If Sam Altman is reinstalled as CEO of OpenAI, he will likely have one person to thank: Satya Nadella.
Some have the Microsoft CEO for allowing the tech giant to become so dependent on Altman’s AI startup without guaranteeing it had at least one seat on OpenAI’s board. But Microsoft has a tremendous amount of leverage in the dramatic whirlwind of events this weekend—far more than any of OpenAI’s other backers.
On Saturday, Altman was reportedly in negotiations to return to OpenAI as CEO, with the OpenAI non-profit board possibly resigning as part of the deal, with much of the impetus coming from Microsoft, with support from other OpenAI investors and company employees. The company’s chief strategy officer Jason Kwon, in a memo obtained by The Information, told staff that the company was “optimistic” Altman and other senior staff who resigned in protest at his abrupt firing could be brought back to the company.Microsoft’s whip hand, more than anything, else, comes down to cold hard cash.
Can OpenAI remain a going concern without Microsoft?
OpenAI commands a soaring valuation thanks to its tender offers with venture capital firms, but this is just paper money based on share sales from existing employees, founders, and earlier investors. (Technically, these were profit participation agreements; OpenAI, owing to its highly unusual capped-profit structure, doesn’t offer traditional equity or stock options.) Khosla Ventures and billionaire Reid Hoffman’s charitable foundation put undisclosed amounts into OpenAI’s holding company initially. But only Microsoft has given significant cash and resources directly to OpenAI. The software giant has committed at least $13 billion to OpenAI since 2019, when it signed its first partnership agreement with the AI startup. Microsoft has delivered just a fraction of this cash to OpenAI so far, , giving the software giant significant power over the startup.
It is unclear if OpenAI could continue as a going concern without continual cash inflows from Microsoft. While OpenAI is, according to reports, making about $80 million per month currently and may be on track to make $1 billion in revenue in 2023—ten times more than it anticipated when it secured an additional $10 billion funding commitment from Microsoft in January—it is not known if the company is profitable or what its burn rate it is. But it is likely to be fast. The company lost $540 million dollars in 2022 on revenue of less than $30 million for the entire year, according to documents seen by Fortune. If its costs have also ramped up in line with revenues, the company would need continual support from Microsoft just to keep operating.
To make matters worse, OpenAI has been on a hiring spree all year, with reports that it has been offering the sort of typical of professional athletes to lure top AI researchers away from Google, Meta, and other rival firms. Ilya Sutskever, the OpenAI co-founder and chief scientist, was paid $1.9 million in 2016, the first year of OpenAI’s operations, according to tax filings. He is certainly making many times that now. GPU costs are also likely stratospheric. Training GPT-4, OpenAI’s latest model, cost more than $100 million, according to Altman. One back of the envelope calculation put the cost of running ChatGPT, OpenAI’s popular chatbot, back in April. It may be more than that now as its user base has grown. (Although Altman has said that the company has been able to optimize its models to bring the cost of serving each ChatGPT query down.)Furthermore, OpenAI is entirely dependent on Microsoft’s cloud computing datacenters to both train and run its models. The global shortage of graphic processing units (GPUs), the specialized computer chips needed to train and run large AI models, and the size of OpenAI’s business, with tens of millions of paying customers dependent on those models, mean that the San Francisco AI company cannot easily port its business to another cloud service provider.
Microsoft has what OpenAI needsAll of this gives Microsoft power. Nadella could threaten to cut off OpenAI’s access to computing power and suspend delivery of its next instalment of committed cash unless Altman is restored as CEO. While OpenAI might have legal recourse against Microsoft under the terms of its partnership agreement, Microsoft could do significant damage to OpenAI’s business before OpenAI’s lawyers could even get to the courthouse door if it wished. Longer-term, Microsoft could choose to replace OpenAI’s technology with software from another leading AI startup, such as Cohere or a newer player, like Mistral, or even the new AI company that Altman and OpenAI’s co-founder, president, and former board chairman, who resigned from the company in protest over Altman’s firing, have been contemplating setting up. Nadella has,, told Altman he will support him in whatever steps he decides to take next. OpenAI’s other inverstors, which include Khosla Ventures, billionaire Reid Hoffman’s charitable foundation, Tiger Global, Andreesen Horowitz, Sequoia Capital, Thrive, and K2 Global have far less potential influence on the company. The same goes for the venture firms that were negotiating to invest in a new tender of existing shares that. But they still have some leverage—mostly through OpenAI’s employees. Those employees want to be able to continue to offer to sell their profit participation shares (which function essentially as stock options since OpenAI is not thought to be currently profitable) to investors at high valuations that could make many of them extremely wealthy. With these investors threatening to pull out of the latest tender offer for those profit participation agreements, the employees will see their own financial prospects damaged. That may give many of them an incentive to leave OpenAI—or at least threaten to do so—unless Altman is reinstalled.
