Anthropic, record losses ahead of stock market listing

Anthropic, the U.S. company that develops the Claude artificial intelligence model, recorded losses of $80.7 billion (approximately €72 billion) in the first half of 2026. The figure emerges from the S-1 prospectus, the document that U.S. companies file with the financial oversight authority ahead of a stock market listing, which El País reported along with elements drawn from the Wall Street Journal.

The figure is misleading if read as a sign of operational crisis. According to the cited data, about $72 billion of the $80.7 billion loss stems from non-cash accounting adjustments linked to convertible securities and other financing instruments: items that reflect how the company structured its capital-raising rounds, not actual cash outflows or a decline in business activity.

The rest of the picture tells a story of marked growth. First-half 2026 revenues are put at $16.3 billion, compared with $1.1 billion in the same period the previous year. At the same time, the company invested nearly $9 billion in computing capacity and infrastructure, a commitment consistent with the nature of the sector: training and running large language models requires computing power on an increasing scale, and Anthropic competes directly with OpenAI, Google and other labs making similar investments.

The prospectus also indicates cloud computing agreements signed for the next decade worth a combined total of approximately $450 billion, a figure that signals how much the company is betting on sustained demand for computing capacity over the long term, well beyond the horizon of the upcoming listing.

It remains to be seen how the market will read these figures at the time of the listing. Accounting losses from convertible instruments are a common phenomenon among technology companies in a phase of rapid expansion that raise capital through successive rounds, but the absolute size of the figure — over $80 billion in six months — has no direct precedent in the generative artificial intelligence sector, where listed companies so far have shorter and less transparent financial histories.

A limitation of the available information framework should be noted: the news currently comes from a single source, El País, which in turn cites the S-1 prospectus and draws on elements from the Wall Street Journal; no independent confirmation is currently available. As this concerns significant financial data — a declared loss in the range of tens of billions of dollars — caution is warranted until other outlets or independent analyses confirm the figures in the filed document.

Anthropic’s listing, if confirmed within the timeframe indicated in the prospectus, would represent one of the largest stock market debuts in the artificial intelligence sector, and the relationship between sharply growing revenues, substantial infrastructure investments and accounting losses tied to capital structure will be one of the key issues investors will need to untangle in valuing the company.