Microsoft has begun scaling back employee access to Anthropic's Claude Code and is redirecting its engineering workforce toward GitHub Copilot CLI, as surging generative AI usage drives up enterprise computing costs, according to multiple reports. The move marks a notable reversal for the US-based technology giant, which had previously extended Claude Code access to thousands of employees across developer, project management, and design roles.

From Broad Rollout to Pullback

Microsoft had earlier expanded Claude Code access widely to encourage experimentation with AI-assisted coding tools. However, the rapid and widespread adoption reportedly drove a significant increase in operational expenses, prompting the company to rein in usage. Engineers are now being steered toward GitHub Copilot CLI — Microsoft's own AI coding assistant — as a cost-effective alternative.

Notably, the rollback does not signal a breakdown in Microsoft's broader relationship with Anthropic. The company continues to maintain its multibillion-dollar Foundry agreement, which covers cloud infrastructure support and access to Claude models through Azure services.

An Industry-Wide Cost Reckoning

Microsoft's decision reflects a broader challenge confronting technology companies that have aggressively deployed generative AI at scale. The core issue is structural: large language models operate on a token-based pricing model, where enterprises are billed based on the volume of text processed and generated. Heavy employee usage can therefore translate into costs that escalate far faster than anticipated.

The problem is not unique to Microsoft. Uber Chief Technology Officer Praveen Neppalli Naga reportedly said the company exhausted its entire 2026 budget for AI coding tools within the first four months of the year, driven by heavy internal usage. This pattern suggests that enterprise AI budgeting frameworks have not yet caught up with actual consumption rates.

What Analysts Are Projecting

Goldman Sachs has estimated that agentic AI systems could drive a massive increase in token usage by 2030, as businesses deploy AI agents at scale across operations. Research firm Gartner projected that while the per-unit cost of running advanced AI models may decline over time, total enterprise AI spending could still rise — because newer agentic systems require substantially higher token consumption per task than their predecessors.

These projections indicate that cost pressure on enterprise AI deployments is likely to intensify before it eases, even as model efficiency improves.

Broader Tech Restructuring Context

This development comes as several major technology firms signal a strategic pivot toward artificial intelligence, often at the expense of existing headcount. Earlier this month, companies including Meta Platforms, Oracle, and Cloudflare announced job cuts and restructuring measures as part of their AI acceleration drives. Microsoft's Claude Code pullback fits within this wider pattern of enterprises recalibrating AI investment against real-world cost outcomes.

As AI coding tools become standard across the industry, companies will increasingly face the challenge of balancing innovation velocity with fiscal discipline — a tension that Microsoft's move has brought sharply into focus.