That corporate governance structureThe current mess at OpenAI ultimately shows how flawed OpenAI’s is. Microsoft is a minority investor in a limited liability corporation that is majority-owned by holding company that is in turn jointly owned by OpenAI’s non-profit board, its employees, and its other venture capital investors. But that holding company is controlled by OpenAI’s non-profit through another limited liability corporation. And OpenAI’s non-profit corporate board has just six seats, all of which must be held by people with no financial interest in the holding company or the LLC in which Microsoft invested. In other words, neither Microsoft, nor any of OpenAI’s other financial backers, have a seat at the table when it comes to the hiring and firing of OpenAI’s CEO. They have a right to any profits OpenAI earns until they earn back their initial investment plus an amount that is capped on a sliding scale, with the earliest investors able to earn up to 100 times their initial outlay and later investors entitled to smaller, but still substantial, pay-outs. After those thresholds have been hit, however, any additional profits revert to OpenAI’s non-profit. This structure was designed to enable OpenAI to raise the tens or even hundreds of billions of dollars it would need to succeed in its mission of building artificial general intelligence (AGI), the kind of AI that is as smart or smarter than people at most cognitive tasks, while at the same time preventing capitalist forces, and in particular a single big tech giant, from controlling AGI. Altman himself was largely responsible for designing this attempt to square a circle. But it was also Altman who struck the deal—for just $1 billion initially—with Nadella in 2019. From that moment on, the structure was basically a time bomb. By turning to a single corporate entity, Microsoft, for the majority of the cash and computing power OpenAI needed to achieve its mission, it was essentially handling control to Microsoft, even if that control wasn’t codified in any formal governance mechanism. Ironically, if OpenAI’s non-profit board chooses to ask Altman to return and resigns,, it will prove that Altman’s structure failed—OpenAI was not able to both raise billions of dollars from a big tech corporation while somehow remaining free from that corporation’s control.
Preventing a big tech company from monopolizing AGI had been the entire reason Altman, Brockman, Sutskever, and Elon Musk co-founded OpenAI in 2015. At the time, they were alarmed by the rapid progress in AI research being made by DeepMind, which Google had purchased for $600 million in 2014. If DeepMind succeeded in achieving AGI, the OpenAI co-founders feared the ultra-powerful technology would end up being monopolized by Google. They created OpenAI as a non-profit entity so that it could develop AGI “to benefit all humanity” and not just the shareholders of a single corporation. Elon Musk initially pledged $1 billion to the non-profit.The problem was, the pursuit of AGI using ever-larger deep learning models is extremely expensive due to the huge datacenter resources needed. Altman has said he vastly underestimated the amount of money OpenAI would need to compete in the race for AGI. And OpenAI’s lack of funding became an existential challenge after Musk acrimoniously resigned from OpenAI’s board after a dispute with Altman and the other co-founders over control of the research lab and its agenda. Musk took his $1 billion pledge with him. (Musk has said he donated $40 million to OpenAI during his the time he was affiliated with it. The company has said that in its time as a non-profit it received $130.5 million in total donations. Not nothing. But not enough to compete successfully with Google DeepMind.) With Musk’s backing gone, Altman had to find more money—fast. He realized that raising money from donors, who only get a tax write-off for their contributions, was extremely inefficient compared to venture capital, where investors expect a return. And a good way to get access to computing resources was to essentially cut a deal with a major cloud service provider that could give him both—cash plus compute. It would be the ultimate irony if the flaws of the very structure Altman designed wind up saving his job at CEO and allowing him to outmaneuver the board that he established to safeguard AGI. The mess at OpenAI will also have investors looking hard at the governance structure of its rival AI company Anthropic, which was started by researchers who broke away from OpenAI in 2021 because they were concerned that the commercial turn OpenAI had made following Microsoft’s investment jeopardized its AI safety mission. Anthorpic is B Corporation, which is a much more straight forward structure than OpenAI’s. A B Corporation is one in which the directors have a fiduciary duty to look after multiple stakeholders’ interests, not just the profits of shareholders. In Anthropic’s case, the board is supposed to look after the interests of society as well as the company. Venture investors in Anthropic have one seat on its board. But Anthropic also has a Long-term Benefit Trust, whose trustees are a panel of experts in AI safety and national security who have no financial interest in Anthropic. The Long-Term Benefit Trust has a special class of stock that allows it to install an increasing number of directors based on certain AI progress milestones. It also has the right to control the majority of the B Corp.’s board within four years. No doubt investors in Anthropic will now be giving this structure additional scrutiny following the destabilizing turmoil at OpenAI.
- Silicon Valley leaders are calling Sam Altman’s firing the biggest tech scandal since Apple fired Steve Jobs—but the leading theory about the drama at OpenAI tells a different story
- Who is Ilya Sutskever, the man at the center of OpenAI’s leadership shakeup—and why is he so worried about AI superintelligence going rogue?
- Mira Murati is suddenly in charge at OpenAI as interim CEO. Her cover interview with Fortune hints at how she’ll lead it
- The inside story of ChatGPT: How OpenAI founder Sam Altman built the world’s hottest technology with billions from Microsoft