Anthropic, the AI lab that has spent years casting itself as the cautious rival, is now preparing the largest initial public offering in history. Reuters reported on September 11 that the company is in talks to bring in Nvidia as a cornerstone investor, with a listing that could raise as much as $100 billion at a valuation of roughly $2 trillion. The deal, if it holds together, would be completed before the U.S. midterm elections in November.
The numbers are hard to put in context because nothing like them exists. A $100 billion raise would dwarf every previous record, and a $2 trillion valuation would make Anthropic worth more than all but a handful of publicly traded companies on earth. People familiar with the discussions, who asked not to be named because the talks are private, cautioned that the plan is still being discussed and could change. Neither company would comment.
Nvidia’s role is the part that has drawn the most attention. The chipmaker is considering investing as much as $1 billion in the offering, according to the report. The two companies already have a relationship: in November 2025 they signed an investment and partnership deal worth up to $1 billion, under which Anthropic committed to purchase $30 billion of computing capacity on Microsoft Azure powered by Nvidia chips. A cornerstone stake in the IPO would deepen that tie considerably.
The circularity of the arrangement is not lost on observers. Nvidia supplies the chips, the cloud providers buy the chips to build capacity, Anthropic rents the capacity to train models, and now Nvidia would also own a slice of Anthropic through the listing. A company that depends on Nvidia hardware would have its biggest hardware supplier sitting on its cap table, which analysts said gives the chipmaker unusual leverage over a customer that is also its customer’s customer.
The timing is as notable as the size. Anthropic and OpenAI have both filed confidential IPO paperwork this year, but Anthropic appears to be moving faster toward an actual listing. Completing the deal before the November elections suggests the company wants to avoid whatever uncertainty a change in Washington could bring to the capital markets. An IPO of this scale needs stable conditions, and the election is the clearest source of instability on the calendar.
The rush has its own risks. Going public at a $2 trillion valuation would lock in expectations that are extraordinarily hard to meet, particularly for a company that has been among the loudest voices arguing that the industry should slow down. The tension between Amodei’s public call for pacing the frontier and a record-setting IPO is one that investors and regulators are likely to probe during the roadshow. A company cannot easily argue both that the technology is dangerous enough to pause and that its shares are worth $2 trillion.
The rivalry with OpenAI frames the urgency. OpenAI filed its own confidential paperwork in June and, by Altman’s account, is now steering away from a 2026 listing. Anthropic’s move to go first would give it the first-mover advantage in the public markets, and possibly a chance to lock in a higher valuation before the safety debate further cools investor appetite for AI risk. The two companies have shadowed each other on safety, on fundraising and now on the path to going public.
Analysts said the practical question is whether a $100 billion raise can actually be absorbed. The largest IPOs in history have been a fraction of that size, and a listing of this magnitude would require participation from every major sovereign fund, index manager and long-only investor on earth. The presence of Nvidia as a cornerstone backer is meant to anchor the order book and give other buyers cover to follow.
The money is destined for compute. Training frontier models has become the single most expensive activity in the technology industry, and Anthropic has tied its own plans to a $30 billion commitment of Azure capacity. A $100 billion raise would fund years of that spending without forcing the company back to private investors on the same terms, and it would give Anthropic a war chest comparable to what OpenAI has raised privately. The listing is, at bottom, a financing event for the most capital-hungry business in tech.
The deeper question is whether public markets will accept the valuation private investors have already set. Anthropic’s backers include Amazon and Google, and its private rounds have priced the company at levels that once looked impossible for a business that burns cash on compute. A public listing would put those numbers in front of investors who mark their holdings every day, and who may prove less patient than the sovereign funds and venture firms that have bankrolled the lab so far.
For Nvidia, the stake is small money with strategic weight. A $1 billion investment is a rounding error against the chipmaker’s balance sheet, but it buys a guaranteed seat in the company that is among the most aggressive buyers of AI compute. Locking Anthropic into the ecosystem matters more than the return on the shares themselves. That is the calculation behind a deal that would tie the two companies together through the biggest offering the market has ever seen.
