Anthropic has set an indicative range for its initial public offering that would value the company at up to $1.404 trillion on a fully diluted basis, testing investors’ appetite for artificial intelligence businesses after a turbulent week for the sector. The developer of Claude disclosed the terms in an updated filing on Tuesday, moving forward with a share sale despite weaker markets and a broader retreat from AI investing.

The company plans to offer 825 million Class A shares at $70 to $78 each, with the sale expected to generate between $57.75 billion and $64. The upper end would put its valuation well above the $965 billion reached in a May fundraising, although below the $2 trillion it had reportedly sought a couple of months ago. Those comparisons frame the pricing challenge: Anthropic is seeking a substantial increase over its previous valuation while moderating its more recent ambitions.

The offering comes as the end of a waiting period allows Anthropic to begin formal marketing as early as Tuesday. Last week’s market swings briefly pushed the Cboe Volatility Index above 25, a level at which companies commonly delay flotations, according to Matt Kennedy, senior strategist at Renaissance Capital. Kennedy nevertheless said the scale of Anthropic’s offering and the case for investing in it could distinguish the transaction from a typical listing. The marketing process will test whether that distinction is persuasive enough to overcome the unsettled backdrop.

Read more: Huang’s troubles put Nvidia’s importance to AI in focus

Anthropic’s operating figures provide evidence of rapid expansion alongside substantial financial demands. The company said its revenue run rate stood at $81 billion at the end of October, about $10 billion above its level at the end of September. Revenue for the third quarter was $17 billion, while the nine-month figure was $33.2 billion. These measures describe different aspects of the business: the run rate indicates its recent revenue pace, rather than the revenue actually recorded over the nine-month reporting period.

Against that growth, Anthropic reported a nine-month net loss of $104 billion and compute and infrastructure commitments of $900.9 billion. It held nearly $132 billion in cash and short-term investments. The figures place the scale of its financing needs alongside its commercial progress, giving prospective shareholders more to assess than revenue momentum alone. They also make the listing a test of willingness to support an AI business with extensive infrastructure commitments even as enthusiasm for the wider sector has weakened.

The public offering is accompanied by a separate agreement to sell $26 billion of non-voting Class N shares privately at the IPO price. Anthropic named AMD, Broadcom, Google, Micron, Microsoft, Nvidia and TSMC as buyers, with Google and Nvidia each investing $10 billion. The arrangement adds a group of technology companies to the financing alongside public-market investors. It also distinguishes the private purchasers’ non-voting securities from the Class A shares being offered through the flotation, an important consideration when assessing what each investment provides.

Read more: Europe joins U.S.-mediated Russia-Ukraine peace effort in Abu Dhabi

Anthropic estimates overall net proceeds of between $84.6 billion and $93.7 billion, with the higher figure dependent on underwriters taking up an option for additional shares. Goldman Sachs, Morgan Stanley and other underwriters can purchase up to 123.75 million extra shares at the IPO price. The disclosed transactions are equity financing, combining the public share offering with a separate private stock sale. For this offering, the final price and use of the underwriters’ option will help determine the proceeds received.

Public investors will have limited influence over governance. Anthropic is a public benefit corporation, and its founders hold a Class F share carrying 50.1% of voting power. The Long-Term Benefit Trust will also choose a majority of the board. The structure means the financial scale of the offering should not be confused with a corresponding transfer of control to new shareholders. Buyers of Class A stock will be investing under arrangements that preserve decision-making authority outside the public shareholder base.

The filing also details planned sales by the company’s co-founders. Chief executive Dario Amodei, president and chairperson Daniela Amodei, Tom Brown, Jack Clark, Jared Kaplan, Sam McCandlish and Christopher Olah each expect to sell nearly 3.4 million shares. At the range’s $74 midpoint, those sales would be worth about $250 million apiece. Elsewhere in the sector, rival OpenAI was reportedly valued at $1.25 trillion last week after raising $11 billion, against an initial fundraising ambition of $20 billion, providing another indication of the gap between financing goals and completed transactions.

Anthropic intends to begin trading on Nasdaq around November 19 under the symbol ANT, and it expects retail investors to receive up to 20% of the offering. The immediate focus is whether investor demand supports the proposed range after the recent sell-off. Its revenue growth and technology-company backing offer arguments in favour; its losses, commitments and restricted shareholder influence provide counterweights. Final pricing will show how investors balance those considerations, with a price within the proposed range implying a valuation above the company’s May benchmark.