Microsoft is developing its own AI reasoning models to compete with OpenAI, signaling a strategic shift to reduce reliance on its key AI partner, according to The Information. The company has been testing models from xAI, Meta, and DeepSeek as potential alternatives for powering its Copilot AI assistant.
Microsoft’s AI division, led by Mustafa Suleyman, has completed training a new family of models, internally called MAI, which reportedly perform close to OpenAI’s and Anthropic’s models on key benchmarks. These models incorporate chain-of-thought reasoning techniques, enhancing their ability to solve complex problems with intermediate reasoning steps.
This move aligns with Microsoft’s broader strategy to diversify its AI infrastructure for Microsoft 365 Copilot and cut costs. Initially built on OpenAI’s GPT-4, Copilot may soon integrate Microsoft’s in-house models. Additionally, Microsoft is considering releasing MAI as an API for third-party developers later this year, further expanding its AI offerings.
Also Read: Microsoft Introduces AI Agent to Enhance Sales Team Efficiency
The development highlights intensifying competition in the AI space, as major tech players work to establish greater independence and innovation in advanced AI models.
