Anthropic Lost $42 Billion in 2025. Why Is It Okay With Losing So Much Money?

Tech

By Jerome London

Updated 33 minutes ago, September 28, 2026

Losing billions starts to make sense if you believe you can replace almost every worker in the American economy, and eventually the whole Earth (and whatever is beyond it).

That is the broader economic logic behind the AI spending race. Work that businesses now pay people to do could become a service they buy from AI companies. The potential market reaches into every industry, with each new capability opening another category of paid work to automation.

Anthropic’s prospectus describes a transformation that could surpass industrialization, electricity and the internet. On that view, today’s spending is a race to build the technology businesses could eventually depend on to operate.

Its nearly $42 billion net loss in 2025 needs an important distinction. Roughly $34 billion came from an accounting charge tied to the rising estimated value of financing that could eventually convert into shares. That wasn’t cash spent running the company. But its operating loss still exceeded $8 billion, according to a prospectus reviewed by Reuters.

The company is spending heavily to build that capacity. Compute and infrastructure costs tripled to $7.33 billion in 2025, accounting for more than half its operating expenses. Revenue rose 12-fold to nearly $4.6 billion, and the prospectus lists $518 billion in cloud, computing and infrastructure obligations for the coming year.

A planned IPO could value Anthropic at more than $2 trillion. A valuation that large would depend on expectations that its current business is only the beginning—and that AI will take on an increasing share of the work that keeps the economy running.

That future remains a bet. Nearly a quarter of revenue came from just two customers, and many of Anthropic’s largest clients have no long-term commitment to keep spending.

The scale of the ambition helps explain the scale of the spending: Anthropic is betting that AI wipes out the whole labor force.

We’re once the product, now we’re the training data of our livelihood.

Like what you read? Follow Thought Catalog on Facebook, and check out more on our website.