OpenAI has poached three founding members from Thinking Machines Lab—the AI startup led by former OpenAI CTO Mira Murati—in what appears to be a calculated strike against a rising competitor. Barret Zoph, Luke Metz, and Sam Schoenholz are all returning to OpenAI, with reports suggesting more researchers are in talks to follow.The move echoes Meta’s aggressive talent raids last year, when Mark Zuckerberg’s company offered compensation packages reportedly worth $200 million to lure AI researchers. OpenAI’s hiring spree has left Thinking Machines reeling, having lost half its cofounders in less than a year since launch.
Fidji Simo, OpenAI’s CEO of applications, announced the hirings on Wednesday, saying the moves had been “in the works for several weeks.” Zoph will report directly to her, while Metz and Schoenholz will work under him. At least two more Thinking Machines researchers—Lia Guy and Ian O’Connell—are reportedly leaving, with Guy heading to OpenAI.
The messy breakup that preceded the talent grab
The departure wasn’t clean. Murati told staff she had terminated Zoph’s employment over “unethical conduct,” though OpenAI disputed these concerns in an internal memo. According to reporting by Wired and the Wall Street Journal, Zoph had an undisclosed workplace relationship with another employee in a leadership role who has since left the company.Murati confronted Zoph about the relationship last summer, sources told Wired. Their working relationship deteriorated afterward, and Zoph began speaking with competitors including Meta before ultimately choosing OpenAI. In her internal message to employees, Murati cited a history of performance and conduct issues dating back to mid-2025.
When billion-dollar valuations can’t compete with cash compensation
The problem isn’t funding—it’s timing. Thinking Machines can dangle equity worth potentially billions, but that’s a long-term bet. Meanwhile, Google and Meta are writing checks that clear in months, not years. They’ve sweetened deals with accelerated vesting schedules that turn stock options into actual money almost immediately.OpenAI and Anthropic have another advantage: both have floated IPO plans within the next year or two. For employees, that’s a concrete exit timeline. Compare that to a startup that’s barely a year old, with no public product roadmap and uncertain path to liquidity. The choice becomes less about believing in the mission and more about when you can actually buy that house.Thinking Machines raised $2 billion in July at a $12 billion valuation—one of Silicon Valley’s largest seed rounds ever. The startup was even in talks to raise more at a $50 billion valuation as recently as November. But eye-popping valuations haven’t translated to employee retention. The startup released just one product—Tinker, an API for fine-tuning large language models—in October.For Murati, who played a role in Sam Altman’s brief 2023 ouster from OpenAI, the talent losses represent a harsh Silicon Valley irony—watching her former company systematically dismantle the team she built to compete against it.
About the Author
