Meta and Microsoft slash usage of Anthropic's Claude

Meta and Microsoft are sharply reducing their internal reliance on Anthropic's Claude models as the startup increasingly positions itself as a direct competitor. At Microsoft, where annual spending on Claude was previously projected to exceed $1 billion, executives Scott Guthrie and Jay Parikh have instructed staff to transition to internal alternatives. Employees in the cloud division are being directed to use GitHub Copilot and OpenAI models instead. Consequently, Microsoft has slashed its Claude spending by more than a third, reducing the monthly per-employee budget in its cloud division from $100,000 down to approximately $10,000.

Meta has executed similar cutbacks, with its internal Claude Code user base dropping from roughly 60,000 to 30,000. This decline is attributed to company layoffs and a strategic push toward Meta's proprietary tools, such as Muse Code and MetaCode. The financial scale of Meta's previous usage was massive; the company spent more than $105 million on Claude Code during a single 28-day period before implementing AI cost-cutting measures.

These cutbacks reflect a broader strategic shift as Anthropic evolves from an ecosystem partner into a formidable rival. With products like Claude Cowork directly challenging established productivity suites like Microsoft Office, tech giants are reluctant to fund a competitor's growth or grant them access to valuable training data. For AI practitioners and enterprise developers, this transition highlights the growing importance of vendor lock-in and the financial unsustainability of relying on external frontier models for internal operations. Organizations are increasingly forced to choose between building proprietary code-generation tools or committing to a single ecosystem, making multi-model strategies more complex to justify financially.