Anthropic's growth and AI impact

According to a report published on October 16 by the Anthropic Institute, the AI laboratory that develops, among other things, the chatbot Claude, the company could serve as "an advanced indicator of how artificial intelligence will change the organization of work, hiring practices, and labor demand across the entire economy." The transformations taking place within Anthropic could thus foreshadow the changes to come in other companies as AI spreads throughout them.

Rapid expansion and hiring trends

The size of the AI laboratory would have tripled in just one year, according to figures from the Anthropic Institute. The company would currently have more than 3,500 employees, according to a spokesperson. The share of junior profiles among technical team hires would have been reduced by three times between spring 2025 and the beginning of summer 2026, going from 27% to 9%.

AI integration and productivity gains

Due to the large number of responses generated by the Claude agents used by employees, which would occupy more than half of their working time, or even more than a full week for engineers, employees at Anthropic have implemented AI agents to sort, reread, and supervise the work done by other agents. The productivity increase observed after the adoption of coding agents by engineers and also by non-technical employees is accompanied by a revenue per employee 2 to 4 times higher than that of Google or Meta at the same stage, after five years of existence.

Automation and role evolution

Thanks to automation, lawyers handle more contracts, salespeople close more sales, and accounting services manage more supplier invoices. The number of coding agent sessions for non-technical employees increased from 2 per month in August 2025 to 90 in May 2026, allowing them to develop their own software. Alongside this increasing automation, human coordination has increased at Anthropic: the median employee spends 20% more time in meetings than a year ago.

AI's potential and economic impact

For individual technical contributors, meeting time has increased by 16%. In August, Anthropic employees estimated that AI could perform 75% of their work by 2030, compared to 28% for other Americans. If Anthropic is observing this within the company, the productivity gains promised by AI are not yet perceptible at the scale of the American economy. In fact, the total factor productivity (TFP), as calculated by the San Francisco Fed, shows no signs of acceleration in the United States since the beginning of the AI investment cycle.