THREAT ASSESSMENT: The vendor is now the rival.
Microsoft is coaching its sales force to talk down AI products from OpenAI, Google, and Anthropic in favor of its own, according to a Bloomberg report detailed by TechCrunch AI. The instructions came at an internal strategy meeting on Tuesday, pitched as the kickoff for Microsoft’s new fiscal year. This is the same Microsoft that has spent years building its flagship AI features on top of those very models.
The Field Orders
- Sell the system, not the parts. “Everyone else is selling parts, we’re selling the full end-to-end system. That’s the story that we all need to get out there and tell in FY27,” Executive Vice President Jay Parikh reportedly told the room.
- Attack on cost and efficiency. The session leaned heavily on positioning Microsoft’s in-house models as cheaper and more efficient than rivals’.
- Name the target. Copilot EVP Jacob Andreou reportedly ran a presentation comparing Copilot directly against Anthropic’s Claude.
- Use the home-turf advantage. Inside Microsoft’s own office apps, Andreou said Anthropic’s model was “slower and less accurate, and lacked the proper security integrations.”
TechCrunch reached out to Microsoft and Anthropic for comment and hasn’t reported a response.
Why This Isn’t Just Sales Talk
Companies train reps to knock competitors. That’s ordinary. What stands out here is the target list. Microsoft is telling sellers to attack the suppliers it built its AI stack on.
And it lines up with what Microsoft is actually doing behind the scenes. An earlier report found the company has been quietly swapping OpenAI and Anthropic models out of Word and Excel in favor of its own, framed as cost-cutting. The sales pitch and the engineering roadmap are pointing the same direction.
How We Got Here
The old arrangement was unusually tight. Microsoft supplied capital and compute to OpenAI. In exchange it got exclusive access to OpenAI’s API and models. Neither side had much reason to compete with the other.
That ended in April. The two amended the deal and dropped the exclusivity clause, which freed OpenAI to sell to Microsoft’s competitors. Once your partner can arm your rivals, the partnership stops being a moat. The new sales script reads like the natural consequence.
There’s a financial layer too. Microsoft’s stock outlook has been soft for the past year as investors question the scale of its AI spending. Loudly claiming your in-house models beat the ones you used to license is one way to argue that the buildout was worth it.
What Practitioners Should Watch
- Expect competitive decks in your next Microsoft meeting. If you’re evaluating Copilot against Claude or ChatGPT, benchmarks from a Microsoft rep now come with a stated agenda. Run your own evals on your own workloads.
- Model swaps may hit you silently. If Microsoft continues replacing third-party models inside Word, Excel, and Copilot, the thing answering your prompt next quarter may not be the thing answering it today. Output quality can shift without a version bump.
- The integration argument is the real fight. Andreou’s strongest claim wasn’t raw intelligence. It was security integrations and speed inside Microsoft’s apps. That’s a distribution advantage, not a model advantage, and it’s the one Microsoft can defend.
- Watch for reciprocity. OpenAI already sells to Microsoft’s competitors. Anthropic sits inside AWS and Google Cloud. If Microsoft goes on the attack, those partnerships get sharper elbows.
The Read
The frontier lab and the platform vendor used to be different businesses. That’s over. Microsoft wants to own the models, the apps, and the cloud underneath, and it’s now willing to say so to customers with a slide comparing itself to a company it still does business with.
For buyers, the useful takeaway is simple. The vendor claiming neutrality about which model powers your tools no longer has any. Test accordingly.
More detail on the internal meeting is available at the original source.