Anthropic's internal AI use is now valued higher than its employee pay

What did Anthropic reveal about its AI usage?

Anthropic's IPO filing reveals a striking milestone in the AI industry: the value of Claude AI used internally, calculated at the prices Anthropic charges its customers, exceeded the company's monthly employee compensation costs in April 2026, excluding stock awards, according to Business Insider.

By September, that internal AI usage value had surged more than 250-fold since January. The company is not spending more cash on AI than salaries — it is valuing its internal consumption at retail prices — but the numbers underscore how deeply AI has embedded into the company's own operations.

Cryptonomist, reporting on the same 335-page prospectus, noted that the personnel costs included salaries, bonuses, benefits and contributions but excluded equity awards. The valuation of internal Claude usage was calculated at list prices, the rates charged to external customers.

The 335-page prospectus filed with the SEC provides the most detailed look yet at how an AI company consumes its own product.

How much code does Claude write at Anthropic?

More than 90% of the code added to Anthropic's software in September was written by Claude, not by human engineers, according to the filing. The statistic highlights how AI companies themselves are the earliest and heaviest adopters of their own technology.

Business Insider reporter Alistair Barr, who broke the story, noted that tech companies are starting to treat AI tokens as a 'fourth component of employee compensation, alongside salary, bonuses, and stock awards.' Anthropic's numbers show why: revenue has grown vastly faster than headcount.

The KuCoin report on the filing added that more than 11,900 customers spent over $100,000 annually on Claude, with 250 of those spending more than $10 million per year, underscoring the rapid enterprise adoption of the platform.

The 90% figure means that for every 10 lines of new code added to Anthropic's software, fewer than one was written by a human engineer.

What does this mean for the broader industry?

The milestone challenges the long-held Silicon Valley assumption that talented humans are the most valuable resource. Anthropic's filing suggests that AI can increasingly substitute for human labor in knowledge work, especially in software engineering.

For decades, companies scaled by hiring more people. Anthropic's model implies a future where a small team supported by AI agents can produce output that would have required a much larger workforce. The company's own headcount has grown modestly compared to its revenue trajectory, which the filing shows reached $17 billion in Q3 2026.

Business Standard noted that the disclosure was part of Anthropic's amended IPO filing with the SEC, suggesting the company wants investors to understand how AI is transforming its own operations as it prepares for a valuation of up to $2 trillion.

What are the caveats?

The comparison uses list prices for Claude, not Anthropic's actual cost of running the models internally. The company almost certainly pays far less than retail to run its own models, meaning the cash expense of internal AI usage is much lower than the headline figure suggests.

Still, the symbolism is significant: the AI company that builds Claude relies on Claude for the vast majority of its own software development. As the IPO filing warns, this dependency also creates concentration risk — if Claude were unavailable, Anthropic's engineering productivity would drop by more than 90%.

For IPO investors, the data cuts both ways: Anthropic's ability to replace human labor with its own product demonstrates Claude's real-world value, but the deep dependency on a single model also represents a risk factor that few other companies face.

The IPO filing also revealed that more than 11,900 customers spent over $100,000 annually on Claude, with 250 spending more than $10 million.