Anthropic Files for IPO With $42 Billion Loss and $4.6 Billion Revenue

By Olivia Hart

Anthropic has filed an S-1 for an initial public offering while disclosing a net loss of roughly $42.0 billion for 2025 against revenue of $4.6 billion, according to a Business Insider report from Oct. 26. The company plans to trade under the ticker symbol ANT.

The gap between the two numbers is stark: for every dollar of revenue last year, Anthropic lost roughly nine. Yet the filing has energized investors, because revenue is climbing fast and the company posted a profit on an operating profit basis for the June quarter, according to research firm Semianalysis. In plain words, the market is being asked to price growth now and worry about the losses later.

What does the filing actually disclose?

The S-1 offers the first detailed public look at Anthropic's financial structure and growth pace. The central tension is familiar to AI watchers: massive net losses alongside rapidly rising revenue, with some investors homing in on signals of improving operating profitability. That operating-profit trajectory is one reason some observers see the deal as potentially large.

Semianalysis flagged the scale of fundraising and its comparison against cash and investment assets as the most surprising figures in the document. The firm also delivered a striking assessment: Anthropic's expansion method could be the most efficient business scaling in the history of capitalism. That is a bold claim, and the filing itself — with a $42 billion net loss — offers plenty of evidence for the counterargument.

What are investors watching?

Venture capitalist Sarah Guo congratulated the company shortly after the filing and named three issues she focused on: an inflection point in operating profit, a focus on new discoveries, and a mismatch between usage-based revenue and long-term computing obligations. That last item matters, because revenue that fluctuates with customer usage must service computing commitments that do not.

Guo also raised governance questions. She said she was curious whether a company that has put forward a strong mission would change how it operates under the quarterly-results scrutiny of public markets, and how sentiment might shift when retail investors hold some of the upside. The S-1 does not answer either question.

Investor Vivek Gopalan highlighted a trend visible in the S-1's charts: revenue per $1 of computing is rising. He wrote that one must believe this metric will continue to increase monotonically. In plain words, the investment case rests not only on revenue growth but on computing resources becoming steadily more productive — a projection, not a guarantee.

Who is backing the deal?

Existing investor participation is confirmed. Brad Gerstner said: "the Anthropic team has set another important milestone." He added that Altimeter would participate heavily in the IPO in addition to holding its existing stake. Altimeter co-led Anthropic's Series H investment round in May, making it a long-term holder rather than a newcomer.

How might it reshape AI valuations?

Market participants see the filing as potentially setting the yardsticks for future AI company listings. Ann Bordetsky, a former technology-sector partner at New Enterprise Associates, described the S-1 as "a lot to read" and wrote that capital markets would set a high bar for valuing AI startups. Her caution is a useful corrective to the efficiency superlatives: a high bar means not every AI company will clear it.

The company's safety-first image also drew comment. Dan Shipper, chief executive of AI agent startup Every, joked that the first paragraph of the S-1 did not include a phrase about being able to kill everyone or eliminate jobs — a reference to Anthropic's long-emphasized safety message. He called the listing a victory for humanity.

Anthropic's offering will now test how public markets weigh fast revenue growth against massive net losses, long-term computing cost burdens, and the willingness of insiders to buy more. Whichever items investors weight most heavily during the offering process will likely become the template for every AI listing that follows.