Anthropic's $1 trillion IPO tests Amodei's safety-first paradox

Anthropic plans to go public as soon as September at a valuation of at least $1 trillion, up from $965 billion in May, putting CEO Dario Amodei's safety-first philosophy under public market scrutiny.

"The question is: what happens when the stock price drops?" said Eric Ries, a prominent entrepreneur and investor who has tracked Anthropic's capital structure.

Anthropic's annualized revenue reached $9 billion at the end of 2025 and soared to $47 billion by mid-May, surpassing OpenAI. The company says operating profit will turn positive for the first time in the second quarter of 2026, driven by enterprise demand for its Claude models. At $1 trillion, it would trade at 21 times trailing annualized revenue.

Amazon, which holds a 21% stake built through $13 billion in investments, could see that stake worth more than $210 billion at IPO. Anthropic is contractually required to use Amazon Web Services as its primary cloud provider and spend billions on Amazon's custom Trainium AI chips, meaning much of Anthropic's capital expenditure flows back to Amazon.

Valuation Soared 200x in Three Years

Anthropic's rise from a safety-focused research lab to a near-trillion-dollar company is unprecedented in the AI industry. Founded in late 2020 after Amodei and six colleagues left OpenAI over disagreements about AI safety, the company was valued at under $500 million at inception. Its valuation has since grown more than 200-fold.

The growth has been driven by Claude's capabilities in complex software engineering, machine learning, and cybersecurity tasks. The company's employee count of roughly 3,000 is a fraction of Oracle's workforce, yet its revenue run rate now approaches that of the 50-year-old software giant.

Amodei's path to this point has been marked by tension between his stated principles and the demands of competitive AI development. In 2019, while at OpenAI, he delayed Microsoft's $1 billion investment by months over concerns that GPT-3 could reach AGI level. After founding Anthropic, he implemented strict information controls, writing internal memos on a computer physically isolated from the internet.

Safety Commitments Erode as Competition Intensifies

The contradictions have become harder to ignore as Anthropic scales. Researchers found Claude 4 showed concerning capabilities in assisting with biochemical weapons development. Amodei insisted on keeping expensive AI filters that frequently misclassified normal conversations, even as finance teams complained about margin erosion.

But concessions have followed. Anthropic this year removed its commitment to withhold model releases if risk mitigations couldn't be guaranteed, with Chief Science Officer Jared Kaplan saying the company couldn't make unilateral promises while competitors moved at full speed. The company has accepted investments from Qatar and the UAE, despite Amodei's public writings about empowering democratic intelligence agencies. It pays Elon Musk's SpaceX more than $100 million per month for compute, despite public disagreements between the two executives.

These compromises have drawn criticism from influential venture capitalists. Marc Andreessen and David Sacks have accused Amodei of pursuing "regulatory capture" by advocating for AI regulation that would restrict open-source competitors. Bill Gurley called it "the most aggressive regulatory capture I've ever seen."

For investors weighing the IPO, the question is whether Anthropic's growth trajectory can outpace the governance risks. The company's revenue growth — from $9 billion annualized at the end of 2025 to $47 billion by May — suggests demand for Claude is real. But the capital intensity of AI development means Anthropic will need continuous funding, giving investors significant leverage over the company's direction.

Amazon's stake provides a unique lens. The e-commerce giant has already recorded massive unrealized gains from its Anthropic investment, with its "other" net income jumping from $1.1 billion to $53.4 billion in the second quarter of 2026. Amazon's own books already value Anthropic at $965 billion, so a $1 trillion IPO wouldn't create new gains — but a successful listing would validate the investment thesis and potentially allow Amazon to monetize its position.

The IPO will test whether public markets can accept a CEO who openly warns about the dangers of the technology he builds. As AI researcher Nate Soares, who has known Amodei for over a decade, put it: "It takes a certain kind of person to accept these arguments and still go found that AI company."

This article is for informational purposes only and does not constitute investment advice.