In an amended registration statement filed with the US Securities and Exchange Commission on 2 November 2026, Anthropic described how it uses Claude across research, engineering, business operations and other teams.

Claude Code can write and run code and handle complex, multi-step engineering tasks, shifting engineers toward directing and reviewing AI-generated work; Claude also supports experiments, training-data generation and processing, evaluations, safety audits, and code review.

By September 2026, Claude had written more than 90% of the code added to Anthropic’s software, and the estimated value of the company’s internal Claude usage at customer list prices had risen more than 250-fold since January.

Anthropic says this use creates a feedback loop: Claude helps train and evaluate future models, and each generation contributes to developing the next.

Anthropic also warned that using AI to build more capable AI could create significant safety risks by accelerating technological progress, saying its Responsible Scaling Policy is designed to monitor and mitigate risks from recursive self-improvement.

Anthropic said employees use multiple AI agents across research, engineering, sales and operations to increase output without needing to expand teams proportionally.

That operating leverage is reflected in the company’s growth: from September 2025 to September 2026, trailing 12-month revenue rose about 1,200% while headcount grew around 170%, lifting revenue per employee from $1.4 million to more than $6.6 million.

Anthropic’s headcount grew more than fivefold from December 2024 to September 2026, while revenue in the July–September 2026 quarter was 86 times revenue in the October–December 2024 quarter.

The filing lists more than 11,900 customers spending over $100,000 a year, including about 1,900 spending more than $1 million and roughly 250 spending more than $10 million.

Anthropic was valued at around $380 billion after raising $30 billion in a February 2026 funding round.

The figures also reflect why some technology companies are considering AI tokens—the units used to measure AI consumption—as a fourth part of employee compensation alongside salary, bonuses and stock awards.

More broadly, the company’s model challenges Silicon Valley’s long-standing assumption that skilled human workers are its most valuable resource.