Microsoft has started scaling back employee access to Anthropic’s Claude Code and is redirecting engineers toward GitHub Copilot CLI as AI usage continues to push up enterprise computing costs. The company had earlier expanded Claude Code access to thousands of employees across functions including developers, project managers and designers to encourage experimentation with AI-assisted coding tools.
The decision comes as the rapid adoption of the tool is said to have significantly increased operating expenses. Even with the rollback in Claude Code access, Microsoft’s broader partnership with Anthropic remains intact, including its multibillion dollar Foundry agreement and access to Claude models through Azure services.
The report said the move reflects a wider cost challenge for companies deploying generative AI at scale. AI model pricing is typically based on tokens, meaning businesses are charged according to the amount of text processed and generated, which can quickly raise costs when internal adoption grows across large teams.
The report also noted that other technology companies are seeing similar pressure on their AI budgets. It cited Uber’s chief technology officer Praveen Neppalli Naga as saying the company used up its entire 2026 budget for AI coding tools within the first four months of the year because of heavy employee usage.
At the same time, the report said analysts expect token consumption to keep rising as agentic AI-systems spread across business workflows. Goldman Sachs has estimated that these systems could drive a sharp increase in token usage by 2030, while Gartner expects enterprise AI spending to keep rising even if model running costs decline over time, because more advanced systems consume far more tokens per task.
