Microsoft AI CEO Mustafa Suleyman has indicated the company intends to develop its own frontier-grade foundation models, a move that could eventually put Microsoft in direct competition with OpenAI. The comments suggest Microsoft is planning for extremely large-scale training needs and a long-term roadmap beyond today’s Copilot stack.
Suleyman said building at the frontier requires massive, reliable infrastructure and world-class research teams, and he framed the next wave of products around autonomous agents that can take on complex work. He also pointed to healthcare as a priority area, describing an ambition to pursue “medical super-intelligence” alongside general-purpose models.
The shift comes as Big Tech weighs the cost of leading-edge AI, with compute becoming a defining constraint for model development. It also underscores that Microsoft’s AI strategy is not limited to a single lab partnership, even as it continues to ship AI features across Windows and Microsoft 365.
