Meta and Microsoft cut internal use of Anthropic's Claude

By Splitfeed EditorialLast updated Oct 5, 2026, 20:00 UTC

On October 5, 2026, reports emerged that both Meta and Microsoft had significantly pulled back their internal use of Anthropic's Claude. Microsoft cut the monthly per-employee Claude budget within its cloud division from $100,000 to $10,000, a 90 percent reduction. Meta separately halved the number of employees using Claude Code, dropping from roughly 60,000 to 30,000. In both cases, the companies are reported to be directing staff toward their own in-house AI tools rather than third-party products.

The practical significance of these moves extends beyond the raw numbers. Anthropic occupies an unusual position in the AI industry: it depends heavily on a relatively small number of large enterprise clients for a substantial share of its revenue. When two of those clients, each of considerable scale, reduce usage simultaneously, the concentration risk becomes concrete rather than theoretical. The broader context is also notable. Anthropic has been moving from a posture of partnership with major tech firms toward one of direct competition, offering its own coding and productivity tools that now rival what Microsoft and Meta are building internally. The pullback may partly reflect a deliberate decision by those companies to avoid funding a growing competitor.

None of the three camps, Pro-AI, Anti-AI, or Middle Ground, have published reactions to this story yet. Pro-AI voices would typically frame a move like this as healthy market competition, arguing that it incentivizes Anthropic to build better products and that enterprise clients naturally experiment with multiple tools over time. Anti-AI commentators would more likely read the retreat as evidence of instability in the commercial AI sector, pointing to concentration risk and the difficulty of sustaining large enterprise relationships as competitive dynamics shift. A Middle Ground perspective would probably focus on what the episode reveals about moat problems in AI, specifically that model quality alone may not be sufficient to retain clients who have the resources to build their own alternatives.

The clearest signal to watch for next is whether Anthropic reports any change in enterprise revenue or client composition in the coming quarters, and whether either Microsoft or Meta makes public statements about their internal AI tooling strategies. Any formal earnings disclosure or investor communication from Anthropic would help clarify how much financial weight these two relationships actually carried.