Microsoft is reducing its reliance on OpenAI and accelerating the development of its own, cutting-edge artificial intelligence models, a shift confirmed by the company’s CEO of AI, Mustafa Suleyman. The move signals a significant strategic realignment for the $3 trillion software giant and introduces further uncertainty regarding the financial future of OpenAI.
Suleyman, a co-founder of Google DeepMind before joining Microsoft, stated the company must build its own leading-edge foundation models using massive compute power and highly skilled training teams. Microsoft heavily utilizes OpenAI’s systems, including ChatGPT and DALL·E 3, in its Copilot tools across its product line, such as Microsoft 365 and GitHub Copilot.
However, the partnership between the two companies underwent substantial revisions in October. The restructured agreement grants OpenAI greater freedom to secure external partners and investment in computing infrastructure, allowing Microsoft to potentially reduce its long-term commitment. Microsoft remains a major investor in OpenAI, holding approximately 27 percent of the equity in OpenAI’s capped-profit subsidiary and retaining rights to its models through 2032, although still maintaining access to OpenAI’s most advanced systems.
Suleyman outlined Microsoft’s pursuit of “true self-sufficiency” in AI, noting the company is collecting vast datasets and reorganizing internal teams to achieve this goal. He indicated that Microsoft’s most advanced, internally developed models could be released this year, following billions of dollars invested in expanding its data centers and compute capacity. Microsoft is currently anticipating capital expenditures of around $140 billion in the fiscal year ending in June, with a substantial portion allocated to AI infrastructure, including servers, network equipment and power supplies.
The shift comes as investor concerns about an AI spending bubble have put pressure on the stock prices of major technology companies. Microsoft shares have fallen by more than 13 percent in the last month. OpenAI as well faces significant financial challenges, having committed to future compute orders estimated to be worth over $1 trillion over time, and has yet to achieve sustained profitability, relying on continued capital infusions from major investors like Amazon and SoftBank.
Microsoft has diversified its AI investments beyond OpenAI, investing in Anthropic and Mistral, while simultaneously accelerating its internal research programs. Competition in the enterprise AI space remains intense, with Anthropic establishing itself in AI-powered coding tools, challenging Microsoft’s dominance with GitHub Copilot, and Alphabet’s Google expanding aggressively into AI services for businesses. OpenAI is also now directly targeting large enterprise customers.
Suleyman predicted a dramatic transformation in office worker roles within the next 12 to 18 months, suggesting that many computer-based tasks currently performed by lawyers, accountants, and project managers could be automated. He further argued that AI systems will be coordinating workflows within large organizations within two to three years, learning from experience and becoming increasingly autonomous.
Suleyman likened the development of custom AI models to creating digital content, forecasting a future where companies can develop AI systems tailored precisely to their needs. Microsoft’s desire to control its core technology, rather than relying solely on an external partner, is intended to reduce strategic risks and strengthen its negotiating position. However, Microsoft must balance this independence with its existing agreements, continuing to benefit from privileged access to OpenAI’s latest developments and the strength of the OpenAI brand and developer ecosystem.
Tensions between the two companies have surfaced previously, with leadership disputes and controversies surrounding corporate governance periodically straining the relationship. The rapid growth of the AI industry, with companies competing to integrate generative AI tools into customer service, programming, marketing, and internal operations, adds to the complexity. While Microsoft is seeing strong demand for its Copilot products, its AI initiatives for consumers have yielded mixed results, and the company must justify its substantial investments to shareholders.
Suleyman characterized the strategy as necessary for long-term competitiveness, arguing that state-of-the-art AI technology requires gigawatt-scale compute power and top-tier research teams. Microsoft appears determined to build these capabilities directly, a move that could fundamentally alter the power dynamics between major technology companies and the world’s leading AI startup.
