Altman (left) appearing onstage with Nadella at OpenAI’s first developer conference in San Francisco earlier this month. Microsoft snapped up Altman for a new venture after his sudden departure from OpenAI shocked the artificial intelligence world. — AP
JUST a few weeks ago, Microsoft CEO Satya Nadella stood onstage next to OpenAI’s Sam Altman at the startup’s conference in a former concert hall in San Francisco. Both wore black jeans, Altman in an army-green shirt and Nadella in navy casual.
“We love you guys!” Nadella said as he turned to Altman.
“Awwww,” Altman responded.
Altman has called OpenAI’s relationship with Microsoft “the best bromance in tech”. Since 2019, the companies have worked together to build advanced artificial intelligence systems that they believe could be the most important tech innovations in a generation, and Microsoft has invested US$13bil in OpenAI. Together, they planned to take on Google’s hammerlock on the Internet.
That relationship was put to the test. On Nov 17, when the non-profit board that controls OpenAI ousted Altman, the company’s co-founder and CEO, Microsoft was given just minutes of warning before the move was made public.
Over the next few days, Nadella made it clear that he wasn’t about to walk away from the partnership. And what could have been an embarrassing moment for him and his company turned into a display of corporate muscle flexing that stunned industry insiders.
Since OpenAI released its ChatGPT chatbot almost a year ago, AI has captured the public’s imagination, with hopes that it could be used for important work like drug research or to help teach children. It could also lead to job losses or even autonomous warfare.
And whoever builds it could control what some computer scientists believe is one of the most important new technologies since the steam engine.
Hours after OpenAI’s board of directors said it stood by the decision to push out Altman, Microsoft swooped in to hire Altman and Greg Brockman, who quit as OpenAI’s president after the board’s decision. Nadella said the two would run a new AI research lab for Microsoft, and most of OpenAI’s more than 700 employees have said they will walk out and offer their services to Microsoft.
“We look forward to moving quickly to provide them with the resources needed for their success,” Nadella said on X, formerly Twitter.
Microsoft and OpenAI declined to comment.
Nadella’s aggressive move was a stunning capstone to a wild weekend. It exposed a fault line between tech industry leaders focused on turning AI into a giant business and an increasingly influential part of the tech community that believes AI could be dangerous.
A key member of OpenAI’s board believed Altman was moving too fast to expand his company while not paying enough attention to the safety of AI, from concerns it will eliminate jobs to a belief it could be a threat to humanity.
Though Nadella and his company unsuccessfully tried to help resolve OpenAI’s management breakdown, he had more leverage over the San Francisco startup than many people realised.
OpenAI has most likely seen only a part of the US$13bil that Microsoft has committed because it was supposed to be paid out over time – though the exact terms of the deal were unclear. In addition, Microsoft signed a deal that gave it copies of OpenAI’s most cutting-edge technology and has been working with it for more than a year. Microsoft has been providing OpenAI with the enormous computer power it needs to build its AI.
With all that, Nadella could rebuild OpenAI inside Microsoft and not lose much time or money. It was also not out of the question that OpenAI’s board would give in to employee pressure to bring Altman and his allies back, with significant changes to the board. Nadella could live with that, too.
“It’s like you forget the nonsense that happened for four days – Sam is still Sam, and he is running the show,” said S. Somasegar, a former Microsoft executive now at Madrona Venture Group who has been in touch with Nadella. “Microsoft will end up the winner no matter what happens here.”
For Microsoft, an implosion at OpenAI presented a big risk to its plans to embed AI into everything it does. Microsoft owns 49% of OpenAI but has no direct influence over its board of directors.
As a hedge against not having explicit control of OpenAI, Microsoft negotiated contracts that gave it rights to OpenAI’s intellectual property, copies of the source code for its key systems as well as the “weights” that guide the system’s results after it has been trained on data, according to three people familiar with the deal, who were not allowed to publicly discuss it.
“That is the core protection for Microsoft,” Somasegar said.
Nadella moved quickly to speak with OpenAI’s board in an attempt to calm the charged situation. He said Microsoft would continue working with OpenAI.
In an interview on Bloomberg TV, Nadella said he’d had conversations with Emmett Shear, who had been named OpenAI’s second interim CEO. “My message to Emmett was very clear, which is, ‘Hey, look, we remain very, very committed to OpenAI and its mission and its road map, and they can count on us,’” Nadella said.
He also reaffirmed his commitment to working with Altman.
“We want to partner with OpenAI, and we want to partner with Sam,” Nadella said.
“Irrespective of where Sam is, he is working with Microsoft. That was the case on Friday, and that’s the case today, and I absolutely believe that will be the case tomorrow.”
Nadella said that in discussing Altman’s ouster with the OpenAI board, he had not been told “about any issues,” and so he remained confident in Altman. He added that Microsoft would push to make governance changes at OpenAI, though he did not say what those would be.
“Surprises are bad,” Nadella said.
Microsoft investors, who feared Microsoft was put in a tough spot by the management mess at OpenAI, applauded Nadella’s move. Microsoft’s stock price rose more than 2% to a record high.
Nadella and his technology chief, Kevin Scott, had close relationships with Altman and Brockman. Nadella and Altman have known each other since 2018, when they met at the high-wattage Allen & Co conference in Sun Valley, Idaho.
At the time, OpenAI was a non-profit research lab dedicated to building safe artificial general intelligence.
But OpenAI needed vast amounts of expensive computing power, so to attract investors, it created a for-profit company still under the control of the non-profit board of directors.
Since its first US$1bil investment in OpenAI in 2019, Microsoft has treated the far smaller company as a tech incubator. OpenAI was singularly focused on AI – like a pack of wolves, as one former Microsoft executive described it – while Microsoft had to manage an array of businesses, from cloud computing and software to computer games.
OpenAI is discovering it needed Microsoft far more than Microsoft needed OpenAI.
Microsoft developed and provided the vast computing power that runs OpenAI, and negotiated a slate of legal and commercial deals to protect it if something went wrong there.
Microsoft had spent months negotiating a US$10bil investment that closed in January, and worked to keep its ownership stake just under 50%.
Among other things, it worried that having majority control would expose it to antitrust scrutiny, according to the three people familiar with the deal. And Nadella avoided meddling in OpenAI’s management.
The chaotic weekend showed he did not need a seat on the board to have power. — ©2023 The New York Times Company