According to a Bloomberg report on Tuesday, Microsoft has started routing a portion of user prompts in popular Office applications to its own MAI (Microsoft AI) models as part of a cost‑saving strategy. This represents a shift away from relying solely on third‑party models from providers such as OpenAI and Anthropic for some Office 365 functionality.
What happened and when
Bloomberg says Microsoft previously highlighted that much of Office 365 was powered by models from OpenAI and Anthropic. In recent months, however, the company has been building out its own AI agents: at its annual Build developer conference last month, Microsoft announced the launch of seven new MAI models, including an agentic coder and a text‑to‑image generator.
Microsoft told TechCrunch it had nothing further to share when asked for comment.
Why this matters
The move fits into a broader industry trend of tighter AI spending. After a brief wave of “tokenmaxxing” earlier this year, recent reporting has shown major tech companies acting more frugally. Other large firms — including Amazon, Uber, Meta, and Accenture — have reportedly taken steps to restrain AI and other expenditures.
Providing and purchasing AI services has become expensive, creating pressure to find lower‑cost options. Some companies have even explored Chinese models as more affordable agentic solutions, despite potential security concerns.
Implications and next steps
While Microsoft continues to use third‑party models, increasing the share of in‑house models could reduce reliance on external services and the associated costs. The trend toward internal models and alternative vendors will raise further questions about balancing cost savings with security and model quality.
Bloomberg's report did not provide exact percentages or long‑term strategic details, and Microsoft has not offered additional public comment.
