Tech giant Microsoft is beginning a large-scale transition to its own artificial intelligence models in key products. Tens of thousands of queries in Excel and Outlook are now being processed by internal MAI (Microsoft AI) systems, marking a shift in the company's strategy regarding third-party developers, including OpenAI and Anthropic.
This decision is not just a technical update, but a calculated economic maneuver. Previously, these applications relied primarily on solutions from OpenAI and Anthropic. However, as practice shows, dependence on external providers is becoming costly. The leadership of the AI division, headed by Mustafa Suleyman, directly points to the need to reduce expenses. In June of this year, he emphasized that the company pays Anthropic significant sums and intends to gradually eliminate these costs by fully transitioning to its own developments.
Cost Savings as the Top Priority
The transition to MAI is part of a broader optimization program. Despite record investments in artificial intelligence, Microsoft is cutting budgets in other areas. The company is already reducing its workforce by 2.1% (approximately 4,800 employees), with the main impact falling on the Xbox gaming division, where about 3,200 specialists will lose their jobs.
Significantly, as early as May 2026, Microsoft began gradually phasing out most internal licenses for Anthropic's Claude Code. This confirms the systemic nature of the changes: the company is consistently reducing its reliance on external models in favor of its own, cheaper solutions.
Analytical commentary from Cryptalist: This move by Microsoft is a clear signal to the market that the era of "free" integrations with AI leaders is coming to an end. Large corporations are moving toward vertical integration to control costs and avoid dependence on the pricing policies of OpenAI and Anthropic. For the crypto industry, this is also an important trend: if traditional giants move away from expensive cloud-based AI solutions, then decentralized networks with their flexible pay-per-compute model could gain new momentum for development.