Anthropic was hours away from the biggest moment in its history when it told the banks to ignore the numbers it had just sent them. Late on Monday, November 9, a company representative emailed investment banks a copy of a securities filing that priced the AI company’s initial public offering at $70 to $80 per share, a deal worth up to $69.3 billion. A follow-up email arrived soon after, saying the pricing details were inaccurate and that an updated filing would come later.
Reuters reviewed both emails and reported the retraction on Monday night. Anthropic declined to comment, and the follow-up message did not say which details were wrong. Reuters said it could not determine what was incorrect, which leaves the most anticipated listing of the year without a confirmed price, days before its investor roadshow is due to begin.
What the retracted filing actually showed
The circulated document described a sale of about 866.7 million Class A shares. Anthropic itself would sell roughly 843.3 million new shares, while existing shareholders would sell another 23.3 million in a secondary offering. At the top of the $70 to $80 range, the deal would raise $69.3 billion in gross proceeds, before underwriting fees and expenses.
That would put it among the largest IPOs ever, though still below the roughly $75 billion that SpaceX raised in June. At the top of the range, the implied valuation would be around $1.4 trillion, according to Futurum Equities. Nasdaq Private Market separately puts Anthropic’s private market valuation at about $1.46 trillion.
| Item | Retracted draft (November 9) | Status |
|---|---|---|
| Shares offered | 866.7 million Class A | Retracted, update promised |
| Price range | $70 to $80 per share | Called inaccurate by Anthropic |
| Gross proceeds at top of range | $69.3 billion | Not final |
| New shares from Anthropic | 843.3 million | Per the retracted draft |
| Secondary shares | 23.3 million | Per the retracted draft |
| Private placement | $26 billion from seven tech firms | Reported, unconfirmed |
| Listing | Nasdaq, ticker ANT | Per the public S-1 filing |
One important distinction: the $69.3 billion is a gross figure from a draft the company itself disowned, not money raised and not a confirmed financing target. Until Anthropic files corrected terms, none of these numbers should be treated as the deal.
The $26 billion side deal with seven tech giants
The retracted filing also described a separate $26 billion private placement running alongside the IPO. Google, Nvidia, AMD, Broadcom, Micron, Microsoft, and TSMC had agreed to buy Anthropic’s Class N common stock at the IPO price, with the placement closing immediately after the offering.
That detail matters for two reasons. First, it shows how tightly Anthropic is now bound to the companies that supply its chips and cloud capacity. Second, it is a reminder of the scale of the company’s spending: separate disclosures show $895.4 billion in computing commitments stretching through 2045. An IPO of this size would be, in large part, a way to keep paying for the compute that its models run on.
The revenue picture is strong but cooling. The filing reported revenue of almost $17 billion for the three months ended September 30, 2026, though the quarterly growth rate had declined sharply. That combination, huge revenue and slowing growth, is exactly what public market investors will interrogate on the roadshow.
Roadshow week, and OpenAI’s opposite move
The timing could hardly be tighter. The investor roadshow was expected to begin on Tuesday, November 10, with final pricing the following week. The Wall Street Journal reported that Anthropic is aiming for its shares to start trading on Thursday or Friday of next week. The company’s public registration statement, filed October 26 and amended November 2, still leaves the price range and share count blank, so the corrected filing is now the next real milestone.
The stumble also lands in a week when Anthropic’s products are in the news for other reasons. The company has just shipped new Claude models that are faster and cheaper than their predecessors, and Claude is moving beyond chat into areas like personal finance, where it can now read bank accounts but cannot move money. The IPO story and the product story are pulling in the same direction, toward a company trying to prove it can grow up in public.
The contrast with its biggest rival is sharp. OpenAI closed a funding round worth about $11 billion at a $1.25 trillion valuation and dropped its own IPO plans, choosing to stay private while Anthropic heads for the public markets. One company is raising private money to fund the same AI buildout, the other is asking public investors to pay for it. The market is about to get a very public answer to which approach investors prefer.
What to watch next
Three things will settle the story. The corrected prospectus will show whether the share count and price range change, or only the details around them. The roadshow, starting this week, will test whether institutional investors still want AI exposure at a $1.4 trillion valuation after the false start. And the first day of trading, expected next week, will show whether a retracted filing is a footnote or a warning sign.
For now, the only numbers that count are the ones Anthropic has not yet filed. The $69.3 billion figure made headlines around the world, but it describes a document its own author called wrong. Investors, and everyone else watching the AI boom, should wait for the corrected version before deciding what this IPO is actually worth.
Sources: Reuters, The Wall Street Journal, CNBC, Nasdaq Private Market. Anthropic declined to comment to Reuters.
