Microsoft is coaching its sales team to compete aggressively against OpenAI and Anthropic by playing up the shortcomings of their AI products, Bloomberg reported, citing an internal strategy meeting for the fiscal year that began this month. The pitch centers on cost, security and breadth rather than a claim of a better underlying model.
Executive Vice President Jay Parikh reportedly told staff that rivals are “selling parts” while Microsoft is “selling the full end-to-end system,” a message he urged the sales force to broadcast throughout fiscal 2027. The framing casts Microsoft’s advantage as its integrated platform, spanning Azure, security, identity, Office and Copilot, rather than the raw capability of any single model.
The sharpest moment involved a direct comparison to Anthropic. Copilot Executive Vice President Jacob Andreou presented a side-by-side analysis of Copilot against Anthropic’s Claude inside Microsoft’s Office apps, characterizing Claude as slower, less accurate and lacking the security integrations needed for enterprise work.
It is worth stressing that this is Microsoft’s own internal characterization delivered in a sales context, not an independent benchmark, and Anthropic declined to comment. Microsoft’s cost argument leaned on a concrete example: CEO Satya Nadella cited a Unilever automated claims system that a Microsoft spokesperson said is projected to save about $300 million after swapping a top frontier model for a cheaper Microsoft alternative, a figure Unilever did not confirm.
What makes the pitch notable is who Microsoft is now targeting: the same companies whose models have powered its flagship products. The sales narrative reflects a real product shift, as Microsoft began this month migrating parts of apps like Excel and Outlook away from OpenAI and Anthropic models toward its own in-house MAI systems, unveiled at its Build conference in June.
The flagship, MAI-Thinking-1, is pitched as matching GPT and Claude on some benchmarks at a fraction of the cost, and Microsoft is now routing tens of thousands of prompts a week through its own models to cut inference bills.
The Model No Longer the Message
The strategy’s core insight is that Microsoft is not arguing its models are the best; it is arguing that model quality is becoming beside the point. As frontier models converge and grow more interchangeable, Microsoft is betting the decisive factors are price, security, compliance and integration into tools enterprises already run, exactly the “substrate” it controls.
This is a classic incumbent move against point-solution challengers: make the bundled suite cheaper and stickier than the sum of best-of-breed alternatives, so a customer’s loyalty attaches to the platform rather than any one model, including its partners’. The approach also lets Microsoft treat model providers as swappable commodities, decoupling from any single vendor as a feature rather than a risk.
A Partnership Turned Rivalry
The move formalizes a striking reversal. Microsoft and OpenAI were once so intertwined that Microsoft held exclusive access to OpenAI’s models and still owns a stake reported around $135 billion, while Anthropic’s Claude was embedded directly into Copilot. An April renegotiation that dropped OpenAI’s exclusivity, freeing it to sell to Microsoft’s rivals and freeing Microsoft to build competing models, opened the door to this competition.
Nadella has reportedly feared Microsoft becoming “the next IBM” by leaning too hard on one partner, and owning the model is how he intends to avoid that fate. The context is financial pressure: Microsoft’s stock has slid sharply over the past year on investor doubts about its heavy AI spending, making a cost-discipline story strategically urgent.
For OpenAI and Anthropic, the discomfort is real, since Microsoft has been one of their largest distribution channels, and it is now quietly teaching its salespeople to sell around them.
