The Microsoft-OpenAI divorce is coming. Who’s getting the best deal?

After more than a year of increasingly bitter relations, the one-time bromance between Microsoft and OpenAI is all over except for the shouting — the companies have agreed to a vague framework for their impending divorce. 

Who got the better of the deal? To find out, we need to revisit the history of their partnership, then examine the possible terms of the breakup.

The Microsoft-OpenAI relationship soured largely thanks to what in the tech world takes the place of the language of love — money. Starting in 2019, Microsoft began showering the startup OpenAI with $13 billion in investments. In return, Microsoft got a piece of the company, though details about how much were never clear. Microsoft was also given the exclusive right to use OpenAI’s ChatGPT in whatever way it wanted. That allowed Microsoft to quickly jumpstart its generative AI (genAI) offerings by using it as the brains for its genAI-based Copilot tool.

Since then, OpenAI decided it would eventually go public, and the two companies have been fighting over what Microsoft should get for its investment. At issue, among other things, is what percent of the company Microsoft would own, whether Microsoft would retain exclusive rights to OpenAI technology, and whether OpenAI would be free to make deals with Microsoft rivals.

Declaration of a tech war

Disagreements over the terms of a deal eventually turned into open warfare. OpenAI is considering launching a productivity suite to compete against Microsoft 365. The company also hinted that it might publicly accuse Microsoft of engaging in anticompetitive behavior and ask the federal government to investigate whether Microsoft’s contracts with it violate antitrust laws. Microsoft considered walking away from negotiations with OpenAI, which could make it impossible for OpenAI to go public. 

Cooler heads have prevailed; it appears neither OpenAI nor Microsoft are considering those nuclear options, at least for now.

Frenenemies forever…, with the emphasis on ‘enemies’

The companies have been fighting with each other for well over a year, and in the last several months, it’s become abundantly clear a big breakup is due. 

One sign of that came when SoftBank announced it would invest $40 billion in OpenAI. More recently, OpenAI signed a five-year, $300 billion deal to buy cloud services from Oracle rather than Microsoft. And most recently, chipmaker Nvidia just announced it will invest up to $100 billion in OpenAI.

Microsoft, in turn, launched its first large language models (LLMs), which ultimately will augment or possibly replace ChatGPT as Copilot’s brains. It also made a deal with OpenAI competitor Anthropic to power parts of Microsoft 365 Copilot with Claude. That deal fixes one of Copilot’s problems — its Excel capabilities are exceedingly weak because ChatGPT’s are; Claude’s abilities are much better.

 Those are just some of the initial steps Microsoft has made towards improving Copilot with a “best-of-breed” approach that relies on multiple LLMs, not just ChatGPT.

A vague deal emerges

In mid-September, Microsoft and OpenAI jointly announced they had reached a memo of understanding about what Microsoft will get for its investment when OpenAI’s for-profit arm goes public. 

Details are sketchy, but the deal might have Microsoft give its OK to OpenAI’s restructuring in which OpenAI’s nonprofit arm will retain control of the new for-profit spinoff. The deal could also change how much ongoing OpenAI revenue Microsoft will get for its past investments. (It’s currently 20%, and will likely be negotiated down.)

For Microsoft to give OpenAI all that, it likely wants to retain the right to use ChatGPT in whatever way it wants. The deal might also give Microsoft’s Azure the exclusive rights to  OpenAI’s APIs.

As to how much money Microsoft might wind up with, whether Microsoft will get exclusive rights to ChatGPT, or what percentage ownership of OpenAI’s for-profit arm the company will get, that’s anyone’s guess.

Who profits more from the investment?

So, who’s gotten the most out of Microsoft’s investment? Although both companies have been improved by it, Microsoft will end up with a brighter future than OpenAI.

Microsoft’s $13 billion investment in OpenAI was largely not in cash, but in providing free computing resources. Microsoft’s valuation before its initial OpenAI investment in 2019 was approximately $1.2 trillion. Today, the company is valued at more than $3.8 trillion. That’s a hefty six-year payoff for a relatively little amount invested. 

Even more important, the investment quickly made Microsoft the world leader in genAI without it having to do research and development on its own.

OpenAI benefited, too. Thanks to Microsoft’s initial funding, it found itself on solid financial footing and with enough resources to pursue full-speed-ahead research. Since then, it’s gotten more than $100 billion from other investors.

Even so, Microsoft appears to be much better positioned for the future. The genAI tech market is in the middle of a bubble. Investors are pouring in billions; valuations are sky-high. But increasingly, enterprises say they’re not seeing the payoff they hoped for from the technology. The bubble will burst. And when it does, OpenAI will have a harder time recovering than Microsoft. 

Microsoft offers a lot more than just genAI tools — it also has Windows, the Microsoft 365 productivity suite, a vast cloud business, and much more. Even in an AI bust, Microsoft should thrive.

OpenAI will survive an AI bust, too — but in weakened state, because it’s a one-trick pony.  It’s got no other business. So, no matter what terms OpenAI and Microsoft eventually agree to, Microsoft will have gotten the better of the investment.