Microsoft and OpenAI Are Rethinking Their Partnership as AI Gets More Powerful

Microsoft and OpenAI have reached a new deal that fundamentally changes how they work together. The two companies have been partners since 2019, but as OpenAI builds more powerful AI systems, both companies need more flexibility. The updated agreement lets OpenAI build its own computer servers for the first time, while Microsoft keeps certain rights to OpenAI's technology until either the companies agree that AGI has been achieved or the year 2030 arrives — whichever comes first.

What Is Happening and Why

To understand the shift, it helps to know what AGI means. AGI stands for "artificial general intelligence" — it's the idea of an AI system that can learn and solve problems across many different fields the way a human can, rather than being limited to one specific task like playing chess or answering customer service questions. Neither OpenAI nor Microsoft claims to have built AGI yet, but it is central to their partnership agreement.

Under the original deal in 2019, Microsoft invested $1 billion in OpenAI and provided all the computer power needed to train OpenAI's AI models. Think of it like renting warehouse space — OpenAI used Microsoft's servers (called Azure) to build and run its AI systems. This was a tight arrangement that worked while OpenAI was smaller.

The new deal lets OpenAI rent server space from other companies too. OpenAI has already signed agreements to use computer servers from Oracle in Texas. This gives OpenAI more room to grow and less dependence on Microsoft alone.

The Money and Corporate Structure

OpenAI recently changed from being entirely a nonprofit organization to a mix of nonprofit and for-profit company. The nonprofit part now owns about $100 billion worth of the for-profit part. This split addresses a long-standing awkwardness: OpenAI was founded to advance AI safely for humanity, but it also needed to earn money and grow like a business. The new structure tries to balance both goals.

Microsoft still owns about 27% of OpenAI's for-profit side, making it the largest shareholder.

What This Means for Microsoft's Rights

Microsoft no longer gets to use everything OpenAI builds. Instead, Microsoft has exclusive rights to the methods OpenAI uses to create its AI systems, not the systems themselves. This is an important distinction. It means Microsoft can access some of OpenAI's "secret recipe," but OpenAI can sell its finished AI systems to other companies without giving Microsoft a cut.

Microsoft still has preferred access to OpenAI's AI services through a product called Azure OpenAI Service, which several large companies already use — including healthcare organizations and defense contractors. But OpenAI can now build direct relationships with other customers and offer its AI products in different ways.

In recent documents filed with the U.S. Securities and Exchange Commission, Microsoft has called OpenAI both a partner and a competitor. This reflects the new reality: the two companies still need each other, but they are also working independently in some areas.

When the Partnership Could End

The new agreement includes a trigger for when the partnership might end: when an independent group of experts decides that AGI has been created. Neither company alone gets to make that call anymore. An external panel will judge whether AGI has truly arrived.

If AGI is achieved, the revenue-sharing arrangement between Microsoft and OpenAI would change. Exactly how remains unclear. If AGI has not been achieved by 2030, the partnership agreement expires anyway.

The basic tension here is that neither company wants the other to control them once AI becomes very powerful. By creating an outside review process, both Microsoft and OpenAI are trying to solve a real problem: how do you know when a landmark has been reached if there are no agreed-upon standards for measuring it yet.

Why the Partnership Is Changing

From a broader view, this shift makes sense. Over the past few decades, we have seen this pattern before. When companies start out, they often depend heavily on a single partner for crucial resources. But as they grow and their needs become more complex, they need flexibility that single partnerships cannot always provide. OpenAI is at that stage now. Its computing needs have grown beyond what one company can comfortably supply, and its customers want options.

Microsoft invested in OpenAI because it believed in the technology and wanted a seat at the table as AI developed. That bet looks good today — OpenAI's ChatGPT is widely used, and AI has become central to Microsoft's business strategy. But as OpenAI grows, it needs to operate more independently. Both companies seem to have accepted this.

The new structure also gives both companies protection. Microsoft secures the right to keep using OpenAI's most advanced AI systems through 2030. OpenAI gains the freedom to scale its operations without being held back by a single supplier. The external panel for judging AGI gives both parties a way to handle the biggest unknown: what happens if AI really does become as powerful as some people expect.

What Comes Next

The deal has been approved by state regulators in Delaware and California, so it is now official. Both companies will probably find ways to work together even as they pursue separate goals. Microsoft will keep integrating OpenAI's AI into its products and services for business customers. OpenAI will expand its own partnerships and offerings.

The real test will come if and when that outside expert panel needs to make a judgment about AGI. That is a decision no one has had to make before, and the stakes are high for both companies. But for now, the partnership restructuring lets both Microsoft and OpenAI do what they need to do: grow, invest, and build the AI systems they believe will shape the future.