The company is directing salespeople to position its proprietary MAI models as superior to competitors, marking a dramatic shift away from its OpenAI partnership
Microsoft is doing something that would have seemed unthinkable two years ago: training its sales force to trash-talk OpenAI.
During an internal planning session held around July 14-15, the company laid out a strategy for fiscal year 2027 that puts its own proprietary MAI models at the center of its enterprise AI pitch. The message to sales teams was unambiguous. Stop selling components, start selling the whole system, and don’t be shy about telling customers the competition can’t keep up.
The breakup gets louder #
Executive Vice President Jay Parikh, one of the architects behind this push, framed the strategy around a deceptively simple idea: Microsoft sells the full end-to-end system, not just individual parts. It’s a direct shot at companies like Anthropic, OpenAI, and Google, which Microsoft now explicitly frames as competitors rather than partners or peers.
Copilot EVP Jacob Andreou went even further during the session. He reportedly compared Anthropic’s Claude model unfavorably, characterizing it as “slower and less accurate” within Microsoft Office applications.
The competitive framing extends to all three major rivals. Sales teams are being armed with talking points that position OpenAI, Anthropic, and Google as offering fragmented solutions that can’t match the integration depth of Microsoft’s own stack.
This isn’t just sales theater. Microsoft has already begun routing Copilot workloads in Excel and Outlook to its own MAI models, a shift reported in early July. The company is literally replacing the third-party AI engines under the hood of its most widely used productivity apps with homegrown alternatives.
Why this matters now #
The timing here is not accidental. Back in April, Microsoft amended its partnership agreement with OpenAI, lifting the exclusive access arrangement that had previously given Microsoft preferential rights to OpenAI’s models.
Building your own models and deploying them across your own applications isn’t just a competitive play. It’s a cost management strategy dressed up in strategic language.
What this means for investors #
For anyone watching the AI investment landscape, this represents a potential inflection point. The cozy narrative of Microsoft and OpenAI as inseparable partners has been deteriorating for months, and this sales initiative makes the divergence explicit. Microsoft is no longer content to be the distribution layer for someone else’s intelligence. It wants to own the full stack. OpenAI, which has relied heavily on its Microsoft relationship for both capital and distribution, now faces the reality that its biggest partner is actively training salespeople to position alternatives. Anthropic, despite its growing enterprise traction, is being painted as a point solution that can’t match an integrated platform.
There’s a risk dimension worth considering too. Microsoft is essentially betting that its MAI models can match or exceed the quality of frontier models from OpenAI, Anthropic, and Google. The claim that Claude is “slower and less accurate” inside Office apps may be true in Microsoft’s testing environments, but independent validation of those claims will matter enormously.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our