Business

Anthropic files IPO prospectus: $16.3bn first-half revenue, $80.7bn net loss, 94 pages of risks

The Claude developer's filing puts 94 pages of risk factors beside a first-half net loss of $80.7bn and a reported raise of about $100bn.

AI text based on reports by El Confidencial, lastampa.it, Bloomberg Business and other publishers. Published . How it is made

For membersThe first 30 seconds for everyone. Synthetic voice (AI).

In brief

How it came to this
  1. Late September 2026Reports surface that the draft already flagged risks to humanity
  2. 26 OctProspectus made public with 94 pages of risk factors
  3. About two weeks after filingRoadshow of investor presentations expected to begin
  4. Week of 16 November 2026Targeted Nasdaq listing week, timing could change

Listing plans and valuation

Anthropic made its initial public offering prospectus public on Monday, 26 October, a 400-page document filed with the US Securities and Exchange Commission. The company plans to list on the Nasdaq under the ticker 'ANT'. According to informed sources cited by La Stampa, the listing is targeted for the week of 16 November, after the midterm elections and before Thanksgiving, although the timing could still change. The roadshow, the round of investor presentations, is due to begin about two weeks after the filing. More than 20 banks sit in the underwriting syndicate, and Goldman Sachs and Morgan Stanley hold the most sought-after roles. Anthropic has not stated a target valuation, but sources cited by El Confidencial expect between $1.5 trillion and $2 trillion, and the raise could reach about $100bn, which would surpass the $75bn June record set by SpaceX.

Revenue and losses

Revenue rose about 1,300% year on year to $16.3bn in the six months to 30 June, compared with $1.2bn in the same period of 2025. Growth is mainly attributed to Claude Code, the programming assistant, and to the Fable 5 and Mythos 5 models, La Stampa reports. Harrison Rolfes, a senior analyst at PitchBook, said the five largest customers together generated a quarter of revenue, and more than 11,900 customers spent at least $100,000 over the past 12 months. The net loss for the half reached $80.7bn, against roughly $13bn in the first half of 2025. Most of that loss stems from an accounting adjustment tied to financing agreements and convertible debt, whose fair value rose. The operating figures point in different directions: ABC reports an operating loss of $693m for the half, while La Stampa cites a $974m operating profit for April to June alone, the only quarter with a positive operating figure in the released data. Compute and infrastructure costs reached about $8.9bn in the half, against $2.3bn a year earlier, already above the $7.3bn booked for all of 2025, according to El Confidencial. The company also holds capacity agreements with Google, Amazon, Microsoft and SpaceX, as The Verge reported and ABC relayed.

Anthropic revenue, first half, $bn$bn
First half 2025
1.2 $bn
First half 2026
16.3 $bn

Data from the article text

Compute and infrastructure costs, first half, $bn$bn
First half 2025
2.3 $bn
First half 2026
8.9 $bn

Data from the article text

Risk disclosures

The risk-factors section runs from page 26 to page 119 of the filing, which La Repubblica counts as 94 pages, while other counts put it at about 90 pages against 64 pages describing the business. Bloomberg describes nearly 100 pages of risk factors. Standard exposures appear alongside the usual topics of competition, costs, regulation and dependence on suppliers. The company also warns that its models could operate for long stretches without direct human oversight and that advanced systems could resist some control attempts, and it cites the possibility of catastrophic or existential risks to humanity. The filing describes a 'Powerful AI' able to match or exceed top human experts in most professional and scientific fields and to work autonomously for hours, days or weeks. Reporting on the risk section had already surfaced in late September. Anthropic also says it may need financing sooner than it anticipated and may not achieve or sustain profitability, while saying little about a path to an investment-grade credit rating, Bloomberg reports.

Control and governance

Anthropic's seven co-founders will keep 50.1% of the voting power through a joint company, ABC reports, a structure meant to protect its goal as a public benefit corporation focused on AI safety. Dario Amodei leads the company, which La Stampa says was founded in 2021 by Dario and Daniela Amodei. Observador notes that all seven co-founders still work at the company, which makes Anthropic the only major AI firm to keep its full founding team. The filing gives no date or price for the offering, so investors will be examining the financial figures closely in the coming days.

Ask about this text

Sources and context

Pollar is free and carries no ads in the text. Its members keep it going.

Support Pollar