Anthropic’s figures for the third quarter of 2026 show that the company is spending more on internal AI use than on its workforce, according to its own calculation. The comparison excludes stock-based compensation and values AI use at list prices, rather than the potentially lower cost of running the models on Anthropic’s own infrastructure. The company has not separately disclosed the total cost or the size of that difference.
Anthropic says internal AI use has increased 250-fold since January 2026. In September, AI produced 90 percent of contributions to the company’s codebase, it reported, while code output per developer was 45 times higher than in 2024. The workforce also grew substantially, reaching more than five times its size at the end of 2024.
The figures come as Anthropic prepares for a planned major public offering. The company’s revenue growth has slowed, while it continues to report substantial losses. The article cites a total loss of $104 billion in the first nine months of the year and says Anthropic expects its data-center capacity expansion to cost $895 billion over the coming years, with much of that commitment due within five years.
Tech journalist Alistair Barr described AI spending exceeding employee costs as a potential turning point, challenging the idea that skilled people are Silicon Valley’s most valuable resource. For Anthropic, however, the economics also depend on customers adopting AI at a similar scale and seeing enough value to justify the expense. Apollo Global Management chief economist Torsten Slok said the benefits of the AI boom had so far accrued to the technology sector, with no measurable financial gains yet for the wider economy.
Sources: t3n.