Mira Murati’s Thinking Machines Lab loses two co-founders to OpenAI in major leadership shake-up

New Delhi: Former OpenAI executive Mira Murati’s AI startup, Thinking Machines Lab, is facing an early leadership shake-up, with two of its co-founders exiting the company and returning to OpenAI. The departures come less than a year after the high-profile startup was launched, raising fresh questions about internal stability at one of Silicon Valley’s most closely watched AI ventures.

Murati confirmed on Wednesday that co-founder and chief technology officer Barret Zoph has left the company. In a post on X, she announced that Soumith Chintala will take over as CTO. Murati praised Chintala as a seasoned AI leader and said the company was excited about the transition, but did not address reports of additional departures at the time.

Co-founders head back to OpenAI

Less than an hour after Murati’s announcement, Fidji Simo, CEO of applications at OpenAI, said that Zoph would be rejoining the company. She also revealed that Luke Metz, another Thinking Machines co-founder, and Sam Schoenholz would also be returning to OpenAI. All three previously held technical roles at OpenAI before leaving to help launch Murati’s startup.

Zoph earlier served as OpenAI’s vice president of research and spent several years at Google as a research scientist. Metz and Schoenholz also worked at OpenAI for extended periods, contributing to core research teams before joining Thinking Machines.

A high-profile startup under pressure

In September 2024, Murati split off and co-founded Thinking Machines with Zoph and Metz. Since then, big players have shown interest in the startup and the company has raised a seed round of 2 billion dollars, led by Andreessen Horowitz. Accel, Nvidia, AMD and Jane Street were other investors and were worth billions of dollars.

Nonetheless, the exits are a sign of two co-founders simultaneously including the original CTO of the company. The separation with Zoph is reportedly not a friendly one, which contributed to the issues of intra-company tension. Andrew Tulloch, the other co-founder who joined Meta in October, had already departed the company.

Although switches in the leadership of leading AI companies are typical, several high-level departures at such an early stage of a startup are not the norm. Thinking Machines had established a positive reputation by hiring ex-OpenAI, Meta, and Mistral AI researchers. The recent developments may put the brakes on the momentum since the company is still setting its technical and product course.

The case of OpenAI, also, has not been an exception, as its founders and top managers have also left to start or join other competing projects. In 2024, it was notable that John Schulman departed OpenAI and briefly worked at Anthropic, and then became chief scientist at Thinking Machines. The recent reshuffle highlights the increasingly competitive nature of the increasingly fluid AI talent race.