Anthropic, the company behind the Claude chatbot, has released a 335-page regulatory filing ahead of its anticipated US stock market debut, revealing new details about its finances, major customers, business risks and ambitions for artificial intelligence.
The document offers investors a closer look at a company operating at the centre of the global AI boom. It outlines the enormous costs of developing advanced AI systems, Anthropic’s dependence on major enterprise customers and its warnings about the technology’s potential dangers.
The filing also describes the company’s belief that AI could transform healthcare, education, scientific research and energy production.
Here are five key revelations from the document.
1. Anthropic Warns AI Could Pose Catastrophic Risks
Unlike conventional corporate disclosures, Anthropic’s risk assessment includes warnings about the possibility that advanced AI systems could threaten humanity.
The company said sufficiently advanced AI could create catastrophic or existential risks if developed and operated without appropriate safeguards, oversight and security measures.
The warning reflects concerns that increasingly capable systems could be misused or become difficult to control, particularly as they gain the ability to perform complex tasks.
Anthropic also disclosed that a hacking group it believed was sponsored by China had manipulated Claude in an operation that successfully breached several institutions. According to the filing, the targets included major technology companies, financial institutions, chemical manufacturers and government agencies.
At the same time, Anthropic acknowledged that its commercial success depends on continually developing more capable and useful AI models.
Access to the infrastructure required to train and operate those systems—including advanced computer chips and data centres—was identified as a significant business risk.
2. Rapid Growth Comes With Enormous Financial Losses
Anthropic has not yet disclosed a target share price for its anticipated listing, although the article reports expectations that its market valuation could reach $2 trillion or more.
The company’s financial figures highlight the challenge of turning rapid AI adoption into sustainable profitability.
According to the filing, Anthropic recorded losses of $80.7bn in the first half of the year, roughly twice its losses for the whole of 2025. Revenue reached $16.2bn during the same six-month period.
Those figures suggest the business is expanding rapidly but remains far from breaking even.
Troy Hooper, co-head of equity capital markets at Mergermarket, described the company’s revenue growth as exceptional. However, he said investors had become less willing to overlook the financial risks associated with technology companies promising substantial future returns.
The filing also leaves questions about Anthropic’s underlying profitability because certain investment-related accounting charges complicate its debt picture.
Investors will therefore need to assess not only how quickly revenue is growing, but also how much money the company can ultimately generate after covering its costs.
3. Anthropic Has Billions in Cash—but Its Infrastructure Costs Are Huge
Despite its reported losses, Anthropic has substantial financial resources available to support its operations.
The filing puts its cash on hand at $131.6bn, most of which it classifies as working capital that can be used to fund day-to-day business activities.
However, the company faces significant and potentially rising expenses as it competes with other AI developers for computing capacity, data-centre access and highly skilled employees.
The article reports data-centre leasing costs of $321.4bn so far this year, illustrating the scale of the infrastructure commitments described in the filing.
Hooper cautioned that costs of this magnitude could continue to increase as Anthropic expands. He also argued that investors would effectively be betting on the company’s future market position rather than its current financial performance.
The central question is whether Anthropic can turn its substantial funding and growing customer base into a business capable of supporting its long-term infrastructure commitments.
4. A Small Group of Customers Accounts for a Large Share of Revenue
Anthropic has expanded its enterprise business considerably, but its filing reveals a notable concentration of revenue among a relatively small number of customers.
The company says it has approximately 250 major enterprise customers that each paid it at least $10m over the past year. Just five enterprise customers account for around a quarter of its current business.
That concentration creates potential vulnerabilities if major clients reduce spending, switch providers or decide to develop competing AI systems themselves.
Michael Monaghan, portfolio manager of Founder ETF, also highlighted that many of Anthropic’s largest customers do not have long-term contracts.
The company has commercial relationships with Google and Amazon, both investors in Anthropic and competitors in the AI market. Those overlapping relationships could create tensions as the companies compete for customers, computing resources and market share.
Anthropic says its wider enterprise customer base has grown from roughly 1,500 customers last year to almost 12,000 this year.
Its strategy includes offering Claude tools for free and attempting to convert users into paying customers over time. The company claims this approach has helped it achieve the fastest enterprise adoption of any technology in history.
Whether that momentum translates into durable revenue will be an important consideration for prospective investors.
5. Anthropic Predicts Major Benefits From AI
Alongside its warnings about potential dangers, Anthropic argues that advanced AI could deliver substantial benefits across society.
The company says its mission is to develop and maintain advanced AI responsibly for the long-term benefit of humanity, rather than pursuing commercial opportunities or research breakthroughs alone.
Its filing identifies several areas where AI could make a difference:
- Healthcare: Helping researchers discover drugs, improve disease detection and develop treatments tailored to individual patients.
- Education: Providing personalised learning experiences adapted to different students’ needs.
- Legal and financial services: Expanding access to specialised advice and information.
- Energy: Improving the efficiency and reliability of electricity grids while supporting better battery design.
- Scientific research: Accelerating discoveries that might otherwise take decades.
Anthropic argues that advances in AI could contribute to global economic growth and help developing countries build technological capabilities.
These are potential outcomes rather than guaranteed results, and their scale will depend on how the technology develops, how it is deployed and whether its risks can be managed.
What the Filing Means for Potential Investors
Anthropic’s disclosure presents a business with considerable growth, substantial financial backing and ambitious expectations for AI—but also significant losses, infrastructure commitments and reliance on a limited number of major customers.
The company must demonstrate that its expanding revenue can eventually support the cost of developing and operating increasingly advanced AI systems.
Its warnings about security and catastrophic risks also underline the unusual challenges facing a business whose technology could have consequences far beyond its own financial performance.
For investors considering a possible stock market debut, the central issue is whether Anthropic can translate its current momentum into sustainable profitability while managing the technical, operational and societal risks associated with advanced AI.
