The file that turns hype into accounting
The most important thing about Anthropic’s “secret file” is not that it is long. It is that the 416-page filing forces a private AI story into the language of public markets: revenue, losses, customers, commitments, governance and risk. The headline is astonishing even by AI standards. Anthropic reported $16.29 billion in revenue for the first half of 2026, while a later amendment showed roughly $17.0 billion more in the third quarter alone . That means the Claude maker is no longer being judged only by benchmarks, demos or developer enthusiasm; it is being judged by whether the numbers can survive IPO scrutiny.
The growth curve is the part that feels almost fictional. The company’s earlier trajectory, from tens of millions of annual revenue in 2023 to billions in 2025, now culminates in a single quarter that would look large for a mature public software company. HelloBro’s latest file-based summary says the amended S-1 showed $33.2 billion in revenue for the first nine months of 2026, up from $2.55 billion in the same period of 2025 . It also says third-quarter revenue rose 1,127% year over year and 46.7% from the second quarter . In other words: the story is not just “AI is growing.” It is “one frontier lab is moving at a speed public investors almost never get to underwrite directly.”
Revenue is not the same as easy money
That is where the file gets more interesting. Anthropic’s first-half revenue number came beside an $80.7 billion net loss, though the key distinction is that much of that loss was tied to non-cash fair-value accounting rather than ordinary operating burn . The same summary notes that the third quarter swung to a $567 million operating loss after a second-quarter operating profit of $974 million . That swing is crucial because it separates the “AI prints money” narrative from the infrastructure reality: serving and improving frontier models is expensive, and quarterly profitability can shift quickly when compute ramps.
The filing also reframes Claude as a capital-intensive platform, not simply a chatbot. Anthropic’s compute and infrastructure obligations are described as massive, with recent analysis putting aggregate committed spend at about $895.4 billion, including $447.5 billion before the end of 2030 . Axios likewise framed the company as sitting at the economic center of the AI boom, noting nearly $900 billion in computing and infrastructure commitments involving Amazon, Google, Microsoft and Broadcom . That is the other side of the miracle: a revenue rocket attached to a very large engine.
The IPO still has blanks where the market wants answers
For all the disclosed numbers, the most important IPO mechanics remain unresolved. As of the latest fresh reporting, the filing still had no final price range, no share count and no confirmed first trading day . SyntaxRadar’s November 7 explainer put it plainly: Anthropic has filed to list Class A shares on Nasdaq under the ticker ANT, but the registration is not effective and investors cannot yet buy public shares .
That matters because the missing price range is where narrative becomes demand. A company can discuss a trillion-dollar valuation in private conversations; a public IPO requires institutions to place orders at a stated range. Reuters reported on November 5 that potential investors were questioning whether Anthropic could proceed with its expected roadshow in time for a November listing, with some viewing a recently discussed $1.5 trillion valuation as the high end of expectations . The same report said investors were awaiting the October revenue figure, which was left blank in the updated S-1, to judge whether growth had reaccelerated after a recent slowdown .
Bloomberg’s November 6 report sharpened the repricing. Potential buyers were discussing a $1.2 trillion to $1.3 trillion valuation range, and at least one person familiar with the matter said some were considering a value as low as $1.1 trillion . That is still enormous. But relative to talk of roughly $2 trillion only weeks earlier, it marks a serious shift in bargaining power from issuer to buyer.
The market backdrop changed at exactly the wrong time
Anthropic is trying to go public as the AI trade looks less invincible. Axios reported that the Nasdaq Composite closed down 5.5% on Thursday, November 5, putting it nearly 12% below its October 6 high, while the SOX semiconductor index fell 12.3% . Le Monde described the same moment as a broader rupture in AI sentiment, saying the market turmoil arrived just as Anthropic was preparing to go public and that a once-discussed $2 trillion valuation had already been revised downward to around $1.5 trillion before even that level began to look stretched .
This does not mean the IPO must fail. It means the IPO has become the market’s live test of AI valuation discipline. Axios captured the stakes well: Anthropic’s offering could help the battered AI trade regain its footing, or give investors another reason to run for the exits . In that sense, the S-1 is not just a company document. It is a stress test for the whole chain: chips, cloud providers, data centers, power, enterprise software budgets and AI application demand.
Why coding demand is at the center
The subject context points to enterprise coding demand as a central driver, and the filing’s economics make that plausible. Coding is one of the clearest places where customers can justify heavy AI spend: developers already have expensive labor costs, measurable workflows and direct productivity targets. Claude Code and related enterprise usage turn tokens into a business process rather than a novelty. That helps explain why consumption-based revenue matters: when models become part of software development, research and operations, spending scales with usage.
But the same dynamic creates pressure. If competitors, especially cheaper open models, keep narrowing the quality gap, Anthropic may face price compression even as agentic workloads consume more compute. The filing’s unresolved question is whether Claude’s capability advantage can remain large enough to support premium pricing while the company funds enormous infrastructure commitments.
The real story inside the 416 pages
The file reveals three truths at once. First, Anthropic’s growth is real enough to force Wall Street to take frontier AI seriously as a public-market category. Second, the economics are not simple SaaS economics; they are closer to a hybrid of software, cloud infrastructure, research lab and capital project. Third, the IPO is no longer being evaluated in a vacuum. It is landing during a repricing of the AI trade.
That is why the “secret file” matters. It turns a mythology into a spreadsheet. The spreadsheet is dazzling: $16.29 billion in first-half revenue, roughly $17 billion in one quarter, and a customer base large enough to make Claude feel embedded in the enterprise economy . But it is also sobering: no final IPO terms yet, nearly $900 billion of compute commitments, and investors now debating whether the right valuation begins with a 1 rather than a 2 .
Anthropic may still deliver one of the defining IPOs of the AI era. But after reading the file, the question is no longer whether Claude can grow. The question is whether that growth can outrun the cost of the machine built to produce it.
Comments
Be the first to comment.