Anthropic’s internal use of Claude topped staff costs at list prices by April
Report on Anthropic's IPO prospectus, filed 26 October 2026. It says the value of internal Claude use at customer list prices exceeded monthly personnel costs (excluding equity awards) by April 2026. Internal usage value rose more than 250-fold from January to September 2026, and Claude wrote over 90% of new code. The filing gives no compute cost or dollar figure.
What it claims
Anthropic's IPO prospectus, filed 26 October 2026, says internal Claude use valued at customer list prices exceeded monthly personnel costs by April 2026. Personnel costs here cover salaries, bonuses, benefits and contributions, but exclude equity awards. The filing also says Claude wrote more than 90% of new code added to Anthropic's software by September 2026. It describes Claude's role in developing later versions of Claude. Anthropic says internal use makes its teams more productive and helps improve subsequent models.
Evidence
The prospectus is 335 pages long. The article reports these figures from it:
- Internal usage value rose more than 250-fold from January to September 2026, measured at external list prices.
- Anthropic has about 5,300 employees. The prospectus says revenue per employee is growing faster than headcount, but the article gives no revenue figure.
- More than 11,900 customers spend over $100,000 a year. About 1,900 spend over $1 million and about 250 spend over $10 million.
- Anthropic was valued at about $380 billion after a $30 billion funding round in February 2026.
With about 5,300 staff, the April crossover means list-price usage per employee exceeded each employee's average monthly cash compensation. That is arithmetic on the reported figures, not a number the filing states.
Limits
The comparison sets a list-price valuation against cash compensation. It is not a measure of what the usage cost Anthropic. The prospectus does not disclose the compute cost of internal usage, and it gives no dollar amount for the April 2026 usage value. Excluding equity awards lowers the personnel side of the comparison, because equity is a large part of pay at AI labs. The "90% of new code" figure has no stated method: the article does not say whether it counts lines, commits or accepted suggestions, or how review time is handled. The article is secondary reporting, produced with AI assistance and reviewed by Cryptonomist editors. It quotes no passage from the prospectus directly.
Organisations named here
- Anthropic
Technologies named here
- Claude
Concepts named here
- Internal usage value at list prices
Related in ai-policy


Future of Life Institute AI Safety Index — Summer 2026

Lessons from the hacks
