Meta and Microsoft curb internal use of Anthropic's Claude as AI costs bite

Meta and Microsoft are curbing their employees’ internal use of Anthropic’s Claude AI models and steering them toward in-house tools, The Information reported on Monday, citing unnamed sources (https://www.pymnts.com/news/artificial-intelligence/2026/microsoft-meta-steer-staff-from-anthropic-claude-in-house-ai/).

Microsoft was on track to spend $1 billion this year on internal use of Anthropic’s technology — covering Claude Code, Claude models inside Copilot and Claude Mythos — but has since lowered that figure by more than a third. A Microsoft spokesperson told The Information the company has been steering staff toward GitHub Copilot, though engineers can still choose other models.

At Meta, the number of employees using Anthropic’s Claude Code coding assistant has fallen from about 60,000 earlier this year to 30,000, largely by pushing staff toward Meta’s own Muse Code and its in-house-only tool MetaCode. Meta had been on track to spend billions of dollars this year on internal AI tools, with Claude Code the most popular, and did not comment on the report.

The pullback matters because both companies are among Anthropic’s largest enterprise customers, and it comes as the startup heads toward an anticipated initial public offering. A Wall Street Journal report the same day found just 11% of almost 400 surveyed businesses could accurately project their AI costs.