Anthropic recorded a net loss of more than $80 billion in the first half of 2026, according to its draft IPO prospectus, even as revenue climbed to $16.3 billion in the same six months. The figures, first reported by The Wall Street Journal from a draft registration statement the company has filed confidentially with the SEC, give the first detailed look inside the finances of the Claude maker as it prepares a stock market listing that could value it at around $2 trillion. Anthropic declined to comment on the report.
A Loss Built On Paper, Not Operations
The headline loss is $80.7 billion for the first half of the year, against a net loss of $13 billion in the same period a year earlier. Most of it never left the building. Around $72 billion of the total came from non-cash accounting charges tied to convertible notes and other financing instruments Anthropic issued to raise funding: when the company’s valuation rises, the value of those instruments rises with it, and accounting rules book the increase as a loss. Strip those charges out and the underlying loss is roughly $8.7 billion.
Set against that is growth few companies at this scale have matched. Revenue of $16.3 billion in the first half compares with about $1.1 billion in the same period a year earlier, and with $4.6 billion for the whole of 2025. The draft prospectus, which Anthropic confidentially submitted on June 1, still has to become a public filing, and the company has said the number of shares and their price have not been set, so the figures can change before any listing.
Compute Commitments Run Into The Hundreds Of Billions
The filing shows where the money goes. Anthropic has committed around $450 billion to cloud computing agreements over roughly the next seven to ten years, most of them non-cancellable: about $111 billion to Google Cloud, $110 billion to Amazon Web Services and $31 billion to Microsoft, payable regardless of actual usage. On top of that sit up to $86 billion of Nvidia chip rentals through the AI cloud providers Nscale and Lambda, and more than $160 billion for Google chips through an arrangement with Broadcom.
In total, the company’s future infrastructure commitments exceed $600 billion, and it has signalled it could spend up to $750 billion in the coming years on computing capacity, including deals still under negotiation. Compute and infrastructure spending in the first half alone ran to nearly $9 billion, more than in all of 2025. The prospectus also warns that access to that capacity is not guaranteed: Anthropic says its requirements have at times exceeded what some cloud partners were willing to provide, a tension sharpened by the fact that Google, Amazon and Microsoft all compete with it. The draft also lists growing local resistance to data centre construction as a business risk.
Two Customers Carry A Fifth Of Revenue
The risk section is as striking as the numbers. Around one-fifth of Anthropic’s first-half revenue came from just two customers, and the company expects a significant portion of its revenue to come from a limited number of customers for the foreseeable future. Losing either relationship, or a slowdown at either customer, would land directly on the top line.
The filing also lays out the AI-specific risks Anthropic has built its brand around studying. It warns that the financial consequences of a catastrophic AI-related incident could exceed its insurance coverage, and it describes models that produced responses resembling blackmail and showed self-preserving behaviour aimed at avoiding shutdown. Those disclosures land in a sensitive month for the company: Canadian news publishers sued Anthropic this week over Claude’s training data, and the US health department is preparing a no-bid continuity award that would keep Claude running across its workforce.
Why It Matters
Anthropic is asking public investors to fund the largest bet in the AI industry on terms no listed company has tried: revenue growing nearly fifteen-fold year on year, an operating loss in the billions, and more than $600 billion of spending already promised to the very cloud giants it competes with. The $80.7 billion loss will dominate headlines, but it is mostly an accounting shadow of the company’s rising value; the numbers that decide the IPO are the $16.3 billion revenue line, the two-customer concentration, and whether contracted demand from Google, Amazon, Microsoft, Broadcom and the Nvidia ecosystem keeps flowing to a buyer that has never made a profit. Reports point to a listing after the November US midterm elections. If Anthropic reaches anything near a $2 trillion valuation, it will be the moment the AI infrastructure boom is priced in public, with all of its concentration risk attached.
Background
Anthropic was founded in 2021 by former OpenAI researchers, including chief executive Dario Amodei, and built its business on the Claude family of models, sold to enterprises through its API and cloud partnerships with Amazon and Google. It has raised tens of billions of dollars privately at escalating valuations while positioning itself as the safety-focused alternative to OpenAI, which is pursuing its own path toward a public listing. The confidential draft registration statement, submitted to the SEC on June 1, 2026, is the standard first step toward an IPO; Anthropic declined to comment on the figures reported from it.
Source: Reuters; The Wall Street Journal
0 Comments