Anthropic's IPO filing devotes 94 pages to what could go wrong

The News

Anthropic has put a 335-page S-1 prospectus before US regulators, the formal first step towards listing its shares, and close to a third of it is devoted to what could go wrong. The filing, submitted on Monday, carries 94 pages of risk factors, according to an analysis by The Verge's Ellen Thomas.

A founders' letter at the front asks buyers to see their purchase as more than a transaction. The same filing warns, on page 46, that powerful models could pose "catastrophic or existential risks to humanity" if built without proper safeguards. On page 63 it concedes the company has a history of net losses and may never sustain profitability.

Anthropic will list as a public benefit corporation, a structure that obliges it to weigh a social mission alongside shareholder returns. The founders frame the heavy disclosure as an "anti-sell", meant to attract patient investors rather than momentum buyers. The S-1 also reveals hundreds of billions of dollars in compute commitments and plans to build its own data centre infrastructure, while third-quarter results are not yet included.

Why It Matters

Prospectuses have been swelling for decades. Jay R. Ritter, a professor at the University of Florida's Warrington College of Business, notes that Apple's 1980 offer document ran to 47 pages with no risk factors at all. Facebook's 2012 filing, at roughly 200 pages with 22 pages of risks, marked a shift. Anthropic's version takes that trend to a new extreme and, Ritter warns, length can bury what matters.

The comparison investors will reach for is WeWork, whose 2019 S-1 was mocked for promising to elevate global consciousness. Ritter draws a line between the two: WeWork was a loss-making property business dressed up as technology, whereas Anthropic is selling a product with explosive revenue growth. The open question is whether that growth survives once the second-quarter spike in token consumption normalises and compute bills keep climbing.

Indian Angle

For Indian retail investors, the listing will be one of the most closely watched offshore IPOs of the year. Residents can buy US-listed shares through the RBI's Liberalised Remittance Scheme, which caps outward remittances at $250,000 per person per financial year, and remittances above the annual threshold attract tax collected at source. Gains are taxed in India as foreign equity, so holding period and rupee movement both shape returns.

The filing is also a benchmark for SEBI. India's own new-age tech IPOs already carry long risk sections under the ICDR regulations, and a US prospectus that openly discusses existential risk and indefinite losses raises the bar on candour that Indian AI firms may eventually be measured against when they approach Dalal Street.

Finally, the disclosure of huge compute commitments matters for Indian AI builders such as Sarvam AI and Krutrim. If the best-funded labs need hundreds of billions of dollars in infrastructure, domestic players relying on the IndiaAI Mission's subsidised GPU capacity face a widening gap. Indian enterprises paying for Claude in dollars should also watch pricing once Anthropic answers to public shareholders.

FAQ

When will Anthropic actually list its shares?

No listing date has been confirmed. An S-1 is the opening registration step with US regulators, and the company still needs to add details such as pricing and, as Ritter expects, more recent quarterly results before the offer opens.

Can Indian residents invest in the Anthropic IPO?

Indian residents cannot usually subscribe directly to a US IPO allocation, but once shares trade they can buy them through overseas brokerage accounts or Indian platforms offering US stocks, using the Liberalised Remittance Scheme. The scheme permits up to $250,000 per financial year, and tax collected at source applies above the annual threshold, though it can be adjusted against final tax.

What does public benefit corporation status mean for shareholders?

A public benefit corporation must balance a stated social mission with shareholder interests, rather than focusing solely on maximising value. For Anthropic, that mission centres on safe AI development. Directors therefore have room to make choices that may trim short-term returns.

Is Anthropic profitable?

No. The S-1 states the company has a history of net losses and may not reach or sustain profitability. Operating losses narrowed in the second quarter, according to the filing, but the company also disclosed very large compute obligations and plans to build its own data centres, both of which weigh on future margins.

This story was reported by The Verge. Read the full original coverage at The Verge.