Microsoft Trains Sales Staff to Promote In-House AI Over Anthropic and OpenAI

Microsoft MSFT is coaching its sales staff to steer enterprise customers toward its own in-house AI models over rivals Anthropic and OpenAI, training reps to emphasize cost and security advantages,,. At an internal strategy session tied to Microsoft's fiscal 2027 sales playbook, an executive framed the pitch as selling a complete end-to-end system rather than piecemeal AI tools, while a separate side-by-side comparison positioned Microsoft's Copilot against Anthropic's Claude within the Office suite as faster, more accurate, and better integrated with enterprise security controls.

The push comes as Microsoft accelerates its shift toward proprietary AI. Starting July 7, the company began migrating select Microsoft 365 applications, including Excel and Outlook, away from OpenAI and Anthropic models and onto its own "MAI" model stack. That stack debuted at Microsoft's Build 2026 developer conference in early June, when the company unveiled seven new in-house models, headlined by MAI-Thinking-1, a 35-billion-parameter model with a 256K-token context window.

The strategy underscores growing tension in Microsoft's AI partnerships. Despite its multibillion-dollar investment in, and continued Azure hosting relationship with, OpenAI, Microsoft appears increasingly intent on reducing its dependence on both OpenAI and Anthropic for the AI features embedded in its own productivity software. For investors, the move raises questions about how aggressively Microsoft can diversify away from partner models without straining commercial ties that still generate significant Azure consumption revenue, and how quickly its in-house models can match the performance sales teams are now pitching against them.

Related Stocks

• MSFT