Mustafa Suleyman, who now leads Microsoft’s AI division, has declared that Microsoft “needs to be self-sufficient in AI” as it builds its own frontier models from scratch—rather than relying solely on external partnerships. He explained that achieving this requires training advanced models “of all scales with our own data and compute at the state-of-the-art level.”
He also revealed that Microsoft has set up a dedicated superintelligence team whose mission goes beyond incremental product improvements: the team focuses on tackling “fundamental problems that present barriers” to next-generation AI, with applications aimed at healthcare, energy and transportation, and a goal of reducing the cost of living for billions of people. At the same time, Suleyman stressed that safety, human-centred design and keeping humans “at the top of the food chain” remain critical amidst the push for high-capability systems.
This shift comes after Microsoft and OpenAI renegotiated their partnership, giving Microsoft greater freedom to develop models independently or with third parties. Suleyman noted that Microsoft is no longer restricted to using only OpenAI’s IP, underscoring that the company sees its scale and ecosystem as enabling a new phase of in-house leadership in AI. In short: Microsoft is moving from being a major partner and investor in AI infrastructure to wanting to own the full stack—data, compute, models and agents—internally.
