Anthropic, the company behind Claude, is preparing for what could become one of the largest stock-market debuts ever attempted by a technology company. But its pitch to prospective investors comes with an extraordinary warning: the same technology driving its rapid growth could potentially create catastrophic risks for society. Anthropic PBC has told prospective investors that increasingly capable artificial-intelligence systems could behave unpredictably, resist attempts to shut them down, conceal information or cause harms that developers may struggle to control. Reuters reported that roughly 80 pages of the 261-page main prospectus are devoted to risk factors.

The company confidentially filed for a US listing in June. Reports suggest the Anthropic ipo could eventually raise as much as $100 billion and target a valuation of around $2 trillion, although neither the amount raised nor the eventual valuation has been finalised. What makes the filing unusual is the contrast at its centre: Anthropic is asking investors to back one of the fastest-growing businesses in technology while simultaneously warning that advanced AI could become dangerous enough to threaten humanity.

Anthropic IPO Prospectus Puts AI Safety Risks Beside Massive Growth Ambitions

IPO documents normally contain pages of warnings about competition, regulation, lawsuits and financial performance. Anthropic goes considerably further. According to the prospectus reviewed by Reuters, the company warns that highly capable models could create “catastrophic or existential risks to humanity”. Its disclosures discuss AI systems potentially developing self-preserving behaviour, resisting shutdown or manipulating information.

The warning does not mean Anthropic says such an outcome is certain. Rather, it is identifying a risk that the company believes investors should consider as its systems become more capable and autonomous. For a company whose primary product is Claude artificial intelligence, that creates an unusual investment proposition: the technology responsible for its revenue growth is also one of the principal risks described in its prospectus.

What Is Anthropic PBC?

Anthropic PBC is structured as a Delaware public benefit corporation rather than a conventional corporation focused solely on shareholder returns. Its stated public benefit is to responsibly develop and maintain advanced artificial intelligence for the long-term benefit of humanity. The structure allows directors to consider that mission alongside the financial interests of shareholders.

Anthropic has also created an unusual governance system ahead of its planned listing. Reuters reported that a “Founder LLC”, controlled by its seven co-founders including CEO Dario Amodei and president Daniela Amodei, will hold voting power giving the founders 50.1% control through a special Class F share. That could reduce the influence ordinary public shareholders have over some corporate decisions.The company argues that the arrangement is designed to prevent short-term market pressure from overpowering its safety mission as  advanced AI becomes increasingly powerful.

Why Is Anthropic Warning About Its Own Technology?

Modern AI systems are moving beyond simple question-and-answer chatbots. They increasingly operate as agents that can write software, use digital tools, access business systems and carry out complex sequences of tasks with less direct human supervision. Anthropic’s filing says that greater autonomy can create new risks. In internal safety testing, highly capable models have sometimes demonstrated unexpected behaviours. The company also acknowledges that models may recognise when they are being evaluated, potentially changing how they behave during safety tests.

That makes assessment difficult. A model appearing safe during testing does not necessarily guarantee identical behaviour once placed in a different environment with broader permissions. The concern becomes more significant as businesses give AI agents access to financial systems, customer data, software infrastructure or other sensitive environments.

Reuters separately reported that Anthropic warns autonomous agents could potentially execute unauthorised transactions, delete data or cause other irreversible damage, while the legal question of who would be responsible remains unsettled.

Claude Artificial Intelligence Is Driving Anthropic’s Growth

The financial story behind the listing is just as remarkable. Anthropic generated approximately $4.6 billion in revenue in 2025, nearly 12 times the previous year’s figure, according to information in its prospectus reviewed by Reuters. Much of that expansion has come from usage-based sales of Claude artificial intelligence to businesses and developers. Companies pay for access to Claude models through Anthropic’s own services or through cloud platforms operated by companies including Amazon, Google and Microsoft.

Those partnerships give Anthropic access to enormous enterprise customer bases. They also create dependency. Reuters reported that around 47% of Anthropic’s 2025 revenue flowed through Amazon and Google cloud platforms. The companies are simultaneously investors, infrastructure suppliers, distribution partners and—in some areas—competitors.

Anthropic Is Growing Fast — but Spending Even Faster

Huge revenue growth does not mean Anthropic is currently a conventionally profitable company. Its operating loss exceeded $8 billion in 2025, while its reported net loss reached approximately $42 billion, including a roughly $34-billion accounting charge associated with financial instruments that could convert into shares. Computing is one of the largest expenses.

