One of tech’s most-watched power couples — OpenAI and Microsoft — appears to be redefining their relationship.
The pair first linked up in 2016, just one year after OpenAI was founded, forging a partnership that would reshape the trajectory of artificial intelligence forever. Microsoft’s vast cloud infrastructure and deep pockets fueled OpenAI’s cutting-edge research, launching GPT-3 — the first large language model to demonstrate near-human text generation capabilities — in 2020, and later ChatGPT in 2022, which pushed conversational AI into the mainstream. Their collaboration powered ChatGPT’s meteoric rise, and embedded OpenAI’s models into just about every Microsoft product, cementing both companies as central players in the AI boom.
Microsoft and OpenAI have signed a new, non-binding agreement that shifts their once-exclusive, tightly controlled partnership into more of a flexible arrangement. Among other things, it gives OpenAI the room to restructure into a traditional for-profit company
(and potentially undergo an IPO down the line), as well as the freedom to work with other cloud providers like Oracle. Microsoft, meanwhile, can diversify its own AI offerings and gains a new equity stake in OpenAI.
But what once looked like one of the industry’s most stable alliances is starting to fray, as both sides pursue their own ambitions in an increasingly crowded, feverishly competitive landscape. All told, Microsoft has poured more than $13 billion into OpenAI in exchange for early access to its technology and nearly half of its profits, but OpenAI now seems eager to be more independent. It recently struck a $300 billion cloud deal with Oracle that directly challenges Microsoft’s Azure empire, and has been pulling in massive funding rounds from other major investors. Microsoft, meanwhile, is forging its own path, acquiring top AI talent through sweeping acquisitions and even striking a deal with OpenAI rival Anthropic to integrate Claude into its Office suite.
The two companies recently signed a non-binding memorandum of understanding that feels more like a polite handshake than a warm embrace. On paper, the partnership remains, but in practice it is a high-stakes balancing act: Both still benefit from shared infrastructure and branding power, yet both seem to be preparing for a day when they might need to stand on their own.
All in all, the love may still be there, but the honeymoon phase is definitely over.
What Changed?
The first cracks began to show earlier this year, when OpenAI announced plans to develop its own supercomputing infrastructure, meaning it wouldn’t be entirely dependent on Microsoft Azure anymore. It also began working on a professional networking platform that will likely compete against LinkedIn, one of Microsoft’s flagship products. Undoubtedly though, the $300 billion deal with Oracle is the clearest sign yet that the relationship is shifting.
Others trace the rift further back. In 2023, OpenAI reportedly urged Microsoft to delay rolling out GPT-4 into its Bing search engine, warning that the model wasn’t ready for prime time. Microsoft pressed ahead anyway, and when Bing Chat launched that February, it quickly went viral for all the wrong reasons — gaslighting users, hurling insults, sulking and even professing love. The fiasco forced Microsoft into damage control, capping conversation lengths and spending months ironing out the worst kinks. That tension only deepened later in the year, when OpenAI’s board abruptly ousted CEO Sam Altman for “communication issues,” only to reinstate him days later. For Microsoft, which had staked billions on OpenAI’s leadership and stability, the chaos served as a wake up call to just how fragile the partnership could be.
Since then, Microsoft has been hedging its bets. It quietly absorbed most of Inflection AI’s team and developed its own in-house models, MAI-1-preview and MAI-Voice-1. Add those deals with rivals like Anthropic, and it’s clear the relationship didn’t rupture overnight — it’s been drifting for some time.
Wait, Doesn’t Microsoft Own OpenAI?
Despite common misconception, Microsoft does not own OpenAI. While the company was previously listed on OpenAI’s website as a minority owner, the listing has since been updated to clarify that Microsoft holds only a minority economic interest, not equity.
It’s best to describe these two companies as close strategic partners. OpenAI remains fully independent, operating as a public benefit corporation (formerly a capped-profit LLC) that’s governed by a non-profit parent, OpenAI Nonprofit, which makes all the strategic and operational decisions, from research priorities to safety and mission alignment. This structure allows OpenAI to raise capital through its for-profit subsidiary, OpenAI LP, while capping investor returns and requiring that most of the company’s value be directed toward the “benefit all of humanity.”
Much of the confusion comes from Microsoft’s reported 49 percent economic stake in OpenAI, which currently gives it nearly half of the company’s profits from commercial products like ChatGPT Plus subscriptions. However, this portion is expected to shrink to a 29 percent equity stake under the new terms.
So, Are Microsoft and OpenAI Breaking Up?
Not exactly. The new memorandum of understanding doesn’t dissolve the relationship, but it does make it clear that things are no longer exclusive. Both parties have clearly started pursuing outside opportunities, shifting the arrangement into more of an open relationship.
Unless their alliance sours, both companies seem invested — but there are clear scenarios where either might walk away. Microsoft could pull back if OpenAI’s moves start cutting into its business, like this Oracle deal or the potential Linkedin rival. Another factor would be if OpenAI’s profit-sharing and governance structure limits Microsoft’s influence, or diminishes the value of its exclusive access. Historically, Microsoft has left partnerships when strategic priorities diverged. For example, it retreated from its $43.7 billion bid for Yahoo in the late 2000s after search and advertising ambitions clashed.
