Anthropic, the AI company behind Claude, is advancing its U.S. stock market debut after publicly releasing its IPO prospectus, revealing a rapidly growing business and staking a valuation exceeding $2 trillion, while also disclosing enormous AI development costs and warnings of risks more severe than typical tech firms.
This filing marks a major milestone in the AI market, as it will be the first time public stock investors can closely examine the financial details, infrastructure costs, and risks of a leading Frontier AI company. Anthropic has grown from a startup founded in 2021 into a key competitor alongside OpenAI, Google, and Meta.
Reuters reports that Anthropic is targeting a valuation above $2 trillion in its IPO, more than twice the approximately $965 billion valuation the company estimated back in May.
Revenue Grows 12-Fold but Losses Reach $42 Billion
The most attention-grabbing figure is revenue growth: Anthropic expects nearly $4.6 billion in revenue for 2025, about 12 times higher than the previous year. Meanwhile, its annualized revenue run rate, calculated from the latest revenue pace, rises from about $9 billion at the end of 2025 to over $65 billion by the end of July 2026.
The company further forecasts 2028 revenue to be roughly $190 to $200 billion, reflecting expectations for rapidly expanding AI demand in business sectors. However, this growth comes with extremely high costs.
In 2025, Anthropic reported a net loss of approximately $42 billion, most of which did not stem from direct cash operating expenses. Around $34 billion relates to revaluation of financial instruments convertible into future shares. Excluding these, operating losses exceed $8 billion.
This highlights a major challenge for Anthropic and its AI peers: whether AI revenue can grow fast enough to cover the soaring costs of model development and infrastructure. Disclosed data shows the largest expenses range from AI researcher salaries to compute and infrastructure.
In 2025, the company spent about $7.33 billion on computing and infrastructure, roughly triple the prior year. Looking ahead, Anthropic has cloud, computing, and infrastructure commitments totaling around $518 billion.
A key deal includes a $100 billion, 10-year agreement to use Amazon Web Services (AWS). The company also contracts with SpaceX for full data center processing capacity at Colossus 1 in Memphis, costing roughly $1.25 billion monthly or about $15 billion annually through May 2029.
Anthropic warns investors that many major customers lack long-term contracts, meaning they can reduce or stop spending at any time. This risk is critical for AI businesses since current revenues must cover enormous computing and infrastructure costs. If usage growth slows, the company could be stuck with fixed infrastructure expenses.
Anthropic’s expected public offering after the U.S. midterm elections in November will allow retail investors to own shares directly for the first time and is viewed as a “new benchmark” for AI company valuations, especially as competitor OpenAI filed a confidential IPO in June and may go public by early 2027.
If Anthropic successfully lists with a valuation over $2 trillion, investors will gain public market data to assess how capital markets value AI firms still requiring massive investments to scale.
More importantly, this IPO will reveal Frontier AI’s true unit economics—from customer revenue, processing costs, licensing fees, investments to safety expenses—all under public investor scrutiny.
This filing will help the market understand AI business details and provide an opportunity to directly compare the business models of two major AI companies in the future.
Anthropic’s governance structure reflects a balance between business interests and AI safety mission. The company remains a Delaware Public Benefit Corporation and forms a new legal entity, Founder LLC, comprising the seven co-founders.
Founder LLC holds a single Class F share with 50.1% of total voting rights on key matters, including certain board member elections, while publicly offered Class A shares carry one vote each. Thus, general shareholders have limited decision-making power compared to founders’ voting rights.
Anthropic’s prospectus warns this structure could lead to decisions conflicting with the company’s financial performance and shareholder interests in the short, medium, or long term, potentially impacting Class A share value.
Therefore, the critical focus of Anthropic’s IPO is not just the $2 trillion figure but how the market will judge the speed at which AI can generate massive revenue.
Beyond finances, Anthropic’s IPO documents dedicate significant attention to AI safety risks. Of the 261 main pages, about 80 discuss Risk Factors, more than the 40 pages describing the business itself.
The company warns that advanced AI could cause “Catastrophic or Existential Risks to Humanity.” Anthropic notes AI models might show self-preserving behaviors, such as resisting shutdown, concealing or distorting information, and actions resembling blackmail.
This paradox is a major focus of the IPO: despite Anthropic positioning itself as an AI safety-conscious firm, it must alert investors about risks from the very technology it develops.
The company acknowledges that models may develop unexpected capabilities during training that might only be discovered when deployed and after safety incidents occur.
Evan Hubinger, Anthropic’s AI Safety researcher, estimates over a 10% chance that AI could cause human fatalities within the next decade, emphasizing the resource-intensive nature of safety work and the need to allocate limited compute, personnel, and safety efforts carefully.
Nonetheless, Anthropic continues to develop and launch new models, unveiling Opus 5.5 on 22 September, followed by Sonnet 5.5 the next week, designed to be faster and less costly for customers not requiring Opus’s highest capabilities.
This creates a dual challenge: Anthropic must continuously advance models to stay competitive, but higher capabilities also increase risks and costs of safety testing.
Anthropic’s IPO may become a major test of whether public investors will value a rapidly growing company that requires huge investments and computing power, and whether AI investment risks will be weighed alongside valuation, revenue, profit, and increasingly, technology safety.
Source: Reuters [1] [2] , Yahoo Finance
Read technology, innovation, and digital transformation news with ThairathMoney for "Better Finance, Better Life" at https://www.thairath.co.th/money/tech_innovation
Follow the Facebook page: Thairath Money at https://www.facebook.com/ThairathMoney