Meta and Microsoft Scale Back Reliance on Anthropic’s Claude

Meta Platforms and Microsoft are reportedly scaling back their reliance on Anthropic’s Claude for internal work, marking a notable shift in their AI tool strategies. At Meta, internal use of Claude Code dropped from roughly 60,000 users earlier in 2026 to about 30,000, while Microsoft cut its projected spending on Anthropic technology by more than a third from an initial $1 billion.

The reduction in Claude usage comes as both companies increasingly favor their own in-house tools. Meta has been promoting MetaCode, which already has over 30,000 internal users, and Muse Code, which began external testing in August 2026 and has over 6,000 internal users. Microsoft, meanwhile, is steering employees toward GitHub Copilot and OpenAI models, reflecting its strategic focus on tools it owns or has invested in.

The shift highlights a broader trend of hyperscalers preferring internal AI solutions to reduce dependence on external vendors. Cost efficiency and competitive instincts are key drivers, as companies like Meta and Microsoft aim to minimize long-term reliance on third-party tools for critical functions like software development.

Anthropic’s enterprise business faces scrutiny as it moves toward a public listing, with investors closely watching customer concentration and growth trajectory. Despite recent product advancements, such as Claude Opus 5.5’s 1-million-token context window, the pullback by major clients raises questions about the company’s future enterprise AI spending and market positioning.