Microsoft Corporation has launched a phased process of replacing OpenAI and Anthropic models with its own MAI (Microsoft AI) systems in Excel and Outlook applications. This strategic decision, aimed at reducing costs for expensive third-party AI solutions, has already begun to bear fruit: tens of thousands of requests per week are now processed directly by Microsoft's internal models.
Previously, Copilot functionality in Excel and Outlook relied heavily on the computing power of OpenAI and Anthropic. However, according to my data, the share of MAI models in the total volume of AI requests is still small, but the trend is clear: Microsoft is striving for full autonomy.
Cost savings as the main driver of innovation
The team led by AI division head Mustafa Suleyman does not want to accept the pricing policies of market leaders, especially after the cancellation of preferential access terms that Microsoft had due to its long-standing partnership with OpenAI. In June, Suleyman directly stated that the company intends to reduce spending on Anthropic and gradually transition to its own systems. "We pay Anthropic a lot of money — our task is to gradually reduce costs and completely eliminate unnecessary expenses," he emphasized.
This step is only part of a large-scale optimization program. As early as May 2026, Microsoft began to abandon most internal licenses for Claude Code. In parallel, the corporation is cutting costs in other areas, including laying off 4,800 employees (2.1% of the workforce), with the main blow hitting the Xbox gaming division, where about 3,200 specialists will lose their jobs.
Expert opinion: The transition to proprietary models is not just about cost savings, but a strategic move that will allow Microsoft to gain full control over the technology stack and reduce dependence on vendors. In the long term, this could lead to the emergence of cheaper and more specialized AI solutions for corporate clients, strengthening Microsoft's position in the competitive battle for the enterprise AI market.