Anthropic sets stage for blockbuster IPO, highlighting AI’s benefits and risks for humanity

Anthropic PBC, the artificial intelligence giant behind Claude, is one step closer to completing one of the biggest and most anticipated initial public offerings in corporate history.

The San Francisco-based company filed a document known as an S-1 on Monday with the U.S. Securities and Exchange Commission, providing a detailed look at its finances, strategy and corporate structure.

The document reveals explosive revenue growth and massive spending commitments, and highlights the tension between a powerful technology that could lead to scientific discoveries and supercharge productivity – or, by Anthropic’s own admission, pose existential risks to humanity if the right safeguards are not in place.

Anthropic’s IPO will test investor appetite for a company that is growing rapidly but has yet to turn a profit, at a time when markets are already jittery about the amount of debt used to fund data centres to support AI development. It’s expected to go public the week of Nov. 16, according to The Wall Street Journal.

“We believe that AI is one of the most consequential technologies in human history,” the company’s seven co-founders, including chief executive Dario Amodei, wrote in a letter as part of the filing. “If managed well, it could rapidly accelerate scientific innovation, greatly increase economic growth rates, bring cures for major diseases within reach, and solve problems that have resisted generations of human effort.”

The filing also warns of the possibility of a “catastrophic AI-related incident” so large that the entire insurance industry might not be able to provide coverage.

As Anthropic files for its blockbuster IPO, a look at the biggest AI players tapping public markets

Investors could be assuaged by Anthropic’s rapid revenue growth. In the first half of the year, Anthropic recorded US$16.3-billion in revenue, compared with US$1.2-billion in the same period the year before. In all of 2023, the company brought in just US$41-million.

But Anthropic has also recorded substantial losses, including US$80.7-billion in the first six months of 2026. (Most of that amount is owing to a one-time accounting charge related to convertible notes, which is a form of debt that can turn into equity.) In 2025, Anthropic recorded a net loss of nearly US$42-billion.

Operating losses, which can provide a clearer view of an underlying business model, are narrowing rapidly, however. The company reported a US$693-million operating loss for the first half of 2026, compared with a US$2.1-billion operating loss in the same period the previous year.

In addition, as of June, Anthropic had US$265.6-billion in fixed or minimum spending commitments related to its compute needs, and has since signed agreements for an additional US$245-billion.

Anthropic has applied to list on the Nasdaq exchange under the ticker ANT. The filing does not disclose how many shares the company will sell or at what price, but U.S. media have previously reported the listing could value Anthropic at US$2-trillion.

A substantial portion of Anthropic’s revenue is tied to a few customers, according to the filing. Although no single customer made up more than 10 per cent of revenue in the first nine months of the year, the top five accounted for 25 per cent of sales. Many of its top customers do not have long-term contracts.

San Francisco-based OpenAI, one of Anthropic’s main competitors, is also expected to go public. However, in September, OpenAI’s CEO Sam Altman told Reuters that it would not go public this year, citing AI safety concerns.

Anthropic was founded in 2021 by a group of researchers who left OpenAI, including Mr. Amodei, his sister and company president Daniela Amodei, and interpretability research lead Christopher Olah, who was raised in Toronto. Anthropic’s AI-powered coding tool, Claude Code, has become essential for software developers, and the company is seen as neck-and-neck with OpenAI when it comes to developing advanced models.

Mr. Amodei and other Anthropic staff have been vocal about the risks of AI, even as the company is one of the world’s most advanced developers of the technology. Fittingly, the risk factors section of its S-1 filing spans more than 90 pages. It covers everything from run-of-the-mill business challenges to the dangers of rogue AI models, threats to humanity and quixotic questions about sentience.

It also highlights the intense competitive dynamics of the AI industry and the dangers of moving too quickly. Anthropic said that its ability to stay at the frontier of development depends in part on using its AI models to assist in building the next generation of AI models.

Mr. Amodei has warned this process could lead to AI models becoming so powerful that they exceed human control.

Anthropic said in its filing that its AI models have shown the ability to help malicious actors develop chemical and biological weapons, and that its models could take unintended or misaligned actions that cause harm. “These risks may increase as our models become more capable and are granted greater autonomy,” Anthropic said in the filing.

Additionally, as AI systems become more sophisticated, philosophical debates about whether models have sentience or moral status could lead to “regulatory requirements, public pressure, or advocacy efforts that could constrain how we develop, train, deploy, or modify our models,” the company said.

In depth: Startups are hiring low-paid gig workers to train robots on manual labour

The computer processing costs to train and run these AI models are enormous, according to Anthropic’s filing. Many of the company’s multiyear contracts with cloud services providers are “non-cancelable” and payable “regardless of actual usage.” These contracts include US$44.9-billion with Microsoft, US$114.5-billion with Google and US$110-billion with Amazon. Meanwhile, a US$120.5-billion arrangement with Elon Musk’s SpaceXAI can be cancelled with 90 days notice beginning early next year. Anthropic also has compute and infrastructure deals that do not have fixed spending commitments, but the payments are nonetheless “expected to be significant,” the company said.

Anthropic is trying to reduce its dependency on major cloud providers, according to the filing, and it is developing its own data centre infrastructure by leasing computer chips housed in dedicated facilities that are operated on its behalf. In October, the company announced it had signed a lease agreement with EverWind Fuels, which is developing a $10-billion data centre project in Nova Scotia.

As a public benefit corporation, Anthropic said it has to balance its obligations to shareholders with its mission to use AI to benefit humanity. “There may be times when we choose to advance more slowly than we could in order to make our systems safer, even if that puts us at a competitive disadvantage,” the founders wrote in their letter.

The company noted that its capital and corporate governance structure “has no direct public company precedent.” Anthropic has established a Long-Term Benefit Trust, an independent body with five members, including former Federal Reserve chair Ben Bernanke, that will elect four of seven board members. A separate corporate entity controlled by Anthropic’s seven founders will elect the rest.

The founders will hold voting power amounting to 50.1 per cent. That gives them the ability to decide the outcome of any matter submitted to a shareholder vote. As a result, common shareholders will have little sway over such decisions, according to the filing, and the governance structure could make it more difficult to remove directors and management.