Microsoft and Meta are significantly restricting internal use of Claude Code. According to reporting from The Information and CyberSecurityNews, both companies are instructing their developers to rely less on Anthropic's AI coding tool – a rare signal of tensions in the AI ecosystem and potential cost discussions.
Key Facts
- Microsoft is reducing planned spending on Anthropic's AI from $1 billion by approximately one-third
- Meta is cutting the number of developers with Claude Code access from approximately 60,000 to around 30,000 employees
- Both companies are canceling or throttling Claude Code licenses internally
- Meta's decline is not primarily driven by headcount reductions, according to internal sources
Microsoft's Reversal Since May 2026
Microsoft hit the brakes as early as May 2026. At that time, the company informed its developers that numerous Claude Code licenses would be canceled over the coming months. The result: AI spending forecasts dropped sharply. Earlier in 2026, Microsoft had planned to invest $1 billion in Anthropic's AI – now it's significantly less.
This move looks like a strategic realignment. Microsoft's own developer tools and AI solutions (such as Copilot) likely play a role. Why invest heavily in a competing tool when you have your own alternatives?
Meta Halves Its Developer Base
Meta shows a similar pattern. Early in 2026, approximately 60,000 Meta employees were using Claude Code in their daily work. That number has now dropped to around 30,000 – a 50% reduction. While Meta did cut headcount throughout the year, internal sources indicate this is not the primary driver of the decline. This suggests the restriction is a deliberate strategic choice.
Here too, the suspicion arises: Meta is developing its own AI coding solutions and wants to reduce dependence on Anthropic.
What This Means for the AI Industry
The moves by Microsoft and Meta are remarkable because they show that even major tech companies are critically reviewing their AI spending and dependencies. Claude Code was long the reference tool for AI-assisted software development – the fact that two of the industry's largest employers are now actively cutting usage could signal a turning point.
It's likely not about Claude Code's quality, but rather about costs, control, and strategic independence. Companies with their own AI models and tools don't want to remain dependent on an external vendor long-term – especially when licensing costs can be substantial at scale.
Implications for German Companies
For German development teams and mid-market businesses, this is an important signal: major tech players are actively diversifying their AI tools. This means you shouldn't automatically bet on a single tool – no matter how dominant it currently appears. At the same time, it could become challenging for Anthropic if major customers defect. German companies should review their own Claude Code licenses and evaluate whether open-source alternatives or in-house solutions might be more cost-effective in the long run.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
