Microsoft Plans to Reduce OpenAI Dependency
Microsoft's AI chief confirmed the company plans to reduce its dependency on OpenAI to gain greater strategic independence. The move follows rising tensions over model exclusivity and vendor lock-in, with OpenAI reportedly facing capital constraints. Following the news, Microsoft's software business was downgraded by analysts concerned about the potential loss of AI exclusivity.
- Microsoft's "self-sufficiency" push is being led by Mustafa Suleyman, co-founder of DeepMind, who was hired in early 2024 to lead Microsoft's internal AI team. This strategic shift was solidified after a restructuring of the Microsoft-OpenAI partnership in October 2025. - As part of this internal strategy, Microsoft is developing its own large-scale AI models, including MAI-1, a large language model with a reported 500 billion parameters. This is a significant in-house effort, though still smaller than OpenAI's GPT-4, which is estimated to have over a trillion parameters. - The revised partnership agreement with OpenAI, settled in October 2025, allows Microsoft to pursue the development of its own frontier models and even AGI independently. While Microsoft's intellectual property rights to OpenAI's models are extended through 2032, the new terms also grant OpenAI more freedom to collaborate with other companies. - The financial terms of the partnership have also been a point of negotiation, with reports suggesting that OpenAI plans to reduce Microsoft's revenue share from the current 20% to as low as 8-10% by 2030. This could potentially shift over $50 billion in additional revenue to OpenAI. - Microsoft has invested over $13 billion into OpenAI, giving it a 27% stake in the company. In return for this investment, OpenAI has committed to purchasing an additional $250 billion in Azure services. - Analyst downgrades from firms like Stifel and Melius Research cite concerns over increased competition in the AI space from companies like Google and Anthropic. There are also concerns about the high capital expenditures required for AI development, which could impact Microsoft's free cash flow. - The development of in-house models like MAI-1 is seen as a way for Microsoft to have more direct control over the technology powering its products, such as Copilot, and to optimize for its specific needs. Microsoft has reportedly been critical of GPT-4's cost and speed for certain consumer applications. - Beyond OpenAI, Microsoft is diversifying its AI investments, including a partnership with French AI company Mistral. This is part of a broader strategy to avoid over-reliance on a single partner and to have flexibility in choosing the best model for a given task.