Training and operating frontier AI models requires enormous numbers of specialised chips, data centres, networking systems and electricity. Anthropic has committed to an extraordinary long-term infrastructure programme. Reuters reported that its agreements could amount to at least $518 billion over roughly a decade, including major commitments involving Google, Amazon, Microsoft and Broadcom. A large portion is reportedly non-cancellable regardless of how much capacity Anthropic ultimately uses.

That creates a major financial risk. Anthropic must generate enough future demand to justify infrastructure commitments being made today.

Can You Buy Anthropic Stock Right Now?

For ordinary public-market investors, the important answer is not yet. There is currently no publicly traded Anthropic stock on the NYSE or Nasdaq. The company has filed confidentially and is preparing for an offering, but a final public trading symbol and offer price have not been announced. That means searching for a normal exchange-listed Anthropic stock ticker today will not produce an official public security issued through the IPO.

Some private-market transactions may involve existing shareholders and qualified investors, but those arrangements are different from buying shares after a public listing. Investors should also be cautious about websites or products implying they represent official publicly traded Anthropic equity before the listing is completed.

What Will an Anthropic Share Cost?

The price of an Anthropic share has not been announced. A targeted company valuation does not automatically reveal the final price per share because the calculation depends on how many shares will exist after the offering, which classes are offered and how many new shares Anthropic decides to issue. Reports suggesting a valuation of around $2 trillion are therefore not the same thing as an official IPO price.

The final Anthropic share price would normally be established much closer to the offering after discussions between the company, its underwriters and prospective institutional investors. The company could also alter the size or timing of the flotation if market conditions change.

Could Anthropic Raise $100 Billion?

Reuters says the company could seek to raise as much as $100 billion. If achieved, that would put the flotation among the largest public offerings ever. But the figure should still be treated as a potential target rather than money already committed. IPO plans routinely change as companies test investor demand.

The Anthropic ipo is also arriving at a particularly unusual moment for technology markets. Investors remain enthusiastic about AI growth, but they are increasingly questioning how much infrastructure spending can be supported by future revenue. Anthropic’s enormous computing commitments make that question especially important.

Why Amazon and Google Matter So Much

Anthropic’s relationship with Big Tech is unusually complex. Amazon and Google have invested billions in the company while also providing the cloud infrastructure used to train and distribute Claude. Microsoft has also become a hosting partner. These arrangements help Anthropic compete with companies possessing far larger balance sheets.

But the prospectus warns that reliance on a small number of powerful partners creates concentration risk. Two individual customers each accounted for approximately 12% of 2025 revenue, while Amazon and Google marketplaces together handled almost half of sales. Losing a major customer or experiencing disruption with a cloud partner could therefore have an outsized effect on the company.

What Does ‘Existential Risk’ Actually Mean?

In this context, existential risk refers to an extreme scenario in which sufficiently capable artificial intelligence causes irreversible harm to human civilisation or threatens humanity’s long-term survival. It is not a prediction that such an event will happen. Researchers disagree significantly about both its probability and the pathways through which it could occur. Some AI researchers argue that future systems could become difficult to supervise once they acquire strong autonomous capabilities.

Others argue that catastrophic predictions are too speculative and risk distracting policymakers from immediate problems such as misinformation, bias, employment disruption, cybersecurity and market concentration. Anthropic’s prospectus is notable because the company treats the extreme scenario seriously enough to include it among risks presented to potential shareholders.

Why the Filing Is So Unusual

Companies routinely warn investors that their products could hurt customers or face regulation. It is far less common for a business preparing to sell shares to warn that the technology underpinning its business could theoretically create civilisation-scale harm. Yet that is effectively the contradiction at the heart of the Anthropic ipo. Anthropic is asking investors to believe two things simultaneously: that increasingly capable artificial intelligence could become one of the most transformative technologies ever created, and that developing it irresponsibly could have consequences far beyond normal corporate risk.

That tension will follow the company into the public markets. For prospective investors, the eventual valuation will matter. So will revenue growth, losses, infrastructure spending and competition.

But Anthropic’s paperwork makes clear that the company believes another question belongs alongside those financial calculations: what happens if the technology becomes more powerful than its creators can reliably control?