Microsoft is coaching its sales team to badmouth OpenAI and Anthropic
The company that built its AI business on partners' models is now telling salespeople those same models are slower, less accurate, and too expensive.
Key points
- Microsoft held an internal sales strategy meeting for fiscal year 2027 planning where executives told staff to pitch the company's own AI models as superior to those from OpenAI, Anthropic, and Google.
- Executive Vice President Jacob Andreou reportedly told salespeople that Anthropic's Claude is "slower and less accurate" than Microsoft's Copilot inside Microsoft's office apps, and lacks proper security integrations.
- Microsoft amended its partnership with OpenAI in April 2025, dropping the clause that had given Microsoft exclusive access to OpenAI's technology.
- Microsoft has already been swapping OpenAI and Anthropic models out of flagship apps like Word and Excel in favour of cheaper in-house alternatives.
Microsoft and OpenAI were, for a long time, practically the same company in the public eye. Microsoft poured billions into OpenAI and received exclusive access to its models. That arrangement shaped every AI feature Microsoft shipped, from the Copilot assistant in Word and Excel to Bing's search tools.
Now, according to reporting by Bloomberg, Microsoft is teaching its sales force to say the opposite: that rival models aren't good enough.
At an internal meeting on Tuesday, billed as a strategy session for fiscal year 2027, executives set out the new pitch. Executive Vice President Jay Parikh told the room: "Everyone else is selling parts. We're selling the full end-to-end system." Executive Vice President Jacob Andreou went further, presenting a direct comparison between Microsoft's Copilot and Anthropic's Claude, a chatbot made by the San Francisco company Anthropic. Andreou reportedly told staff that Claude was "slower and less accurate" inside Microsoft's office apps and lacked proper security integrations.
Neither Microsoft nor Anthropic had responded to requests for comment at publication time.
The targets are notable. OpenAI and Anthropic aren't random competitors: they're the companies Microsoft depended on almost entirely for the AI models powering its own products. The April 2025 amendment to the OpenAI partnership removed Microsoft's exclusivity and freed OpenAI to sell directly to Microsoft's rivals, and that shift appears to have accelerated the split.
Cost is almost certainly a factor too. Licensing powerful external models is expensive, and Microsoft has faced investor pressure over whether its enormous AI infrastructure spending will pay off. Switching to in-house models cuts that bill. As we reported on 13 July 2026, Satya Nadella has been warning enterprises about the hidden price of depending on outside AI vendors: that framing now looks like groundwork for exactly this pivot.
What does this mean for people who use Microsoft products?
If you use Word or Excel with Copilot active, the AI answering you may already be a Microsoft-built model rather than one from OpenAI or Anthropic. The swap has been happening quietly.
Whether Microsoft's own models are genuinely as capable as the ones they replace is a separate question from what the sales team will claim. Independent benchmarks, the standardised tests researchers use to compare AI systems, are the real check on that. Company sales decks are not peer-reviewed results, and the two are worth keeping clearly apart.
