Anthropic's IPO prospectus shows the Claude maker lost just over $80 billion in the first half of 2026 — a figure driven almost entirely by non-cash accounting charges — even as revenue climbed to $16.3 billion, according to a review of the filing by The Wall Street Journal. The same filing lays out what may be the most eye-watering number of all: Anthropic has told investors it could spend as much as $750 billion on computing power in the coming years.
The figures arrive as Anthropic edges toward what is expected to be one of the largest public listings in history, with a debut currently targeted for November. For continuing coverage of the filing and the rest of the industry, see our AI news section.
The Numbers Behind the Headline Loss
The $80 billion first-half net loss dwarfs the roughly $13 billion Anthropic lost in the same period last year, the Journal reported. But the composition of that loss matters more than its size. According to the Journal, about $72 billion of it came from accounting charges tied to the changing value of the company's fundraising instruments — obligations that could later convert into stock as Anthropic's valuation has soared. It is a paper expense, not cash that left the building.
The operating picture underneath is strikingly different. Revenue jumped to $16.3 billion for the six months ended June 30, up from roughly $1.1 billion a year earlier, while spending on AI infrastructure nearly quadrupled to around $9 billion. The filing also revises the record for 2025, when revenue grew twelvefold to nearly $4.6 billion and the operating loss widened to $8.06 billion, with $7.33 billion spent on compute and infrastructure alone.
A $750 Billion Compute Commitment
The forward-looking commitment is where the filing reads less like a software company and more like a sovereign infrastructure project. The Journal reported that Anthropic told investors it may spend up to $750 billion over the coming years to lock in computing capacity — a figure that towers over the roughly $518 billion in cloud and compute commitments disclosed in the draft prospectus circulated in September.
That trajectory is the central bet of the listing: that AI revenue growth will outrun one of the largest capital expenditure programs ever assembled by a private company. It also explains why Anthropic's margins, profitability, and funding needs will be scrutinized more closely than those of almost any debutant before it. Notably, the company flagged its own exposure to a potential backlash against the data center buildout that underpins the AI boom, according to reporting cited by the New York Post.
Valuation Expectations Have Cooled
The filing lands at a delicate moment for the offering itself. Reuters reported that investors at a recent Anthropic investor day discussed a valuation of roughly $1.5 trillion — down from the $2 trillion figure previously floated by backers, and roughly in line with the $1.8 trillion at which SpaceX went public in June. The listing is not expected until after the US midterm elections in November.
Analysts told Reuters that the first major AI lab to reach public markets will effectively set the pricing benchmark for the rest of the sector, including OpenAI, which confidentially filed for its own IPO in June. A soft reception for Anthropic would likely compress expectations across the board.
Risk Factors Front and Center
According to the Financial Times, which reviewed an earlier draft of the prospectus, nearly a third of the document is devoted to risk factors — including unusually candid language warning that advanced AI systems could manipulate, blackmail, or otherwise behave in unpredictable ways, and even pose what the company called existential risks to humanity.
The filing also quantifies the concentration risk in Anthropic's business: just two customers accounted for nearly a quarter of 2025 revenue, and many large customers are not locked into long-term contracts. The New York Post, meanwhile, highlighted a human-scale detail from the filing — that CEO Dario Amodei and President Daniela Amodei draw comparatively modest salaries relative to the company's valuation.
What It Means for the AI Sector
Beyond Anthropic itself, the filing is being read as a stress test for the entire AI financing model. As financial analysis from Finimize put it, the $750 billion compute tab may matter more than the valuation talk: if investors believe Anthropic genuinely needs hundreds of billions of dollars of computing capacity, they will price it less like a high-margin software company and more like a capital-intensive project that must raise money repeatedly.
That makes the November debut something closer to a clearing price for the sector — the moment the market decides how much future AI revenue is worth today, with implications for OpenAI's expected listing next year and for the wave of AI infrastructure debt and equity now working through the system. Anthropic, for its part, expects AI to reshape the global economy more deeply than industrialization, electricity, or the internet did, according to the prospectus. Investors will soon get the chance to say what they think that is worth.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
