Nvidia is in talks for a second $10 billion bet on Anthropic -- this time as an IPO anchor, not just its chip supplier

Nvidia is in talks to invest up to $10 billion as an anchor investor in Anthropic's initial public offering, Reuters reported over the weekend -- a deal that, if it closes, would make the world's most valuable chipmaker a cornerstone shareholder in what both companies' own numbers suggest could be the largest IPO in history. Anthropic is reportedly seeking to raise as much as $100 billion. Both companies declined to comment on the talks, and people familiar with them cautioned the terms could still change.

The talks surface days after Sam Altman ruled out an OpenAI IPO for 2026 while Anthropic pushed ahead on its own listing -- and they sharpen a question that comparison left open: who is actually putting up the money to anchor Anthropic's record-breaking debut. The reported timeline has Anthropic pricing before November's US midterm elections.

An anchor investor buys a guaranteed block of shares before an IPO prices, which is meant to signal to other buyers that a sophisticated, well-capitalized investor has already done the diligence and is willing to commit. For scale: Nvidia's reported $10 billion check alone would be bigger than the entirety of Mistral's EUR 3 billion Series D, Europe's largest-ever tech funding round, closed just three days before Reuters' report on the Anthropic talks surfaced. And a $2 trillion valuation would put Anthropic's IPO above SpaceX's $75 billion raise in June at a $1.77 trillion valuation -- itself already described as the largest private financing on record before any of this.

This would not be Nvidia's first check to Anthropic. In November 2025, Nvidia, Microsoft and Anthropic jointly announced a strategic partnership: Nvidia committed up to $10 billion and Microsoft up to $5 billion, and Anthropic committed in return to purchase $30 billion of Microsoft Azure compute capacity running on Nvidia's Grace Blackwell and Vera Rubin chips, plus additional capacity up to one gigawatt. Two $10 billion Nvidia commitments to the same company, thirteen months apart, are similar in size but structurally different -- one bought compute-partnership terms, the other would buy IPO shares.

The structure invites the same question already asked about Nvidia's much larger position in OpenAI and its stake in CoreWeave: is this independent capital, or a supplier financing demand for its own hardware? (Nvidia announced up to $100 billion for OpenAI in September 2025; Huang later said that figure was "never a commitment," and Nvidia's actual contribution to OpenAI's March 2026 round was $30 billion -- a separate, larger arrangement with its own circularity questions that financial press has tracked closely.) One analysis of the Anthropic talks put the distinction plainly: "An anchor investor is meant to be an independent vote of confidence in a price. A supplier's stake in a customer that is contractually committed to buying its hardware is a different instrument."

What isn't circular is Anthropic's own reported growth. TechCrunch reported the company's annualized revenue run rate reached $9 billion at the end of 2025, $47 billion in May 2026, and $65 billion by the end of July -- a trajectory investors reportedly expect to reach $100 billion to $120 billion by year-end. Those figures come from Anthropic's own disclosures to investors, not from Nvidia or Microsoft, and they're the strongest evidence that whatever Nvidia does with its own money, the underlying demand for Claude isn't manufactured by this specific deal.

Coverage of the same IPO doesn't agree on the number it's actually targeting.

Either number raises the same question a headline figure obscures: what multiple of revenue is actually being priced in, and does Anthropic's own growth make that multiple look ordinary or extreme?

At today's $65 billion run rate, a $2 trillion valuation prices Anthropic at roughly 31 times revenue. If the company actually reaches the $120 billion investors are reportedly modeling for year-end, that same $2 trillion valuation compresses to under 17 times -- a very different-looking number attached to the identical headline price, purely because the denominator moved. Which multiple is the right one to judge the IPO against depends entirely on whether Anthropic's growth keeps compounding at anything close to its 2026 pace.

Nvidia's own CFO has already put a number on how the company frames these bets. On the company's August 26 earnings call, Colette Kress said Nvidia had invested nearly $50 billion in frontier AI labs to date.

“This was a meaningful commitment, but it represented a small fraction of our expected free cash flow over the same period.” — Colette Kress, Nvidia CFO, Q2 FY2027 earnings call, Aug. 26, 2026

There's also a straightforward strategic reason for Nvidia to want in, separate from any circularity question: Anthropic is one of a small handful of customers actually capable of absorbing Nvidia's newest Grace Blackwell and Vera Rubin systems at gigawatt scale, and a deepening financial relationship gives Nvidia a closer view of -- and stake in -- whichever lab ends up capturing enterprise AI spend if the market doesn't consolidate entirely around OpenAI. Betting on more than one frontier lab is, in that sense, a hedge as much as a bet.

None of this resolves the actual, open question: neither Anthropic nor Nvidia has confirmed the anchor talks are real, let alone final. Anthropic's own S-1, whenever it's filed, will settle the valuation and name its investors directly -- until then, every figure in this story, including the ones above, is reported, not disclosed.