Microsoft Reassesses AI Costs Amid Token Bill Surge
Key Summary
Microsoft has reduced its spending on Anthropic's Claude models by more than a third, with token costs exceeding budgeted expectations. The deployment began in December 2025 and saw thousands of engineers adopt the AI coding assistant, with token consumption doubling in the first few months. Per-engineer token costs ranged between $500 and $2,000 each month. By May 2026, most internal licenses were canceled, with employees redirected to GitHub Copilot CLI. Microsoft is part of a wider effort to trim AI costs, with the company routing more prompts to its proprietary MAI models. Anthropic still has a $31.4 billion infrastructure commitment with Microsoft, which may impact investors.
Microsoft Cuts Claude Spending by More Than a Third Amid Token Bills Pile Up
Microsoft's shift to in-house AI tools highlights a growing trend of tech firms reassessing AI costs, impacting future enterprise AI strategies.
Industry Impact and Implications
This move by Microsoft underscores the need for companies to closely monitor their AI-related expenses, as token bills continue to rise.
Future of Enterprise AI
The trend towards in-house AI tools and reduced reliance on external token bills may lead to more cost-effective AI solutions for enterprises, but also raises questions about the long-term viability of these approaches.