Claude too expensive? Report says Microsoft is cancelling licenses for internal use after rising cost
Microsoft is reportedly winding down most Claude Code licenses amid rising AI coding costs, even as the company discusses supplying its Maia AI chips to Anthropic.
For months, tech companies have been telling the world that artificial intelligence will make software development faster, cheaper and more efficient. Developers are increasingly using AI tools to write code, fix bugs and automate repetitive work. But now, another reality of the AI boom is beginning to emerge — the technology may also be becoming extremely expensive to use.
Recently, Uber reportedly exhausted its entire 2026 artificial intelligence budget by April, just four months into the calendar year, after Anthropic’s Claude Code spread across roughly 5,000 engineers faster than the company’s finance models had anticipated. And now, Microsoft is reportedly planning to remove most of its Claude Code licenses as well.
Microsoft reportedly pulling back
According to The Verge, June 30 marks the final day of Microsoft’s current financial year, and removing Claude Code licenses is seen as an easy way to reduce operating expenses before the new financial year begins in July.
Microsoft first opened access to Claude Code in December, allowing thousands of its developers to use Anthropic’s AI coding tool daily. But the company is now reportedly pushing many developers toward its own GitHub Copilot CLI instead.
The report says Microsoft’s Experiences + Devices team — which includes engineers working on Windows, Microsoft 365, Outlook, Microsoft Teams and Surface — is winding down its usage of Claude Code by the end of June.
The hidden cost of AI coding
The situation is highlighting something the AI industry rarely talks about openly — token costs.
Every AI-generated response, line of code or chatbot interaction consumes computing resources. As companies scale AI usage across thousands of employees, those costs can rise rapidly.
That appears to be what happened at Uber, where widespread use of Claude Code reportedly consumed the company’s AI budget far faster than expected.
The irony is striking. AI coding tools are supposed to improve productivity and reduce costs, but for some companies, the tools themselves are now becoming a major operating expense.
Cutting Anthropic tools while talking business with Anthropic
What makes Microsoft’s move even more interesting is that while the company is reportedly reducing internal usage of Anthropic’s coding tools, it is also discussing a deeper hardware partnership with the company.
According to CNBC, Microsoft is currently in talks to supply its custom artificial intelligence chips to Anthropic.
Microsoft introduced its second-generation Maia AI chip in January. During Microsoft’s April earnings call, CEO Satya Nadella said the Maia 200 chip offers “over 30 per cent improved tokens per dollar” compared with the latest silicon currently used by the company.
AI’s strange new economics
The broader picture reveals how complicated the AI race is becoming for tech giants.
In November, Microsoft said it would invest $5 billion in Anthropic, while Anthropic committed to spending $30 billion on Azure cloud services. At the same time, Anthropic also relies on cloud infrastructure from Amazon and Google.
So even as companies compete aggressively in artificial intelligence, they also remain deeply dependent on one another.