Anthropic's draft IPO filing shows an $80.7 billion half-year loss, but about $72 billion is an accounting charge, not cash spent running the business.
Anthropic, the AI company behind the Claude chatbot, reported a net loss of $80.7 billion for the first half of 2026. The number looks shocking. But about $72 billion of it is an accounting charge, not cash going out the door. Reuters reported the figures Tuesday, Oct. 20, after seeing a draft of the company's stock market filing.
The $72 billion charge is tied to a rise in the estimated value of bonds that could eventually turn into Anthropic shares. It is a bookkeeping entry, not a cost of running the company. The first-half loss compares with a $13 billion loss a year earlier. Reuters also reported last month that Anthropic's 2025 net loss was $42 billion, mostly from a similar charge.
The cash costs are still huge. Anthropic spent close to $9 billion on computing capacity and infrastructure in the first half of 2026, up from $2.3 billion a year earlier. It has also pledged to spend as much as $518 billion on cloud services, computing and infrastructure over the next decade.
Investors are watching closely. They want to know whether spending on data centers and chips will pay off.
Anthropic is expected to go public next month. An IPO is when a private company first sells shares to the public. At an investor meeting earlier this month, the company said third-quarter revenue rose 45% to $16.8 billion from the prior quarter, according to people familiar with the matter.
That eased some worries about costs. But some investors were disappointed that growth is slowing, since revenue had more than doubled between the first and second quarters. Some are now weighing a lower valuation of around $1.2 trillion.
If you use or pay for AI tools, the pressure on these companies to prove they can make money may shape what you see later. That could affect prices, plans and features. The material does not say that any prices will change. That is not known yet.
Anthropic's huge loss is mostly a paper charge, but its soaring computing costs and slowing growth are what investors will watch.
Visa tested 22 AI models on realistic shopping tasks. They passed just over half, and often bought the wrong thing anyway.
Read more: Visa tested 22 AI shoppers, and most still make costly mistakesDARPA wants AI to hunt for design flaws in stock, bond, energy and cloud markets, so enemies cannot use them to cause economic damage that reaches households.
Read more: Pentagon wants AI to find weak spots in stock and energy marketsTuring Award winner Richard Sutton says more computing power alone will not make AI truly smart. Big AI companies disagree, which matters for anyone betting on AI.
Read more: AI pioneer says piling on computing power won't make machines smart
