Microsoft Coaching Salespeople to Discuss Its AI Competitors Internally
Microsoft has been training its salespeople internally to downplay competing AI models like OpenAI, Google’s offerings, and Anthropic’s Claude, even though Microsoft itself relies on these models in its products.
An executive vice president (EVP) compared Copilot, a Microsoft model powered by Anthropic’s Claude, unfavorably to Claude directly during a strategy kickoff meeting. They claimed Copilot was slower, less accurate, and lacked security integrations. This pitch instructs salespeople to tell customers that one of the engines in their products is inferior while continuing to sell it.
This internal strategy shift reflects Microsoft’s wider move to integrate its own MAI (Major AI) models into more of its products, replacing reliance on external providers like OpenAI and Anthropic. They believe these in-house models are cheaper, more efficient, and come bundled with other Microsoft products.
What makes this notable is the target: Microsoft has built dependence on these same firms it’s now competing against. It recently amended its partnership with OpenAI, allowing them to sell to competitors, and quietly pulled back from using Claude Code internally.
The company aims to convince customers that the alternative models are not as good as their own, despite paying for access to them in the past. This is a strategic attempt to justify its substantial AI investments to skeptical investors.