Though it might seem like OpenAI is more inclined to break free from Microsoft, leaving comes with some major risks. It could get swallowed whole by larger competitors like Google or Meta, or get lost in the rapidly consolidating AI market with one bad product launch. Then again, it might not have a choice if Microsoft’s growing AI ecosystem starts to feel restrictive or interferes with its pace of innovation. Regardless of the multi-billion support and many years of collaboration, this dependence on Azure and Microsoft’s infrastructure could limit OpenAI’s flexibility, especially as it pursues multi-cloud strategies and high-profile enterprise deals elsewhere. Both companies are also susceptible to talent and intellectual property conflicts, since they’re drawing from the same pools to build competing AI tools.
In short, this partnership will only survive so long as both companies see collaboration as more beneficial than going solo.
Inside Microsoft and OpenAI’s Partnership
Here’s a quick look back at Microsoft and OpenAI’s nine-year relationship, with all the receipts to prove it.
2016: Freshly founded OpenAI selects Microsoft Azure as its exclusive cloud provider, marking the beginning of a beautiful strategic collaboration.
2019: Microsoft invests $1 billion in OpenAI to build artificial general intelligence through a joint hardware and software platform within Azure. In this deal, OpenAI also designates Microsoft as its preferred partner for commercializing AI technologies.
2021: Microsoft invests an additional $2 billion into OpenAI. That same year, Microsoft-owned GitHub launches a new-and-improved version of Copilot, now powered by OpenAI’s Codex model designed to translate natural language into code for developers, showcasing how the startup’s tech could turn out real-world products.
January 2023: The companies extend their partnership with a multiyear, multibillion-dollar investment.
February 2023: Microsoft integrates OpenAI’s GPT‑4 into Bing and Edge, and rolls out Copilot features across Microsoft 365 apps like Word, Excel and Outlook, bringing AI-assisted writing, summarization and insights directly to users.
June 2023: Tensions surface when Microsoft rushed GPT-4 integration into Bing, and OpenAI warned against premature launch.
November 2023: OpenAI abruptly fires CEO Sam Altman, with the board stating he “was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities.” Following intense pushback from employees, investors and industry figures, Altman is reinstated days later.
March 2024: Microsoft poaches the majority of Inflection’s 70-person team, including two of its co-founders, and goes in on a $650 million licensing deal that would allow it to use Inflection’s AI models, potentially challenging OpenAI’s dominance by building its own in-house AI capabilities. The hiring spree raised regulatory concerns, leading to investigations by both the U.S. Federal Trade Commission and the UK’s Competition and Markets Authority.
January 2025: OpenAI and Microsoft agree to let OpenAI restructure its corporate setup, giving it more control over its operations and compute resources. At the same time, OpenAI launches a $500 billion Stargate Project with SoftBank, Oracle and MGX to build new AI infrastructure across the U.S. over the next four years. These moves signal a major shift in OpenAI’s strategy, giving the company more control over its hardware and cloud resources while doubling as an opportunity to diversify its partnerships.
February 2025: Continuing its sweeping AI talent acquisitions, Microsoft acquires 24 AI engineers from Google’s DeepMind, including the former head of engineering for Google’s Gemini chatbot, to strengthen its Copilot AI efforts.
March 2025: OpenAI acquired $350 million worth of CoreWeave shares, beefing up its computing power with additional AI infrastructure independent of Azure.
April 2025: Microsoft announces it will no longer support ChatGPT training, as OpenAI gains more financial autonomy following a $40 billion investment from Softbank.
June 2025: OpenAI signed a $300 billion deal with Oracle for 4.5 gigawatts of additional compute capacity, further reducing reliance on Microsoft Azure.
July 2025: The Stargate Project scales, as OpenAI, Nvidia, Cisco and co. announce a 5-gigawatt UAE-U.S. AI data center, “Stargate UAE,” in Abu Dhabi, designed to support large-scale AI training and infrastructure outside the U.S.
August 2025: GPT-5 launches. Microsoft continues integrating OpenAI models into its own products, maintaining exclusive cloud and commercialization rights, but starts to face mounting competition from OpenAI’s independent infrastructure.
September 2025: OpenAI announces the Stargate UK project, where it will build the nation’s largest supercomputer alongside Nvidia and Nscale. Microsoft responds with a $30 billion “Tech Prosperity” spending package toward its own UK AI infrastructure commitments, underscoring a growing competitive edge between the two.
Later that month, after years of negotiations and evolving terms, Microsoft and OpenAI sign a non-binding memorandum of understanding, signaling a new phase in their partnership. The landmark agreement that will allow OpenAI to transition into a for-profit organization and unlock access to funding it urgently needs. At the same time, OpenAI announces plans to cut revenue sharing with Microsoft down to eight percent by 2030, which is projected to generate over $50 billion in additional revenue for the AI startup. This marks a decisive step toward greater independence while still maintaining strategic ties.
Does Microsoft still own 49 percent of OpenAI?
As of September 2025, Microsoft is entitled to 49 percent of the profits from OpenAI’s for-profit subsidiary, but it does not have any direct ownership of the company. However, this portion is expected to shrink to 29 percent under the terms of the new agreement, and OpenAI has announced plans to cut revenue sharing with Microsoft down to 8 percent by 2030.