Anthropic, the artificial intelligence company behind the Claude chatbot, is generating revenue at a breakneck pace — and losing money almost as fast. According to the company's S-1 registration statement, publicly filed on 26 October 2026, Anthropic recorded a net loss of more than $80 billion in the first half of this year, one of the largest half-year corporate losses ever committed to paper.
The disclosures, part of a public S-1 registration statement the San Francisco company filed with the SEC on 26 October 2026, offer the first detailed look at the finances of a frontier AI lab. They reveal a company growing at extraordinary speed while taking on infrastructure obligations so large they rival the GDP of entire countries. Anthropic has applied to list its Class A shares on the Nasdaq Global Select Market under the ticker "ANT", with Goldman Sachs and Morgan Stanley leading the offering. The price range and share count were left blank in the filing; the Wall Street Journal, citing people familiar with the matter, reported the company aims to raise about $100 billion or more at a valuation of $1.5 trillion to $2 trillion, with a listing targeted for the week of 16 November.
What the draft filing shows
The headline number is staggering: a net loss of $80.7 billion for the six months ended June 30, 2026, compared with a net loss of $13 billion in the same period a year earlier. That figure, reported by the Journal, dwarfs the combined first-half losses of every money-losing company in the S&P 500, according to one analysis of the filing.
But the headline requires careful disaggregation. Approximately $72 billion of the loss is a non-cash accounting charge reflecting the rising fair value of convertible notes and other financing instruments Anthropic issued during prior fundraising rounds. Because those instruments count as liabilities, their increased value flows through as a loss on paper — without a corresponding outflow of operating cash. Stripping out that charge, the company's underlying net loss was roughly $8.7 billion for the half — still substantial, but a fundamentally different story than the headline suggests. On the company's own non-GAAP measure, Adjusted Net Income, the same half shows a $2.7 billion profit — a figure that also excludes roughly $9 billion of mark-to-market losses on other financial instruments and $2.2 billion of stock-based compensation, as the Wall Street Journal reported from the filing. Adjusted Net Income is Anthropic's own measure, not a GAAP figure, and the $80.7 billion net loss remains the audited headline.
On the revenue side, the growth is genuinely dramatic. Anthropic generated roughly $16.3 billion in revenue in the first half of 2026, up from $1.1 billion in the year-earlier period and more than triple the $4.6 billion it generated in all of 2025. The Journal also reported that third-quarter revenue alone reached $16.8 billion, a figure Anthropic presented to investors in mid-October — though the quarter's roughly 45% growth rate represented a slowdown from earlier quarters, which gave some investors pause ahead of the listing, according to Italian financial daily Il Fatto Quotidiano's reporting on the investor presentation.
The $750 billion compute question
The most consequential disclosure may not be the loss but the spending commitments. The public filing puts Anthropic's committed computing-capacity agreements at more than $475 billion over the next seven to ten years. The draft filing had put the figure at approximately $450 billion in mostly non-cancelable cloud agreements; the public filing's committed figure is higher. The largest named minimums in the filing are at least $114.5 billion with Google Cloud, at least $110 billion with Amazon Web Services, and at least $44.9 billion with Microsoft — each payable regardless of actual use, according to readings of the filing. The S-1 itself leaves the aggregate total of purchase commitments and leases blank, and its figures cover overlapping categories, so they should not be added mechanically.
Additional commitments of up to $86 billion for Nvidia chips purchased through so-called neocloud providers, plus more than $160 billion for Broadcom-facilitated Google chips, push total future infrastructure obligations past $600 billion. And deals still under negotiation could lift the figure toward $750 billion — a sum that would rival the GDP of Switzerland.
This is the structural bet at the heart of the frontier AI business model: that demand for AI models will grow fast enough to justify spending hundreds of billions of dollars on computing capacity before the revenue exists to pay for it. The company signaled it could spend $750 billion in the coming years on obtaining more computing capacity, including deals it is still negotiating. Anthropic spent about $9 billion on computing capacity and infrastructure in the first half of the year, up from $2.3 billion a year earlier, Fortune reported from the filing.
Customer concentration and risk factors
The filing also reveals a narrow revenue base. Around one-fifth of Anthropic's $16.3 billion in first-half revenue came from just two customers, and the company expects a "significant portion" of its revenue will come from a "limited number of customers for the foreseeable future," according to the Journal's reporting. That concentration means the loss of even one major customer could materially dent the growth story the IPO is selling.
The risk-factors section of the draft is sprawling. It includes uncertainty over how Anthropic will prevent its AI models from causing harm and protect itself from the resulting liabilities — from sophisticated cyberattacks to possible blackmail by uncontrolled models. The company also flags uncertain access to the computing power it needs, a notable risk given that its entire business plan depends on securing hundreds of billions of dollars of it.
Why it matters for the AI trade
Anthropic is preparing for what could be one of the largest IPOs in history: the Wall Street Journal, citing people familiar with the matter, reported the company aims to raise about $100 billion or more at a valuation of $1.5 trillion to $2 trillion, which would surpass the $75 billion SpaceX raised in June, the largest IPO on record. Barron's noted that the debut is no longer shaping up as a crowning celebration of the AI revolution but as a major test of it. Barron's noted that the debut is no longer shaping up as a crowning celebration of the AI revolution but as a major test of it.
The timing is awkward. Investors have grown increasingly wary of a potential AI bubble, questioning whether the enormous capital expenditures of the AI labs will ever generate commensurate returns. "The tenor of conversations with clients this week indicated increasing skepticism that the S-1 and IPO could be the definitive catalyst to restart the AI trade," Dauvin Peterson, an analyst at 22V Research, wrote in a note, according to Barron's. "Investors are taking a much more critical look at the business and returns."
Anthropic attempted to woo Wall Street at an investor presentation in mid-October, but the slowing quarter-over-quarter sales growth gave some investors pause, Barron's reported. The public prospectus is now the market's first chance to judge the numbers directly.
What happens next
Anthropic publicly filed its S-1 on 26 October 2026. Between now and the listing — targeted for the week of 16 November — the company will conduct a roadshow, expected to begin in about two weeks, and set a price in an amended filing, all against a backdrop of volatile AI stocks and a market asking harder questions about the economics of frontier AI than it was a year ago. The share count and price range were left blank in the initial filing.
The $80.7 billion loss will dominate headlines, but the more important number for investors is the compute bill: more than $475 billion committed over seven to ten years, with potential spending reaching toward $750 billion once deals still being negotiated are included. That figure represents a bet that AI demand will keep growing at something close to its current torrid pace for the better part of a decade. If the bet pays off, Anthropic's IPO could still rank among the great technology listings. If it doesn't, the draft S-1 will read as the document that marked the moment the math of the AI trade turned.
For readers, the story is a useful window into how the AI economy actually works: staggering revenue growth, funded by staggering capital spending, built on infrastructure commitments that assume the future will look a lot like the recent past. Whether that assumption holds is now a question the public markets will get to answer. Related reading: India's CERT-In warning on AI agents breaching live systems, Meta's Muse Wonder frontier AI model, and Japan's ¥400 billion injection into chipmaker Rapidus.