How tensions over profitability at OpenAI turned into a full-blown crisis
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Natalie Peeples/Axios
When the nonprofit OpenAI started selling shares in a potentially world-changing for-profit subsidiary, the tensions between the two were clear to see.
Why it matters: Those tensions erupted over the weekend into a full-scale governance crisis, as the board of the governing nonprofit fired the CEO, Sam Altman, over what Axios' Scott Rosenberg has described as "a deep cultural rift" within the organization.
State of play: OpenAI is now on its third CEO in as many days, while Altman has joined Microsoft.
- The new OpenAI CEO, Emmett Shear, has written that he's "not crazy enough to take this job without board support for commercializing our awesome models."
The big picture: It's relatively common for a nonprofit to use the profits of a corporate subsidiary to fund its operations. U.S. examples include Patagonia, Bloomberg, and Newman's Own; international examples would include Bertelsmann or Novo Nordisk.
Be smart: OpenAI diverges from that model. The nonprofit doesn't want to spend the (nonexistent) profits of the for-profit subsidiary; instead, it wants the subsidiary itself to research and develop safe AI.
- The nonprofit's goals are research-focused. They militate in favor of releasing fewer products, and making sure that any such products are "public goods."
- The for-profit, by contrast, is much more product-focused and provides privileged access for investors like Microsoft. It was created to be able to afford to continue to pay the astonishing costs of the non-profit's research agenda.
Where it stands: OpenAI has four separate power centers: The board of directors; the senior management, led until Friday by ousted CEO Altman; the employees; and the outside investors, led by Microsoft. None of their interests are entirely aligned.
Driving the news: Substantially all OpenAI employees work for the for-profit subsidiary and were looking forward to being able to sell some of their shares at an $86 billion valuation.
- Meanwhile, Altman, who weirdly owns no shares in OpenAI, had been looking to raise billions of dollars from Saudi Arabia and other investors for a separate AI-focused for-profit that would run independently of OpenAI, Bloomberg reported.
Between the lines: As far back as 2020, OpenAI started losing employees who worried that it was becoming too commercial under the leadership of Altman and his colleague Greg Brockman.
- The remaining employees have been largely supportive of Altman. As OpenAI's CEO-for-a-weekend Mira Murati posted after her own replacement, "OpenAI is nothing without its people."
Zoom out: For OpenAI, the stakes could hardly be higher. The organization was founded to save the world from the existential risk posed by profit-hungry technologists — but at this point, capitalist imperatives are so deeply embedded in the organization that they're probably impossible to expunge.
The bottom line: Under the terms of OpenAI's governing charter, its "primary fiduciary duty is to humanity." Humanity as a whole, however, is the one stakeholder that has zero ability to influence the outcome of the current fustercluck.
Go deeper: How a Silicon Valley nonprofit became worth billions
