On Jan. 12, Mira Murati, the chief executive and co-founder of the artificial intelligence start-up Thinking Machines Lab, was asked by Sam Schoenholz, a leading researcher at the company, to join a meeting.
Ms. Murati sat down with Mr. Schoenholz and two other founders, Barret Zoph and Luke Metz. All four had worked together at OpenAI, the ChatGPT maker, where Ms. Murati had been the chief technology officer, before the creation of Thinking Machines in February.
But in less than a year, Mr. Zoph, Mr. Metz and Mr. Schoenholz had become deeply unhappy with the start-up’s direction. Thinking Machines lagged OpenAI and other rivals in releasing products, and was struggling to raise new funding at an eye-popping $50 billion valuation. The men had urged Ms. Murati to strike a deal — Meta, the owner of Facebook and Instagram, had discussed buying Thinking Machines, and Ms. Murati had developed closer ties with the chief executive of Anthropic, a leading A.I. company — but no transaction had resulted.
At the meeting, the men lobbied for Mr. Zoph, Thinking Machines’ chief technology officer, to oversee decisions about the company’s technical direction. They threatened to leave if changes were not made, four people with knowledge of the discussion said.
Ms. Murati balked. She told Mr. Zoph that he was already the chief technology officer and should just do his job.
Two days later, Ms. Murati, 37, fired Mr. Zoph, 32, who had been talking behind her back since October with OpenAI’s chief executive, Sam Altman, about returning to the company. OpenAI immediately rehired Mr. Zoph, along with Mr. Metz and Mr. Schoenholz.
Another nine or so of Thinking Machines’ roughly 100 employees have decamped to OpenAI or have received offers to join the company, three people with knowledge of the situation said. Meta has also made offers worth hundreds of millions of dollars to some of the start-up’s employees, two people said.
The defections have set off furious mudslinging, as each side has cranked up its spin machine to portray its version of the story. A relationship that Mr. Zoph had with a Thinking Machines employee became public, and Ms. Murati — who posted a public message about having “parted ways” with Mr. Zoph — has faced questions over her reasons for firing him and the state of the company.
The turmoil at Thinking Machines, one of the highest-profile A.I. start-ups, is the latest episode in the perpetual soap opera that is the artificial intelligence industry, where young billionaires clash over professional and personal relationships as they vie for money, power and supremacy in the field. Occasionally, the drama spills into public view, riveting Silicon Valley.
“I am reminded of this line from ‘Anna Karenina’: ‘All happy families are alike; each unhappy family is unhappy in its own way,’” said Venky Ganesan, a partner at the investment firm Menlo Ventures. “Happy companies require many things to go right simultaneously. You only need one or two things to go wrong to have an unhappy company.”
Wired and The Wall Street Journal earlier reported some details of the exits of Mr. Zoph, Mr. Metz and Mr. Schoenholz from Thinking Machines.
The A.I. industry, a relatively small world of technology specialists and wealthy executives and investors, has long dealt with internecine feuds. Elon Musk, a co-founder of OpenAI, clashed with Mr. Altman over the company’s direction in 2018 and left. Mr. Musk has since sued Mr. Altman and OpenAI, accusing them of breaching the company’s founding contract by putting commercial interests ahead of the public good.
(The New York Times has sued OpenAI and Microsoft, claiming copyright infringement of news content related to A.I. systems. The two companies have denied the suit’s claims.)
In 2023, several OpenAI board members ousted Mr. Altman, saying they did not trust him to build A.I. for the benefit of humanity. Ms. Murati played a part in Mr. Altman’s firing and briefly took over as OpenAI’s chief executive.
When Mr. Altman and other powerful Silicon Valley figures resolved to reinstate him, Ms. Murati came out in support of him. She stayed at OpenAI for another year before leaving to start Thinking Machines with Mr. Zoph, Mr. Metz and others, including Andrew Tulloch, a former OpenAI researcher.
Her timing was fortuitous. The A.I. industry was booming, and Ms. Murati had the name recognition in tech circles to gather funding for a new venture. Thinking Machines amassed $2 billion from the venture capital firm Andreessen Horowitz, the chip makers Nvidia and AMD, and others.
The financing, announced in July, valued Thinking Machines at $12 billion, though it had not released a product and Ms. Murati was tight-lipped about what the start-up would produce. In a social media post at the time, she said the start-up “exists to empower humanity.”
The company opened an office in San Francisco’s Mission District, a few blocks from OpenAI’s headquarters, and spared no expense on interior decorating. The in-house gym included “squat racks” for weight lifting, with iron barbells branded with “Thinking Machines” in bold white typeface.
Last summer, Mr. Zoph approached Ms. Murati about his relationship with a colleague who was not an A.I. researcher and who eventually left the company, three people familiar with the matter said. Mr. Zoph continued in his role.
Around then, Meta executives reached out to Ms. Murati about potentially buying Thinking Machines, five people with knowledge of the conversations said. Mr. Zoph wanted to sell, but Ms. Murati preferred to remain independent and viewed the discussions with Meta as not serious.
In October, Thinking Machines released a product that helps developers adjust A.I. models to better perform specific tasks. Offered to businesses and software developers over the internet, the product was similar to technologies from Google, Amazon and Microsoft.
Mr. Tulloch also pushed Ms. Murati to sell Thinking Machines, two people familiar with the matter said. No deal materialized, and he left for Meta, which was heavily recruiting A.I. talent. The talks between Meta and Thinking Machines were reported earlier by The Verge.
In the ensuing months, Thinking Machines struggled to raise new funding, for which it was seeking a valuation of $50 billion. Some of its investors have privately worried about whether the start-up has a path forward as an independent company, two people familiar with the discussions said.
Ms. Murati also developed closer ties with Anthropic’s chief executive, Dario Amodei, who had also previously worked at OpenAI, five people said. While Ms. Murati and Mr. Amodei had clashed when they were at OpenAI, they are now in regular communication and have discussed their two companies and the A.I. industry. Ms. Murati has long been friends with Mr. Amodei’s wife.
For now, Ms. Murati appears to be content with Thinking Machines going it alone as an independent company.
On Wednesday, OpenAI announced that a new leader would take over selling A.I. services to businesses, according to internal communications relayed to The Times. The appointment, reported earlier by The Information, is a key position for overseeing an important area for future revenue growth.
That leader was Mr. Zoph.
Cade Metz is a Times reporter who writes about artificial intelligence, driverless cars, robotics, virtual reality and other emerging areas of technology.
The post The A.I. Start-Up Soap Opera Riveting Silicon Valley appeared first on New York Times.
