Anthropic Files Blockbuster IPO Prospectus
Left 50%
Center 0%
Right 50%
1 left · 0 center · 1 right
What happened
Anthropic filed a more than 300-page S-1 registration statement on Monday, the most detailed public account to date of how a pure-play AI laboratory generates revenue, ahead of an initial public offering that reporting placed in late November. The prospectus reported $16.29 billion in revenue and $2.38 billion in cost of revenue for the first half of 2026, an 85% gross margin, compared with $1.16 billion in revenue in the first half of 2025, and showed sequential growth slowing from about 144% in the second quarter to about 47% in the third; it said substantially all cost of revenue is compute and infrastructure and that the firm has historically operated in a compute-constrained environment. Chief Executive Dario Amodei and six co-founders each own more than 200 million shares, and seven co-founders hold a single Class F share equal to 50.1% of voting power in a public-benefit-corporation structure the company said has "no direct public company precedent." Anthropic did not disclose a target valuation, saying it would be "determined by negotiations among us, the selling stockholders, and the representatives," while cited analyst estimates range from $1 trillion to $2 trillion. The company is challenging in court a U.S. Defense Department ban on military use of its AI systems that followed a public dispute with Defense Secretary Pete Hegseth after Anthropic refused the Pentagon unrestricted access to its models.
Sourcing check · No one on the record · 2 articles
No article cites a named source; the reporting relies on people who aren't identified. It started with unnamed sources and none has been named since.
What these terms mean and how they're counted
- On the record · names a source 0 of 1
- On the record means the article cites a source by name for the story's main claim: a person, or an institution speaking officially, such as a company statement or a named agency's spokesperson. The opposite is an unnamed source. The count is how many of the counted articles cite at least one such source; the names are everyone named across them, each once however the articles spell it. Here: No article does.
- Primary evidence · cites a document or data 0 of 1
- Primary evidence means the article cites a record you could check yourself: a court filing or ruling, a dataset or report, a recording or transcript, an official statement or press release, market data. A quote is not a document. The count is how many articles cite one. Here: None cited.
- Credits another outlet 0 of 1
- The article says the story comes from another outlet's reporting (“first reported by Reuters”). Those articles relay the story; they are not independent confirmation. Here: Each article presents its own reporting.
- Unnamed sources 1 of 1
- The main claim is attributed to people who are not identified: “people familiar with the matter”, “officials said”. The count is how many articles do this. Here: e.g. “sources with first hand knowledge of the matter”, “people close to the firm”
- How it developed
- When the first report landed and whether it named anyone; when the first named source and the first document appeared. Here: First report Oct 30, 10:00 UTC by New York Post, with no named source. No article has named one since.
- Not counted 1
- Opinion, analysis and video pieces, and articles too short to assess. Every count above is out of the counted articles. An article republished under a new headline is one article.
This is about who is cited, not whether the story is true. Each source below says what it names. How we count
Omitted — what each side leaves out
Unpacked
Right-leaning coverage of Anthropic's prospectus leaves out what the filing said and follows OpenAI instead. Left-leaning coverage at CNBC cites first-half revenue of $16.29 billion. It also describes a Class F share with 50.1% of voting power. Those disclosures are absent from the NY Post piece, which treats the S-1 as a precursor to an OpenAI listing. A reader of only that account would not know the scale of the business that just filed or who will control it once shares trade.
Left-leaning coverage leads with questions the prospectus still does not answer. Right-leaning coverage leads with OpenAI's timetable and tags Anthropic as the "archrival" already on file.
What operating profit or loss did the S-1 show after compute costs?
Bottom line
You would not learn Anthropic's first-half sales from the NY Post story about its own IPO filing. That $16.29 billion figure ran in CNBC, which read the prospectus as a financial document rather than an OpenAI preview.
The receipts — 4 facts one side's coverage lacks
- First-half revenue was $16.29 billion
- A single Class F share equals 50.1% of total voting power
- First-half gross margin was 85%
- Sequential revenue growth slowed to roughly 47% in the third quarter
The Left View
CNBC treated the filing as unusually comprehensive yet still incomplete for investors, arguing that "a few big questions remain" on revenue durability, compute, control, and price. Quoted analysts asked whether growth can "keep pace with expectations" when "it is unclear how much it costs to generate each additional dollar of revenue as computing demand increases," and said the document omitted "management's expectations around the sequential slowdown in the growth rate." On infrastructure, Nick Patience of the Futurum Group said investors can see compute falling as a share of revenue "but not how much of the forward commitment is earmarked for training the next generation of models versus serving customers," a split that "dictates future margins," and questioned whether costs can keep falling faster than commitments grow. Morningstar's Michael Field said it is unknown "whether investors will be fine with the lack of control implied" by founder voting power and further share classes, "or will this impact the share price." CNBC called the unset listing price "the trillion-dollar question," writing that a high-end outcome "could bring back some of the swagger the AI sector has lost in recent weeks" while the low end could mean "things could get ugly," and noted that Anthropic "has more cash than almost every company in the U.S.," with "never before" a company going public with comparable cash.
The Right View
The New York Post framed the S-1 as the start of a late-November deal that "could value the company as high as $2 trillion, possibly even surpassing SpaceX's record-breaking IPO," and as a catalyst squeezing OpenAI's public-market window. It described Amodei as among the industry's "top alarmists" on AI risk "although that isn't stopping the latter from cashing in," and reported that people close to him say "his recent coyness about the timing of the sale was motivated less by market concerns and more about" the Defense Department ban. Citing Wall Street executives, it said bankers have been "prodding top AI companies to tap the market before a possible AI meltdown hits," with Anthropic competing for capital against a planned Meta sale of $95 billion and OpenAI's private fundraising; it added that the prospectus shows that "while losing money, Anthropic's revenue growth remains strong as it seeks to dominate corporate and other AI usage," whereas OpenAI "has none of these issues with the Pentagon, which would appear to make its IPO path easier."
Our Take (balanced)
Left-leaning coverage's strongest argument is that the prospectus still withholds the variables that set future margins and the public float's value: the training-versus-serving split of compute commitments, the unit cost of incremental revenue, unexplained sequential deceleration, and whether buyers will accept founder voting control in a structure with "no direct public company precedent"—with Field noting that the control question is unanswerable until trading begins. Right-leaning coverage's strongest argument is that the relevant signal is the listing itself: Amodei is moving toward a late-November, potentially $2 trillion offering while safety warnings and a Pentagon fight continue, and named Wall Street executives treat bubble fears and competing supply (including OpenAI's private raise and Meta's stock sale) as closing the IPO window, not unanswered S-1 line items. The central unresolved tension is whether the filing is best read as an incomplete economic and governance disclosure that public investors have not yet priced, or as a competitive rush to public capital that tests whether AI-safety positioning and a Defense Department dispute can sit alongside a blockbuster cash-out.
2 sources
More stories
The week's bottom lines, in your inbox
One email a week: the five stories that mattered and what they actually mean. Free.