Microsoft CEO Satya Nadella revealed earlier this year that the company now writes up to 30% of its code using generative AI. The company has invested $13 billion in OpenAI. It built GitHub Copilot. It has made AI central to every product in its portfolio.
On May 14, Microsoft (MSFT) sent thousands of its own engineers a message that cuts against that narrative in a very specific way.
What Microsoft told its Experiences and Devices engineers
Microsoft is cancelling 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, according to TechRadar. Engineers are being directed to switch to GitHub Copilot CLI, Microsoft’s own command-line AI coding tool.
The story was first reported by The Verge’s Tom Warren on May 14 via his Notepad newsletter. Warren noted that Claude Code had become “perhaps a little too popular” inside Microsoft, with engineers choosing Anthropic’s tool over Microsoft’s own product. That popularity, it turns out, was part of the problem.
More AI:
- Micron sits at the center of a red-hot chip rally
- IBM CEO sends blunt message on AI and quantum computing
- Anthropic CEO makes shocking admission about AI
The internal memo came from Rajesh Jha, Microsoft’s Executive Vice President. “Claude Code was an important part of that learning,” Jha wrote, according to Memeburn. “At the same time, Copilot CLI has given us something especially important: a product we can help shape directly with GitHub.”
The official reason given is “toolchain unification.” But the June 30 deadline is also the last day of Microsoft’s fiscal year, and the timing suggests cost reduction played a role alongside the desire to consolidate on internally controlled software, TechRadar confirmed.
Why Claude Code became too expensive at enterprise scale
Claude Code launched inside Microsoft’s Experiences and Devices group in December 2025. Six months later, the experiment is ending. The tool was not cancelled because engineers disliked it. It was cancelled because they used it too much.
Token-based pricing is the standard model for frontier AI APIs. Every prompt, code generation request, review, and debugging session consumes tokens, and the cost accumulates with each interaction. At the individual developer level, the numbers are manageable. At the scale of thousands of engineers using the tool daily, they compound into a line item that can surprise finance teams, according to AI Magazine.
Uber’s experience illustrates exactly how quickly that can happen. The company deployed Claude Code to 5,000 engineers and watched monthly usage rates climb to 84-95% by April 2026. Per-engineer API costs reached between $500 and $2,000 per month. The result: Uber burned through its entire $3.4 billion 2026 AI budget in four months. Its CTO said the annual budget had already been exhausted before the year was half over, AI Magazine confirmed.
That is not a modest overrun. It is a sign that the economics of enterprise AI tool deployment are not yet fully understood by the companies buying them.
Taggart/Getty Images
The broader pricing reckoning now facing enterprise AI
Microsoft and Uber are not isolated cases. AI software prices across the US have climbed 20-37%, according to AI Magazine. GitHub is responding to the pressure by shifting all Copilot plans to usage-based billing through GitHub AI Credits starting June 1, 2026. The move signals that even the largest AI platforms are redesigning their commercial models around the reality that heavy usage at enterprise scale creates unpredictable cost exposure for buyers.
The structural issue is that AI tools are not like traditional software licenses. A seat license for a conventional productivity tool costs the same whether the employee uses it for one hour or eight. A token-based AI tool costs more the more useful it becomes. That creates a paradox for enterprise buyers: the better the tool works, the more engineers use it, and the higher the bill grows.
That dynamic is beginning to change how companies think about AI deployment. The first wave of enterprise AI adoption was driven by enthusiasm about what the tools could do. The second wave, now underway, is being shaped by what they cost, according to TechRadar.
Key figures on Microsoft’s Claude Code cancellation and enterprise AI costs:
- Microsoft timeline: Claude Code launched internally December 2025 in Experiences and Devices; cancellation deadline June 30, 2026; engineers redirected to GitHub Copilot CLI, according to TechRadar
- Affected team: Experiences and Devices division covering Windows, Microsoft 365, Outlook, Teams, and Surface engineers, TechRadar confirmed
- Uber AI budget: $3.4 billion 2026 AI budget exhausted in four months; 5,000 engineers deployed; monthly usage rates 84-95% by April 2026; per-engineer costs $500 to $2,000 per month, according to Storyboard18
- GitHub pricing shift: all Copilot plans moving to usage-based billing via GitHub AI Credits starting June 1, 2026, Windows Central confirmed
- AI software inflation: US AI software prices have climbed 20-37%, Windows Central noted
- Microsoft AI context: $13 billion invested in OpenAI; writes up to 30% of its code using generative AI per Satya Nadella; retains Claude model access via Microsoft Foundry and Microsoft 365 Copilot, according to Memeburn
What this means for investors watching enterprise AI spending
The Microsoft decision carries a specific implication for investors tracking enterprise AI adoption. The assumption built into many AI growth narratives is that once companies deploy these tools, adoption accelerates and spending follows in a predictable upward line.
Microsoft’s experience suggests the opposite can also happen: deployment accelerates, spending surpasses what was budgeted, and companies pull back or redirect to cheaper alternatives.
That dynamic does not undermine the long-term AI investment thesis.
Microsoft is not abandoning AI. It is consolidating onto its own product, which it controls and can optimize for cost. But it does complicate the assumption that token-based API revenue from enterprise customers will grow smoothly as adoption expands. Some of that revenue will convert to churn when companies hit budget ceilings and shift to alternatives.
For Anthropic, the cancellation is a visible counternarrative at a sensitive moment. The company is in the process of raising capital at a $900 billion valuation, according to AI Weekly, with enterprise adoption as a core part of the pitch. Its most prominent internal enterprise user finding the cost model unsustainable is not a fatal development, but it is a data point that enterprise buyers and investors will factor into their assessment of where AI spending goes from here.
Related: AI is getting worse as Google and Anthropic nerf AI models and limit usage
#Microsoft #CEO #sends #shocking #message #employees