Microsoft held an internal sales meeting on July 14, 2026, where executives told the company's salespeople to make AI products from OpenAI, Google, and Anthropic look bad compared to Microsoft's own AI tools, according to reporting from Bloomberg and TechCrunch.

The meeting was billed as a strategy session for Microsoft's new fiscal year. Executive Jay Parikh told attendees: "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 Jacob Andreou presented a specific comparison of Microsoft's AI assistant, called Copilot, to Anthropic's AI assistant, called Claude. He claimed Claude was "slower and less accurate, and lacked the proper security integrations" within Microsoft's office applications. The argument is that Microsoft offers a complete package — its AI models, its cloud computers, and its familiar apps like Word and Excel all from one company — while competitors only sell the AI part.

The sales directive follows a major change in Microsoft's relationship with OpenAI. In April 2026, the two companies changed their partnership agreement, dropping a clause that had given Microsoft exclusive rights to OpenAI's AI models (OpenAI). That change allowed OpenAI to sell its models to Microsoft's competitors, altering what had been one of the most important partnerships in the AI industry.

Microsoft has also been trying to cut costs by using more of its own AI models instead of paying other companies for theirs. Bloomberg reported on July 7, 2026 that the company is replacing OpenAI and Anthropic models with its own in products like Excel. TechCrunch confirmed that reporting the same day, noting the shift as a cost-cutting measure across applications including Word. The costs are significant: The Information reported in January 2026 that Microsoft's spending on Anthropic's AI alone was on pace to reach $500 million (The Information).

Yet Microsoft's business relationships with both companies remain active even as it builds a competitive story. On July 2, 2026, Microsoft launched a new company backed by $2.5 billion to help other businesses adopt AI, using models from both Anthropic and OpenAI (Reuters). That venture, announced less than two weeks before the sales meeting, uses the same models Microsoft is now training its sales force to criticize.

The competitive landscape has shifted in other ways too. OpenAI is preparing to launch GPT-5.6, described as its most advanced AI model (Reuters). Anthropic launched a Claude "Max plan" with higher usage limits and has been hiring aggressively throughout 2026, bringing in executives from OpenAI, Google, Microsoft, xAI, and Stack Infrastructure (CRN). Notably, Anthropic has been recruiting talent from Microsoft itself.

The broader context here is that Microsoft is going through a transition. It used to rely on exclusive access to OpenAI's AI models as the foundation of its AI strategy. Now it has to compete based on the quality of its own platform while still doing business with the very companies it competes against. The partnership change in April was the structural turning point. Switching to its own models in Word and Excel addressed the cost side. The new sales strategy addresses how Microsoft pitches itself to customers.

What is still unclear is whether Microsoft's "full end-to-end system" argument will convince customers who might prefer the best individual AI models from independent providers. Parikh's framing assumes that having everything work together in one package matters more to buyers than having the most powerful AI model on its own. That is a reasonable bet for customers who care most about security, compliance, and having AI built directly into their daily tools. It is a weaker argument for customers whose top priority is the most advanced reasoning or the best performance on specialized tasks, which is where OpenAI and Anthropic have focused their efforts.

The tension is also visible in Microsoft's own choices. The AI consulting firm it launched on July 2 uses models from both competitors. Microsoft's flagship applications are being switched to its own models. The sales force is being trained to point out flaws in those same competitors' products. These moves are not necessarily at odds with each other, but they require careful handling. A salesperson trained to highlight Claude's security weaknesses in Microsoft Office might face a reasonable question about why Microsoft's own consulting arm uses Claude for client work.