A loss that doubles and unmatched cash
Anthropic has published its prospectus ahead of a listing on the U.S. stock market, a 335-page document that, according to the BBC, contains corporate information never previously disclosed. The main figures concern financial performance: in the first half of the year, the company recorded losses of $80.7 billion, twice what it lost in all of 2025. Revenue in the same period was $16.2 billion, growing rapidly but still far from break-even.
Against these losses, the company has $131.6 billion in liquidity, almost entirely classified as working capital and therefore usable to finance day-to-day operations. According to the material, this is more than Meta and Google combined. Analyst Troy Hooper, co-head of equity capital markets at Mergermarket, cited in the source, calls the revenue growth exceptional but notes that investors today are less tolerant of technology companies promising high future returns than they were a few months ago.
Hooper further notes that the debt picture remains unclear because of large accounting charges linked to certain forms of investment: investors, he says, still do not have a clear view of underlying profitability or of the ability to generate enough cash to sustain the business as it grows. Anthropic has not yet indicated a target price for the shares, while a market valuation of $2 trillion or more is expected, which would make it the largest listing in history.
Concentrated clients and rising costs
The business structure has an element of fragility: a significant portion of revenue comes from 250 large enterprise customers who, according to the company, paid at least $10 million in the past year. Just five enterprise customers generate a quarter of current business. Michael Monaghan, portfolio manager at Founder ETF, stresses that many of Anthropic's main customers do not have long-term contracts and that this concentration makes growth dependent on a small share of buyers, with no guarantees of continuity.
The prospectus also indicates enterprise relationships with Google and Amazon, which are at the same time investors and competitors, a configuration that according to Monaghan could create future conflicts of interest. On the cost side, data center leasing alone has cost $321.4 billion so far this year; Hooper observes that expenses of this magnitude are bound to grow and that, even with such high liquidity, they will not be enough to sustain the development of the business. Whoever invests in Anthropic, he concludes, will be paying for future dominance and not for the current economics.
The overall enterprise customer base has nonetheless grown rapidly, from about 1,500 last year to nearly 12,000 this year. Part of the strategy involves offering Claude tools free and then converting users into paying customers over time; the company maintains that this approach has led to the fastest enterprise adoption ever recorded for a technology.
Stated risks and promised benefits
In the risk factors chapter, Anthropic states that sufficiently advanced artificial intelligence systems, if not developed, distributed and managed with adequate safeguards, oversight and security measures, could pose catastrophic or existential risks to humanity. The document recalls that less than a year ago a group of hackers believed to be sponsored by China allegedly manipulated Claude to successfully gain entry into several institutions, including major technology companies, financial institutions, chemical firms and government agencies. The company also specifies that its success depends on its ability to develop and distribute increasingly advanced models, and that a significant risk to that ability is access to the necessary infrastructure, including chips and data centers.
Alongside the risks, the prospectus reiterates the thesis that AI will produce extraordinary benefits: the stated mission is the responsible development and maintenance of advanced AI for the long-term benefit of humanity, and the company says it does not pursue projects solely for commercial opportunity or research novelty. Among the hypothesized benefits are personalized treatments and better diagnoses in healthcare, tailored education and legal and financial advice, more efficient power grids and batteries, and an acceleration of technological development in developing countries. These are, by the document's own wording, hypothetical scenarios. Among other elements, the listing price and actual valuation remain to be verified, as does a full reading of the debt position and underlying profitability.
