Meta and Microsoft Reduce Use of Anthropic’s Claude AI Models Amid In-House Development

Meta and Microsoft, two of Anthropic’s largest enterprise customers, have significantly reduced their usage of Claude AI models, particularly Claude Code, as they pivot towards developing their own in-house AI solutions. This strategic shift is part of a broader trend among major tech firms aiming to strengthen their internal capabilities amid intensifying competition in the AI sector.

According to a report by The Information, Meta's internal users of Claude Code have dropped from approximately 60,000 at the beginning of the year to around 30,000, with many employees transitioning to Meta's proprietary tools like MetaCode and Muse Code. In a recent 28-day period, Meta reportedly spent over $105 million on Claude Code, highlighting the extent of its previous reliance on Anthropic's technology.

Similarly, Microsoft has cut its projected annual spending on Anthropic from over $1 billion to less than $700 million, reducing monthly budgets from $100,000 to about $10,000. The company is now focusing on alternatives such as GitHub Copilot CLI and OpenAI models.

This reduction in reliance on Claude comes as both companies implement cost-management strategies and as Anthropic's offerings increasingly overlap with their own products. While internal usage has declined, demand for Claude through Microsoft’s Azure platform remains strong, indicating that customer-facing applications of Claude are still thriving.

As these tech giants continue to develop their own AI capabilities, the competitive landscape for AI models may shift, putting pressure on Anthropic’s market position. Observers will be watching for any announcements from Meta or Microsoft regarding new AI tools, as well as updates on Anthropic’s future developments and partnerships.