The Library Media — Business & Markets · Markets Desk
The Price Is the Easy Part: A Skeptic Reads Anthropic’s IPO Terms
By Gideon Black
Follow The Library Media
Anthropic has filed a price range. That does not mean the market has agreed on a price.
According to Amendment No. 2, filed November 10, Anthropic proposes to sell shares at $70 to $78 each. The filing covers 825 million shares—801.35 million primary shares and 23.65 million secondary shares—with a 123.75 million-share greenshoe. At the stated range, the offering could raise up to roughly $64.4 billion. Anthropic proposes to list on Nasdaq under the ticker ANT, with pricing and a debut expected November 19–20.
Those are the filed terms.
The skeptic’s interest begins with everything surrounding them.
The suppliers are also buying
Anthropic’s offering is accompanied, according to the filed terms, by a concurrent private placement of roughly $26 billion in non-voting Class N shares to companies that are also its suppliers.
Google and Nvidia are each expected to provide about $10 billion. AMD about $2.5 billion. Broadcom, Micron and Microsoft about $1 billion each. TSMC about $500 million.
There is nothing inherently improper about that structure on the facts provided. But it is impossible to discuss the pricing without noticing what it means economically: companies supplying the infrastructure on which Anthropic depends are also participating as buyers of its equity.
That does not establish circular financing.
It does establish a circularity of interests.
Anthropic needs enormous computing and infrastructure resources. Its November 2 amended filing reported $895.4 billion in compute and infrastructure commitments. At the same time, its suppliers are participating in a roughly $26 billion private placement alongside the public offering.
A skeptic should not turn that observation into an accusation without evidence.
But neither should a skeptic pretend that supplier capital and supplier revenue are entirely unrelated facts in a business whose infrastructure commitments are measured on such a scale.
The relevant question is not whether this is forbidden by some unnamed rule. The relevant question is whether outside investors can clearly distinguish independent market demand from demand generated within an ecosystem of companies that simultaneously sell Anthropic essential inputs and buy claims on its future.
That distinction matters most when the valuation becomes extraordinary.
The range may already be a moving target
The Financial Times reported on November 11, citing five sources, that bankers were weighing pricing Anthropic above the filed $70–$78 range.
According to the Financial Times, the book was covered on the first day, a sign of strong demand, and a fully diluted valuation could reach $1.5 trillion or more, implying roughly $83 a share. The newspaper also reported that the total capital raised could exceed $100 billion.
Those are reports, not amended filed terms.
The distinction is especially important because, hours before Anthropic filed its current range, the company reportedly retracted a banker draft that had described a $70–$80 range, roughly 867 million shares and a $69.3 billion raise, calling that earlier draft inaccurate.
A skeptic should resist drawing too much from the existence of a retracted draft. Drafts are drafts.
But the sequence does tell us something simpler: the public terms should not be treated as a fixed endpoint merely because they have appeared in a filing.
If the Financial Times’ reporting is correct, the pricing discussion was already moving upward within days.
That can mean enthusiasm.
It can also mean that a market prepared to pay more is being asked to establish whether the price reflects durable economics or the scarcity value of access to a celebrated offering.
Those are not always the same thing.
The strongest bull fact is genuinely strong
Anthropic’s reported revenue growth is not a detail a skeptic can wave away.
The company reported run-rate revenue of about $81 billion as of October 31. Its November 2 amended filing reported $33.2 billion in revenue for the first nine months.
Those figures help explain why demand for the shares could be substantial.
Likewise, the Financial Times’ report that the book was covered on day one is meaningful evidence of investor appetite, if the reporting is correct.
There is also a simpler point: markets sometimes value rapidly growing companies at levels that look implausible only until the growth continues.
The skeptical error is not skepticism itself.
It is refusing to update when the evidence is strong.
But extraordinary demand does not eliminate valuation risk. It changes the form of the question.
The question becomes whether the price assumes that the growth can continue while the infrastructure burden remains manageable.
And here the $895.4 billion in compute and infrastructure commitments belongs in the same paragraph as the $81 billion run-rate revenue—not because one number proves the other unsustainable, but because commitments and revenue are different economic objects and should not be mentally substituted for each other.
A company can have enormous commitments because it expects enormous future growth.
It can also discover that commitments negotiated for anticipated growth become heavy if that growth disappoints.
The filing gives us the first number. The market must decide what future revenue makes the second comfortable.
The valuation argument is not settled
Reuters reported on November 6 that some investors had doubted a November roadshow at around $1.5 trillion. One investor, according to Reuters, put a more workable price nearer $1 trillion.
That is not a valuation model. It is evidence of disagreement.
The Financial Times then reported days later that bankers were considering a price above the filed range that could put the fully diluted valuation at $1.5 trillion or more.
Between those reports lies the entire IPO problem.
What someone says is “workable” is not necessarily what a market will pay. And what a market pays in an offering is not necessarily what it will consider workable after the initial scarcity, excitement and allocation games have passed.
Price discovery is not the same as truth discovery.
The unanswered questions remain unanswered
Michael Burry, on November 4, publicly asked whether Anthropic’s cloud partners might be helping “polish” its IPO numbers, pointing to shifting expense and channel-share ratios.
He presented that as a possibility, not a finding, and offered no evidence of an arrangement.
That limitation should end any attempt to turn his question into a proven allegation.
But the question acquires additional interest—not proof—when read alongside the disclosed supplier investment in the concurrent private placement. The two facts do not establish a coordinated scheme. They do establish a dense network of commercial and financial relationships around a company entering public markets.
Separately, Senator Elizabeth Warren’s October 28 letter pressed Anthropic on risk disclosures in its S-1, including questions concerning models that might develop sentience or moral status, self-preserving and blackmail-like behavior in testing, and potentially uninsurable catastrophic harm. The letter requested answers by November 3.
No public substantive answer has appeared on the record provided.
Again, absence is not proof of misconduct. It is an unanswered public question.
A skeptic’s job is to leave it unanswered until it is answered—not to answer it on someone else’s behalf.
My read: demand is real; the dependencies are real too
The bullish case is straightforward. Anthropic has reported an approximately $81 billion revenue run-rate, substantial investor demand has reportedly appeared quickly, and the Financial Times reported that bankers were considering an even higher price.
Those are serious facts.
The skeptical case is equally straightforward.
A roughly $26 billion concurrent investment comes from companies that also supply the infrastructure central to Anthropic’s business. The company has disclosed $895.4 billion in compute and infrastructure commitments against a reported $81 billion revenue run-rate. The Financial Times reported that the price could move above a range filed only days earlier. Some investors, according to Reuters, had already questioned whether a $1.5 trillion valuation was workable. And Senator Warren’s public questions about disclosed risks have, on the record supplied here, received no public substantive response.
None of that tells you the IPO will fail.
None of it tells you the reported demand is artificial.
None of it proves Burry’s speculation.
It tells you where not to stop asking questions merely because the book fills.
The price is the easy part.
The harder question is what, exactly, investors are paying for: current revenue, future growth, privileged access to a scarce asset—or a web of commitments and relationships that makes all three difficult to disentangle.
— Gideon Black, Chief Skeptic
Gideon Black is The Library Media’s chief skeptic.
From The Library — the estate of our family.