How the OpenAI and Microsoft deal is coming undone Business News & Hub

There is a kind of awkwardness in a relationship when both parties know something has changed, but neither wants to say it out loud. The partnership between Microsoft and OpenAI seems to have arrived at exactly that moment.

The duo that joined hands even before the ChatGPT-triggered AI boom took the tech world by storm.

Microsoft invested a billion dollars in OpenAI, back in June 2019, to “build beneficial AGI.” The move was aimed at extending Microsoft’s cloud capabilities in large-scale AI systems that were still not completely built. The logic was clean and mutually beneficial. OpenAI needed Microsoft’s money and cloud infrastructure to build large language models, and Satya Nadella needed Sam Altman researchers and cutting-edge technology. The partnership worked fine for both sides.

In late 2022, ChatGPT exploded into public consciousness, making Microsoft increase its funding in OpenAI with a $10 billion investment in January 2023 as the software maker rushed to integrate the technology into Bing — the company’s search engine — and its broader Office suite, which is now rebranded as Copilot.

At that point Mr. Nadella taunted Google’s Sundar Pichai, asking to show what his company could do with AI, making Mr. Pichai’s researchers double down on building Gemini. But the past few years have been quite dynamic and signs of strain in the partnership between the ChatGPT maker and the Windows maker were becoming visible.

The first crack

The first real crack appeared when Microsoft was blindsided in November 2023 when OpenAI’s board abruptly fired Mr. Altman. Mr. Nadella quickly moved to bring Mr. Altman back. After Mr. Altman took back his position, dissolving most of the board, Microsoft obtained a non-voting observer seat — a concession that looked like influence but was, in practice, something far more modest.

What then followed was a slow, deliberate unravelling. In March 2024, Microsoft worked on a strategy that would make it build its own in-house AI. In a $650 million deal, the Richmond-headquartered company acquired Inflection AI and hired its founder, Mustafa Suleiman to lead Microsoft AI.

OpenAI read that signal and responded accordingly. In June 2024, it signed a roughly $300 billion deal with Oracle, a move that would loosen Microsoft’s grip on the startup for compute infrastructure. Rather than end the exclusivity clause with Microsoft, this collaboration was presented as a three-way deal between Microsoft, OpenAI and Oracle, in which OpenAI would continue to pre-train its frontier models on Microsoft Azure. It was the beginning of a pattern that would repeat itself with increasing boldness.

According to a report by The Information, OpenAI had proposed capping Microsoft’s future equity stake at 33% in exchange for it foregoing rights to future profits. Microsoft, which by this point had invested over $13 billion, had balked at these terms. OpenAI also wanted eliminate the clause that gives Azure exclusive cloud hosting rights.

A truce was eventually reached in which Microsoft and OpenAI signed a non-binding memorandum of understanding in September 2025, fundamentally restructuring how enterprises would deploy artificial intelligence. The formal renegotiation came in October, when Microsoft acquired a 27% stake in OpenAI Group PBC, valued at approximately $135 billion at the time of signing. In exchange, Microsoft’s former right of first refusal to be OpenAI’s sole compute provider was formally removed.

Unresolved

The deal was packaged as a new chapter, but the underlying discomfort had not been resolved, only paused. Per a CNBC report citing a document circulated to prospective investors, OpenAI said that Microsoft is responsible for “a substantial portion of our financing and compute,” and warned that any modification or termination of the commercial partnership could adversely affect OpenAI’s business. The AI company also stated that its operating results would depend on its ability to successfully develop relationships with partners beyond Microsoft.

While an OpenAI spokesperson told the outlet “This is a standard legal risk factor disclosure, unrelated to any potential IPO prospectus,” it does reflect the reality that a company is telling investors that its most important relationship is also one of its greatest vulnerabilities.

And OpenAI has also been busy raising funds and partnering with other players in the ecosystem. In February, the company announced a combined $110 billion funding at a valuation for $730 billion from Amazon, Nvidia and Softbank. After this funding round, the ChatGPT maker has expanded its access to compute infrastructure as Amazon’s compute commitments to OpenAI now exceeds Microsoft’s.

Competitive partnership

Separately, OpenAI is also selling ChatGPT Enterprise, a product that rivals Microsoft’s Copilot offering. This effectively makes the two companies competitors even as they are bound together by billions of dollars in contracts, IP licences and revenue-sharing agreements.

Microsoft, to its credit, has been playing its own long game. It has begun to offer Anthropic’s Claude 4 within Office 365, revealing that it isn’t trapped by OpenAI’s models. In fact, Microsoft began its reassessment exercise back in 2024 when it added OpenAI to the list of companies it considers competitors in its annual report. And with the Inflection AI acquisition the software maker sent a clear signal that OpenAI is not its AI strategy, but merely a part of it.

What remains in place, for now, is the contractual scaffolding that neither side has fully dismantled. Azure remains the exclusive cloud provider for stateless OpenAI APIs, and Microsoft maintains its exclusive license to OpenAI’s intellectual property across all models and products. Microsoft will relinquish rights to OpenAI’s research once an independent expert panel verifies AGI, or by 2030, whichever comes first.

The AGI clause, in particular, is one of the stranger provisions in recent tech history — a contractual trigger for an event that most experts believe does not yet exist. Both companies felt it necessary to restate in February 2026 that its definition remained unchanged. The fact that they needed to say so tells you something about the state of trust between them.

The growing-up phase

There is a school of thought that says all of this is simply what successful partnerships look like when the companies involved grow up. What starts as a patron-and-dependent relationship inevitably evolves, and evolution is not the same thing as collapse.

Microsoft got what it needed from the partnership — an early lead in the AI era, the credibility of being the company that backed ChatGPT, the infrastructure deals and Azure growth that followed. OpenAI, in turn, got the capital and computing resources it needed to become the defining AI company of its generation. Both parties have now outgrown the original arrangement.

But outgrowing a partnership and peacefully departing from it are two different things. OpenAI is headed toward a public market debut — expected in the second half of 2026 — at a potential valuation approaching one trillion dollars. That valuation will be scrutinised intensely, and prospective shareholders will spend a great deal of time reading precisely the kind of risk disclosures from OpenAI.

The message to investors is mixed and, on close reading, somewhat alarming. We are, says OpenAI, a company with 900 million weekly active users and $13.1 billion in 2025 revenue. We are also a company whose future is materially dependent on the goodwill of a partner that now competes with us and which has been preparing its own AI capabilities for years.

The clouds over this particular affair have been gathering for a while. The renegotiated deal of last year bought time, but it did not buy resolution. OpenAI is diversifying its way toward independence, and Microsoft is hedging its way toward self-sufficiency. And somewhere in the fine print of investor documents, the truth will keep quietly presenting itself.

Published – March 26, 2026 08:03 am IST