Microsoft Pivots to In-House AI Models to Curb Rising Infrastructure Costs

The internal shift follows the unveiling of seven new MAI models at last month’s Build conference, ranging from automated coding assistants to text-to-image generators. While Microsoft maintains existing relationships with OpenAI and Anthropic, the move underscores a growing industry-wide push for fiscal discipline. The company declined to provide further details regarding the scope of the transition.

This trend toward internal development mirrors a broader cooling of the AI "tokenmaxxing" phase seen earlier this year. Tech giants including Meta, Amazon, Uber, and Accenture are all aggressively auditing their AI spending as the sticker shock of large-scale deployments hits corporate balance sheets. In some corners of Silicon Valley, the search for efficiency has reached an extreme, with firms even exploring cheaper Chinese models despite significant security concerns.