Anthropic's IPO Filing: $16B Revenue, $80B Loss, One Share

Anthropic, PBC filed its S-1 on October 26, 2026, targeting a mid-November listing on the Nasdaq Global Select Market. The filing reveals first-half 2026 revenue of $16.3 billion — up 1,300% from $1.2 billion a year earlier — alongside a headline net loss of $80.7 billion, of which $72.4 billion is a non-cash mark-to-market charge on convertible notes. The company has also committed up to $750 billion in future compute infrastructure spending.

Those numbers are not typos. They are, however, deeply misleading if you read them the way you'd read any other S-1. Anthropic's filing is the first time the frontier-AI race gets daily public-market transparency, and it forces investors to make a judgment call that until now has been confined to private funding rounds: is building the most powerful AI models a generational business, or the most expensive science project in corporate history?

Key takeaways

• Revenue surged 1,300% year-over-year — from $1.2 billion in H1 2025 to $16.3 billion in H1 2026. Q3 2026 alone hit $16.8 billion, nearly matching the entire first half.
• The $80.7 billion loss is mostly an accounting artifact. Of that total, $72.4 billion stems from non-cash fair-value adjustments on convertible promissory notes as Anthropic's private valuation soared.
• Gross margins flipped from 12% to 85% in a single year — a swing driven more by revenue explosion than cost reduction.
• Customer net revenue retention hit 900%, meaning existing customers are spending nine times more than they did the prior year.
• SpaceX is Anthropic's largest single customer, with a $3.1 billion commitment — an unusual alliance given Elon Musk's public clashes with the broader OpenAI/Anthropic ecosystem.
• Seven co-founders retain 50.1% voting control through a single Class F share, one of the most unusual governance structures ever brought to a public market.

Why does Anthropic show an $80 billion loss?

Because accounting rules require private convertible notes to be marked to their current fair value each quarter. As Anthropic's enterprise valuation exploded during 2025 and 2026, the liability attached to those notes ballooned — on paper. The S-1 itself spells it out: "these net losses included change in fair value of convertible notes, derivative liabilities, and SAFEs of $1.0 billion, $5.4 billion, and $34.3 billion for the years ended December 31, 2023, 2024, and 2025, respectively, and $72.4 billion for the six months ended June 30, 2026." Once the company is public and the notes convert to equity, this line item effectively disappears. The real cash-flow picture is dramatically different from the headline number.

Revenue growth that defies software-company precedent

A 1,300% year-over-year jump from $1.2 billion to $16.3 billion is, by any historical standard, extraordinary. For context, Q3 2026 revenue of $16.8 billion nearly matched the entire first half on its own — meaning the growth curve was still accelerating at the time of filing. Customer net revenue retention of 900% suggests enterprises aren't just trying Claude; they're embedding it and scaling usage aggressively. But a material portion of current revenue comes from other AI labs and venture-funded startups, raising a legitimate question about whether some of this demand is circular — a pattern the AI ecosystem has seen before with Nvidia's investment cycle.

The $750 billion question

Anthropic has committed up to $750 billion in potential compute infrastructure spending. That figure is larger than the GDP of the United Kingdom (roughly $3.3 trillion? No — it dwarfs the GDP of most individual countries and is approximately three times Apple's market capitalization). The company is also building its own data centers under a project called "Anthropic Direct," aiming to escape reliance on Amazon and Google for infrastructure. Anthropic described this as providing "flexible use of capital by allowing us to separate our commitments for data center capacity." In plain English: vertical integration as a survival strategy. These are not optional line items. They are binding commitments that must be funded through future revenue, debt, or equity raises — regardless of whether demand keeps growing at the current pace.

Governance: one share to rule them all

Seven co-founders — led by CEO Dario Amodei and President Daniela Amodei — control 50.1% of voting power through a single Class F share. Dual-class structures are common in tech IPOs, but concentrating majority control in a single share is genuinely novel. Public shareholders will own economic upside but have limited say over company direction. For a company structured as a Public Benefit Corporation with an explicit AI-safety mandate, this creates an unusual dynamic: the founders can prioritize safety research over short-term profitability, and shareholders cannot override them.

Is the valuation justified?

The filing implies a market capitalization in the $300–400 billion range. No company has ever justified that kind of valuation on less than six months of profitability, selling a single core product — Claude — to a customer base that, while growing explosively, is concentrated among enterprises and AI peers. The bull case is straightforward: if Claude becomes indispensable infrastructure for coding, enterprise agents, and general knowledge work, the revenue base could grow into the valuation within a few years. The bear case is equally straightforward: a 1,300% growth rate off a relatively small base is mathematically impossible to sustain, the compute commitments are enormous and binding, and the S-1's own risk factors flag model-safety failures and even climate-change impacts on data centers. Management itself sees fragility. Goldman Sachs and Morgan Stanley are leading the offering.

What to watch next

The pricing window is mid-November 2026. Two things will matter more than the opening pop: first, whether Anthropic discloses customer concentration data that reveals how dependent revenue is on SpaceX and a handful of other large accounts; and second, how quickly those $750 billion in compute commitments start hitting the income statement as actual cash costs. Gross margins of 85% look spectacular today. They will not survive if revenue growth decelerates while infrastructure bills arrive on schedule. The ongoing FTC investigation into AI safety practices at both Anthropic and OpenAI adds another layer of regulatory uncertainty that public-market investors will need to price in from day one.

Educational content only — not financial advice.