Microsoft and OpenAI have announced a new definitive partnership agreement that restructures their relationship and sets out how the companies will collaborate as artificial intelligence advances toward Artificial General Intelligence (AGI).
The deal marks the latest evolution of a partnership that began in 2019, initially as Microsoft’s investment in a research-focused startup and later becoming one of the most influential alliances in global technology.
Under the new arrangement, Microsoft is supporting OpenAI’s plan to transition to a public benefit corporation, or PBC.
Following a recapitalization tied to this structural shift, Microsoft’s investment in the newly formed OpenAI Group PBC is valued at approximately $135 billion.
This represents about 27 percent ownership on an as-converted diluted basis across all stakeholders, including employees, outside investors, and the OpenAI Foundation.
Prior to recent funding rounds, Microsoft held roughly 32.5 percent in the company’s for-profit entity.
The agreement preserves core components that defined the collaboration to date.
OpenAI remains Microsoft’s exclusive partner for frontier AI models, and Microsoft continues to hold exclusive rights to OpenAI’s model-related intellectual property and Azure cloud API access until OpenAI formally declares AGI.
However, that declaration will no longer be solely determined by OpenAI. Under the new terms, the claim that AGI has been achieved must be independently verified by a panel of external experts.
The deal also introduces new flexibility for both organizations. Microsoft’s intellectual property rights related to OpenAI models and products now extend through 2032 and include rights to post-AGI models under certain safety and governance restrictions.
Microsoft’s rights to research-related IP—defined as confidential methods used to build models and systems—remain in place until either AGI is independently verified or until 2030, whichever occurs first.
Research IP does not include model architecture, weights, inference or finetuning code, or any hardware and data-center-related IP, which Microsoft retains rights to.
The revised terms expand OpenAI’s freedom to collaborate beyond Microsoft. OpenAI may now jointly develop products with third parties.
API-based products must continue to run on Microsoft’s Azure cloud, but non-API products may be hosted on any cloud provider. Microsoft also gains the right to pursue AGI development independently or with other partners.
If Microsoft uses OpenAI’s intellectual property to pursue AGI before OpenAI declares it, the development is subject to substantial compute limits designed to protect against uncontrolled escalation.
Financially, the agreement allows OpenAI to provide API access to U.S. national-security customers regardless of cloud provider, and permits the company to release open-weight models that meet agreed capability criteria.
OpenAI has also contracted to purchase an additional $250 billion worth of Azure cloud services, though Microsoft will no longer hold the right of first refusal to serve as the company’s primary compute provider.
The revenue-sharing agreement between the companies remains in place until AGI is verified, with payments now spread over a longer period.
The companies framed the announcement as a progression into a new phase of collaboration—one that balances shared innovation with greater autonomy.
In the joint statement, Microsoft emphasized that, as the partnership moves forward, both organizations are positioned to continue creating “real-world value” while expanding opportunities for users, developers, and businesses.
The agreement underscores the rapidly shifting landscape of AI development, where strategic partnerships, cloud computing scale, governance, and intellectual property rights are increasingly intertwined with the race toward advanced, general-purpose AI systems.
Also Read: TVS Motor Posts 37% Jump in YoY Net Profit