The more capable AI is, the bigger the riskier the bill? Anthropic's $2 trillion valuation faces 'artificial intelligence out of control' storm

Zhitong Finance App learned that at a time when Anthropic, the global AI model/ AI application leader, hit the world's largest IPO, the AI valuation dispute is adding a more difficult variable to quantify: the ability of intelligent entities to generate revenue is getting stronger, and whether their potential liability will also exceed the scope of enterprises and insurance systems.

According to the latest media reports, Anthropic is seeking a valuation of about 2 trillion US dollars and raising up to 100 billion US dollars to challenge SpaceX's IPO record of 86.2 billion US dollars set in June this year; the promotion arrangements recently disclosed by Bloomberg and other media point to a listing as early as mid-November. Anthropic's strongest competitor, OpenAI, is currently negotiating new financing of at least US$30 billion and is seeking a pre-investment valuation of approximately US$1.4 trillion as a financing arrangement in addition to the recent IPO; this is still the target of negotiations, and the deal has not yet been completed.

The capital appeal of the two leading AI model companies is closely related to commercial growth: Anthropic's annualized revenue operating rate exceeded $65 billion at the end of July, and the September report further indicated that it is expected to exceed $100 billion during the year; OpenAI's annualized revenue operating rate is approaching $70 billion, up more than 70% from the beginning of the third quarter, and corporate sales have more than doubled since July.

However, revenue growth can enter the valuation model, and compensation, business restrictions, and security investments caused by AI loss of control still lack a stable pricing basis. This is a new test for Anthropic's super IPO — how much premium investors are willing to pay for future smart service revenue, and how much risk they are willing to take on undefined AI regulatory risks and unclear responsibilities.

At the same time that Anthropic reached the top of the IPO, the market began to question “who will pay the risk bill”

Investors considering participating in Anthropic PBC's huge IPO are facing an unusual problem: how should a company that some people fear may contribute to human extinction be valued?

Researchers posted on the X platform that the probability of AI causing human extinction is 10% or more, which sparked a discussion. Anthropic co-founder and CEO Dario Amodei also participated. In an article, he explained his concerns: if top researchers don't slow down the pace of development, humans may lose control of AI systems, and these tools may be misused for cyber attacks and bioterrorism, and cause serious economic turmoil.

Although some observers have scoffed at the idea that a company could be held responsible for causing human extinction — and think it's impossible to even quantify this risk in an initial public offering disclosure — the risks faced by Anthropic and its competitor OpenAI are similar to those faced by companies dealing in hazardous substances or military equipment.

Sam Lessing, partner at Slow Ventures, believes that countermeasures after the 2008 financial crisis can provide lessons for dealing with AI risks. The venture capitalist suggested that the US government should force AI laboratories to deposit $1 trillion in cash into escrow accounts to cover possible future aftercare expenses, similar to how regulators require banks to hold more capital to cover losses after a financial crisis. He said that even so, this money might not be enough.

In an email interview with Bloomberg News, Lessing said, “Even without considering what they call mass extinction, these potential liabilities may mean that society faces a risk of $2 trillion or $4 trillion. Combined, the value of these companies may be less than zero.”

However, Lessing said that some investors may still be willing to buy Anthropic shares and ignore this unknown, but probably extremely huge cost, because they believe someone else will eventually pay the bill.

An Anthropic spokesperson declined to comment.

Arguments for doomsday risk? Although discussions surrounding AI security have risen to the level of risk of the end of human society, Anthropic plans to raise funds through an IPO that is comparable to SpaceX's record listing, and has a very large or even larger IPO market value and financing scale, which has raised questions and ridicule. However, the issue did not seem to hurt SpaceX's stock price performance.

The trillion-dollar rocket and satellite company, owned by Elon Musk, has pinned a large number of future development visions on AI, and its GROK model competes with Anthropic and OpenAI's respective products. SpaceX's IPO file did not mention the threat that AI poses to human survival, but rather suggests risks such as misinformation and “AI deception.” The company's June IPO raised US$86.2 billion, and its stock price is now more than 20% higher than the issue price.

Luke Rango, chief technology analyst at InvestorPlace, said that President Trump recently reached a “morally binding” AI security agreement with tech giant leaders, including Amodei, easing some of investors' anxiety about this broader issue.

Kevin Moss, managing director and portfolio manager of Private Shares Fund, which invests in late-stage high-growth private companies, said: “The current administration has publicly expressed concerns, but it is also a government that emphasizes deregulation while competing with China.”

Moss said AI clearly needs a regulatory framework, but discussions surrounding security are unlikely to disrupt Anthropic's IPO plans.

So far, the pressure to reduce security risks has mainly been borne by AI companies themselves, and investors may even welcome companies to invest some money in this regard.

Harrison Rolffs, a senior financial market analyst responsible for late-stage enterprise research at PitchBook, said the new security mechanisms will increase Anthropic's costs, but they may also strengthen customer trust in its products; if Chinese competitors or open source competitors fail to meet the same standards, these mechanisms can also help create competitive barriers.

Although the company is expected to buy insurance, historical disasters have shown that disputes over coverage limits and claims often leave shareholders with huge bills. In 2010, BP's “Deepwater Horizon” drilling platform exploded and sank, killing 11 people and triggering the largest offshore oil spill in US history. At the time, the company's so-called exclusive self-insurance arrangement covered only 700 million US dollars, and the final cost climbed to tens of billions of dollars.

