Microsoft Moves to Reduce OpenAI Reliance With In-House AI Push

As Neowin reports, Microsoft wants to reduce its reliance on OpenAI and move toward what AI chief Mustafa Suleyman calls “true AI self-sufficiency.” The company plans to build more of its AI stack in-house while still maintaining strong ties with OpenAI.

Microsoft targets AI self-sufficiency beyond OpenAI

Microsoft has already started scaling its own AI infrastructure. This includes the Maia 200 AI accelerator chip and the Fairwater network of large-scale AI data centers and supercomputers.

The company also secured a deal with SK Hynix for HBM3e memory, a key component for high-performance AI workloads. These investments show that Microsoft wants tighter control over both hardware and model development.

The strategy follows a revised October 2025 partnership with OpenAI. Under the updated agreement, Microsoft can build its own frontier AI models and work toward AGI, while retaining a 27% stake in OpenAI and long-term access to its models and intellectual property through 2032.

OpenAI models still power Copilot and Azure services

Despite the push for independence, Microsoft is not cutting ties with OpenAI. OpenAI models will continue powering products like Copilot in the near term.

This shift represents diversification rather than separation. Microsoft sees reliance on a single AI supplier as a strategic risk, especially since any disruption at OpenAI could directly impact Copilot, Office, Azure, and other AI-driven services.

By developing in-house foundation models optimized for its ecosystem, Microsoft expects greater flexibility, improved cost efficiency, and stronger long-term control.

Microsoft already supports multiple leading AI models on Azure. Alongside OpenAI offerings, Azure hosts Anthropic’s Claude, Meta’s Llama, and Mistral models.

This multi-model approach gives enterprise customers more choice while reducing dependency on a single provider. It also positions Microsoft as a neutral AI platform that supports competing frontier models.

Overall, Microsoft’s strategy reflects a hedge against uncertainty in the fast-moving AI race. The company is maintaining its partnership with OpenAI while simultaneously investing heavily in its own chips, data center infrastructure, and foundation models to strengthen long-term technological independence.

In parallel, Microsoft is also exploring the adoption of high-temperature superconducting (HTS) technology as a way to improve data center efficiency by minimizing energy loss and enhancing power delivery for increasingly demanding AI workloads.