Meta and Microsoft Slash Employee Claude AI Spending

Meta and Microsoft have begun sharply reducing employee use of Anthropic’s Claude AI in favor of their own internal coding tools, The Information reported on October 5, citing people familiar with the matter and internal data. Neither company has publicly confirmed the figures below.

Microsoft cuts internal AI spending limits. The company had projected annual internal spending on Anthropic technology above $1 billion. That figure has since fallen by more than a third after managers directed staff to limit Claude use and favor in-house tools such as GitHub Copilot and OpenAI’s models, according to the report.

In the cloud and AI division, monthly spending ceilings dropped from $100,000 per employee to roughly $10,000 in most cases. These caps represent maximum allowances rather than actual outlays.

Meta halves Claude Code users. Internal Claude Code users fell from around 60,000 earlier this year to about 30,000. Layoffs contributed, yet the main driver was Meta’s shift to its own tools: MetaCode, now used by more than 30,000 employees, and Muse Code, used by over 6,000.

Meta began external testing of Muse Code with clients in August. Despite the drop in users, the company spent more than $105 million on Claude Code over a recent 28-day period, according to the report.

Customer access remains separate. Neither company is ending customer access to Claude through Microsoft platforms. Enterprise spending by clients on Anthropic models via Microsoft continues to grow even as internal employee budgets tighten.

Security considerations persist. The move to new coding assistants does not remove risks tied to automated access to sensitive files and credentials. Prior reports noted vulnerabilities in Claude Code that Anthropic later fixed, and similar issues have appeared in Microsoft tools.