Joe Tsiolkovsky, the founder and CEO of Relm Insurance, an insurance company focused on emerging industries such as AI and space, said in an interview with the media that the insurance industry has been providing protection for AI companies through traditional insurance contracts and developing new products for risk protection gaps associated with existing AI models and AI agents. He said that due to the lack of historical data and the fact that it is still unclear what form AI-related liability will take, it is still quite challenging for AI companies to obtain insurance coverage.

Tsiolkovsky said that as AI evolves from simply providing information to making decisions and interacting deeply with other agents, when losses inevitably occur, who actually bears legal responsibility is becoming an increasingly urgent issue.

However, there is almost no direct precedent for judicial decisions against AI agents getting out of control. As a result, investors considering investing in Anthropic at a valuation of up to $2 trillion are actually faced with a choice of two: accept the risk that the company may take full responsibility for some of the harshest AI scenarios, or abandon the investment.

Tsiolkovsky said, “When looking at publicly disclosed documents, you can get a rough idea of how much money these companies have invested in compliance, insurance, or self-insurance, but investors are unlikely to give valuation discounts on these IPOs because of unquantifiable risk exposure.”

“While some of these risks have yet to be objectively quantified, now is a good time to raise capital.”

The hidden bill of super IPOs: Anthropic IPO investors struggle to price the risk of AI getting out of control

The overall revenue of Anthropic PBC, a global AI model/ AI application leader, increased wildly to about 12 times that of the previous year in 2025, or close to 4.6 billion US dollars. At the same time, the AI application leader that developed Claude's current public offering may make the company's overall valuation exceed $2 trillion, surpassing the overall valuation of SpaceX, which was about $1.77 trillion when it was founded and headed by Musk, and Anthropic's plan to raise close to 100 billion US dollars will also surpass the total amount of about 85.7 billion US dollars raised after SpaceX's overallocation. These all mean that Anthropic is expected to surpass SpaceX and become the largest initial stock offering in history.

As cutting-edge AI agents represented by Muse and Astra quickly became popular around the world, AI is accelerating from dialogue tools to automated programming, research, and execution of complex enterprise workflows, and is simultaneously expanding application revenue and AI computing power resource investment. Anthropic's IPO is preparing to bring this expansion into the global open capital market. People familiar with the matter said that in addition to showing a 12-fold revenue surge, Anthropic will also assume a total spending obligation of up to 518 billion US dollars for cloud services, AI computing capacity, and extensive AI infrastructure over the next few years. The potential listing valuation of more than $2 trillion means that investors will directly price the commercialization capabilities, computing power procurement commitments, and long-term profit prospects of cutting-edge models, and the impact will also extend to OpenAI and related AI computing power infrastructure providers.

Recent advances in cutting-edge AI agent/AI model technology represented by Muse, Astra, and Anthropic Claude can be described as providing an important technical foundation for large-scale commercial expansion of AI applications into various industries and the continued surge in AI computing power requirements — in particular, the expansion of the scope of intelligent applications is expected to simultaneously increase AI superaccelerators such as AI GPUs/TPUs and AI core infrastructure requirements such as high-performance CPUs, high-performance HBM/DRAM/NAND memory chips in data centers, and high-speed optical interconnect devices. Anthropic is seeking to raise up to 100 billion US dollars at a valuation of about 2 trillion US dollars, but the final plan may still be adjusted together with Wall Street banks. Whether it can set a new fund-raising record depends on the scale of the final issuance.

Judging from the commercial mechanism, the basis for the increase in revenue and valuation of Anthropic and OpenAI companies is that AI is undertaking more jobs with actual payment value.

Programming, research, and enterprise process execution can bring about continuous subscription, API calls, and inference requirements; improved model capabilities expand the range of tasks that can be completed, and lower unit task costs improve the economy of customer adoption. Anthropic has distributed Claude to AWS, Google Cloud, and Azure, and expanded delivery capabilities through new computing power protocols; OpenAI is also launching continuously operating agents and higher-capacity models at lower prices. As a result of these changes, capital is simultaneously pricing current revenue growth, long-term position in the enterprise's workflow, and future ability to continue charging fees.

The same growth mechanism also expands liability exposure: agents move from generating suggestions to calling tools, executing operations, and collaborating with other agents, and errors can directly translate into customer losses. If compensation, mandatory funding, or business restrictions are ultimately borne by developers, growing revenue will need to cover more costs, and cash returns available to shareholders will also decline. This poses a potential “liability slash line” — business revenue is still expanding, but liability costs and capital use may reduce its valuation and cash flow expansion and shareholder return space.

The latest $1 trillion escrow fund proposed by Slow Ventures partner Sam Lessing is a policy proposal and is not a regulatory requirement that has already been implemented; the core issues relating to the $1 trillion escrow fund are how much of the social risk created by AI will return to the corporate balance sheet, that is, how much of the social losses caused by AI should be borne by AI companies, and how much capital should be prepared in advance for compensation. At the same time, verifiable security capabilities may also enhance customer trust, increase companies' willingness to purchase, and form a competitive advantage. For investors focusing on the AI super-bull market, the key is whether a core company can turn smart demand into a continuous cash return after deducting computing power, security, and liability costs.