On May 14, Microsoft sent thousands of its own engineers a message canceling most internal Claude Code licenses across its Experiences and Devices division — the team responsible for Windows, Microsoft 365, Outlook, Teams, and Surface — with a deadline of June 30, 2026. The story was first reported by The Verge's Tom Warren.
The instruction came from Rajesh Jha, Microsoft's Executive Vice President for the division. Microsoft had opened Claude Code access to Experiences + Devices in December 2025, inviting thousands of developers, programme managers, and designers to use it in real engineering workflows. According to The Verge, the tool became popular quickly — popular enough that it started displacing GitHub Copilot CLI in daily use.
That created a problem that was both financial and strategic: Microsoft sells Copilot to the rest of the world and cannot credibly do that while its own engineers migrate away from it at scale.
The bill that arrived before the year was half over
Microsoft wasn't alone in discovering this problem. Uber burned through its entire 2026 AI budget in just four months after rolling out Claude Code to roughly 5,000 engineers. Adoption climbed from 32% of engineers in February to 84% by March. By spring, 95% of Uber engineers were using AI tools monthly, and roughly 70% of committed code originated from those tools.
Monthly spend averaged $150 to $250 per engineer, with power users running between $500 and $2,000 per month. Uber's CTO Praveen Neppalli Naga reported spending $1,200 in a single two-hour session, according to Forbes. Uber reportedly exhausted its entire $3.4 billion AI budget for 2026 within four months.
The mechanics are straightforward. Traditional enterprise software generally uses fixed pricing regardless of how heavily employees use the product. AI systems work differently. The more useful the tool becomes, the more employees use it, and the more expensive it gets.
Claude Code's enterprise pricing charges a base seat fee of $20 per seat per month plus actual API token usage, meaning costs scale directly with how much engineers use the tool. GitHub Copilot Enterprise, by contrast, charges a flat $39 per seat per month with no usage surcharge. The flat rate is more expensive on paper. In practice, it's cheaper if your engineers are productive.
What Copilot CLI is inheriting
Rajesh Jha told staff: "Claude Code was an important part of that learning. At the same time, Copilot CLI has given us something especially important: a product we can help shape directly with GitHub for Microsoft's repos, workflows, security expectations, and engineering needs." The official framing is toolchain unification.
GitHub Copilot, after Microsoft folded it into the broader GitHub AI brand with a usage-based credit shift on June 1, 2026, still owns the inline autocomplete loop most engineers use 200-plus times a day and ships natively into VS Code, JetBrains, Visual Studio, Neovim, Xcode, and Eclipse. The distribution advantage is real. The gap in agentic capability is also real, and the engineers being redirected already know it.
Anthropic launched Claude Opus 4.8 with availability via the Claude API, Amazon Bedrock, Google Cloud Vertex AI, and GitHub Copilot — meaning engineers in Experiences + Devices who are being moved from Claude Code to GitHub Copilot CLI now have access to Opus 4.8 through the tool they are being directed toward. The model stays. The interface changes.
The practical lesson for any engineering team deploying AI coding tools is the same one Microsoft and Uber learned at scale. The issue is not that enterprise AI coding tools are failing. It is that organisations incentivised adoption without building the financial and governance infrastructure to manage what successful adoption looks like at scale. For a solo founder or a small shop running a five-person engineering team, that lesson is cheaper to absorb early: a five-engineer startup on Copilot Business pays $95 per month for five seats; the same team on Cursor Business pays $200 per month. Before expanding access to Claude Code's token-based tiers, model the ceiling, not just the floor.