Microsoft is systematically replacing third-party AI models like ChatGPT and Claude with its own internal AI systems across Microsoft 365 applications, according to recent reports. The software giant has begun routing at least some AI prompts from tools like Excel and Outlook to its proprietary MAI (Microsoft AI) models rather than paying external providers like OpenAI and Anthropic for each query processed.

Why Is Microsoft Making This Switch?

The primary driver behind this shift is cost control. Token costs, which measure how much data an AI model processes, have spiraled dramatically across the industry. Microsoft's CEO of AI, Mustafa Suleyman, made the company's intentions clear during the June Build conference, stating that the plan is to "reduce and ultimately eliminate" the amount of money it gives to Anthropic. This represents a significant departure from Microsoft's historical reliance on external AI providers.

The financial pressure is real and widespread. Earlier in 2026, Microsoft began reducing its workers' access to external tools like Anthropic's Claude Code after costs skyrocketed. The company later moved GitHub Copilot users to token-based billing and pulled its employees' access to Claude Fable over data-retention concerns. These moves reflect a broader industry trend where even major tech companies are struggling with AI spending. Uber reportedly spent its entire annual AI budget in just the first three months of 2026, and an undisclosed company spent half a billion dollars on AI tokens in a single month due to uncapped engineer limits.

What AI Models Is Microsoft Using Instead?

Microsoft has already developed and deployed several internal alternatives to compete with external offerings. The company's MAI models are available in Copilot for Business and Enterprise users. MAI-Code-1-Flash is Microsoft's coding assistant tool, designed to compete directly with Claude Code and OpenAI's Codex. Microsoft has also announced plans to release its own Teams-based transcription model, further reducing its dependence on external providers.

While the volume of prompts being rerouted to Microsoft's internal models remains small relative to the total number of queries across Microsoft 365, this represents the first concrete evidence of how the company intends to execute its cost-reduction strategy.

How to Understand the Impact on Enterprise Users

• Immediate Changes: Microsoft 365 users may notice that some AI-powered features now use Microsoft's internal models instead of ChatGPT or Claude, though the shift is gradual and affects only a portion of queries.
• Model Quality Differences: Microsoft's MAI models are designed to compete with external offerings, but users accustomed to specific ChatGPT or Claude capabilities may experience different response patterns or performance characteristics.
• Cost Implications: For enterprise customers, this shift could eventually lead to lower licensing costs as Microsoft reduces its token expenses, though pricing changes have not yet been announced.
• Data Handling: Microsoft's internal models may have different data retention and privacy policies compared to third-party providers, which could affect how enterprise data is processed and stored.

The shift reflects a broader industry pattern where companies are moving away from the "tokenmaxxing" approach that dominated early 2026. AI tokenmaxxing leaderboards, which tracked companies' spending on AI tokens, have quickly fallen out of fashion as organizations recognize the unsustainability of unlimited AI spending.

Microsoft's strategy of pushing users toward its own models makes costs easier to manage, even though the return on investment from these internal AI systems remains difficult to quantify. The company's approach suggests that future enterprise AI tools may increasingly rely on proprietary models rather than third-party integrations, fundamentally reshaping how organizations access and pay for AI capabilities in their daily workflows.