Anthropic, the company behind the Claude chatbot, reported a net loss of $80.7 billion for the first half of 2026. The figure looks alarming, but $72 billion of it was an accounting charge, not money spent running the business day to day. The numbers come from a draft stock market listing document reviewed by Reuters.
Where the big number comes from
Anthropic raised money through bonds that can eventually turn into shares. As the estimated value of those bonds rose, the company had to record a charge in its accounts.
In plain terms, $80.7 billion did not leave the company's bank account in six months. Much of the loss reflects a change in the value of its financing deals. For comparison, the net loss in the same period a year earlier was $13 billion.
The real spending is still large
The actual bills are big on their own. Anthropic spent nearly $9 billion on computing power and infrastructure in the first half, up from $2.3 billion a year earlier. It has also committed to spend at least $518 billion over a decade on cloud services, computing and infrastructure.
That money pays for the chips, data centers and capacity that make a chatbot answer feel instant.
Revenue is growing, but more slowly
Anthropic told prospective investors at an Oct. 14 meeting that third-quarter revenue reached $16.8 billion, up 45% from the previous quarter, Reuters reported.
Some investors were still disappointed. Second-quarter revenue had more than doubled from the first quarter, to $11.6 billion. So growth slowed in the third quarter.
What happens next
The San Francisco company is preparing its first sale of shares to the public, known as an initial public offering, or IPO. It is expected in November. Some investors are considering a valuation of around $1.5 trillion, Reuters reported. That is the price put on the whole company, not necessarily the amount it would raise by selling shares.
Anthropic now has to convince investors that demand for Claude can justify both that valuation and the spending ahead.
What it means for your business
A small business weighing AI tools for paperwork or customer questions faces a smaller version of the same test. Does the service save enough time to justify its price? Investors are asking whether enough customers will pay enough to cover a very large bill.
The takeawayMost of the $80.7 billion loss is a paper charge, but the real computing costs are huge, and the same value-for-money test applies to your own AI spending.