The corporate artificial intelligence market is undergoing a tectonic shift. Microsoft, one of the world's largest consumers of AI tokens, has officially launched the process of replacing models from OpenAI and Anthropic with its own developments in key products — Excel and Outlook. This is not just a technical update, but a clear signal of a change in long-term strategy.
Currently, tens of thousands of requests per week are already being processed directly within applications using internal MAI (Microsoft AI) models. Although the share of proprietary solutions in the total volume of AI computations is still small, the trend is obvious and irreversible.
Economics First
The key reason is financial. Microsoft's AI division head Mustafa Suleyman stated back in June the company's intention to reduce spending on Anthropic and gradually transition to its own systems. His words were more than candid: "We pay Anthropic a lot of money — our task is to gradually reduce costs and completely eliminate unnecessary expenses."
This decision is part of a large-scale budget optimization plan. As early as May 2026, Microsoft began phasing out most internal licenses for Claude Code. Now the software giant is taking the next logical step, reducing dependence on external model providers.
Context of Global Cost-Cutting
It is important to understand: the transition to proprietary AI is just one front of a massive cost-reduction campaign. Simultaneously, Microsoft is laying off 2.1% of its workforce — 4,800 people, with the main impact (about 3,200 employees) hitting the Xbox gaming division.
This demonstrates the company's pragmatic approach: record investments in AI do not negate the need for strict financial discipline in other areas.
Cryptalist Analysis: This move by Microsoft is a classic example of vertical integration in the AI era. The company understands that long-term leasing of third-party models entails not only direct costs but also strategic vulnerability. Developing proprietary MAI allows not only for savings but also for controlling the entire technology stack, which is critically important for leadership in the AI arms race. The market should perceive this as a signal: the era of "free" access to advanced models through partnerships is coming to an end.