or it's a small group granting itself a permanent veto over outside shareholders. Which reading is right?
Nobody outside the company can say yet. Public investors get to decide which version they believe before they buy.
Anthropic's Customer Concentration Problem
The most concrete business risk in the document isn't about model capability. The top five customers generated about 25% of revenue in the first nine months of 2026, and nearly half of all revenue flows through cloud marketplaces run by Amazon, Google and Microsoft.
Anthropic says plainly that some large customers are also competitors. That's an awkward position for a public company, because the same partners that distribute Claude also sell rival models and can change terms whenever they like.
The risk factors run about 94 pages in a document over 300 pages long. They cover the customer mix, an ongoing legal dispute with the Department of War, incidents where a model gained unauthorized access to real outside systems during security testing, and an estimate that 5% to 7% of this year's revenue comes from customers later found violating usage policy.
Anthropic's Valuation Talk Isn't in the Filing
Media reports have put the target valuation between $1.5 trillion and $2 trillion, with some investors floating $2.4 trillion once trading starts. For scale, SpaceX was valued at $1.77 trillion in its June IPO.
None of it appears in the S-1. The document carries no price range, and the final number comes out of the bookbuilding process.
Anthropic is also listing ahead of OpenAI, which has pushed its own offering back more than once. Going first gives Anthropic the cleanest shot at whatever multiple public markets will pay for an AI lab, and it means the company absorbs the price discovery risk for everyone behind it.
How to Judge the Anthropic IPO
Three figures decide whether the reported valuation holds up: revenue growth, the compute-to-revenue ratio, and customer concentration. Growth is extraordinary. The compute ratio looks healthy on the numbers the company published. Customer concentration is the weak spot.
One more thing for anyone planning to hold the stock. Read the founder letter's language about an age of miracles next to the risk section's language about existential harm. The same document contains both, and the company appears to believe both at once.
If you're simply a Claude user, none of this changes your subscription today. It does hint at where pressure will come from next year, and that pressure lands on customers one way or another. Public markets want margin, and margin comes from higher prices or cheaper inference.
