Barret Zoph, Thinking Machines Co-Founder, Lands at Google as VP of Research

The AI executive carousel keeps spinning: Barret Zoph, fired from Thinking Machines and gone from OpenAI within five months, is back at Google — his old employer — to bring reinforcement learning and post-training expertise to Gemini.

Published: 2026-09-01 Category: Quick Take Sources: TechCrunch

A Revolving Door With a Twist

Barret Zoph, a co-founder of the AI startup Thinking Machines, has found yet another job: vice president of research at Google. The turn is notable because it is a return — Google is another company where Zoph previously worked. "We look forward to Barret returning to Google and bringing his RL and post-training expertise to Gemini," a Google spokesperson told the Wall Street Journal.

A Whirlwind Two Years

Zoph spent two years at OpenAI, left in October 2024 to co-found Thinking Machines with Mira Murati, and then in January dramatically departed the startup with fellow co-founder Luke Metz to return to OpenAI. It was later revealed that Zoph had been fired from Thinking Machines. That return didn't stick either: he spent only five months at OpenAI — where he was tasked with heading AI enterprise sales — before leaving in June. Now he is back at Google after just a couple of months on the sidelines.

The Talent Market Is a Game of Musical Chairs

It is not always easy to divine why tech executives are spending less time in their roles. The turnover rate in the AI industry is high, and it has been especially acute at OpenAI — a company that, despite readying itself for an IPO and being one of the most powerful presences in tech, has lost a lot of critical staff over the last eight months. From the departure of its COO to the recent loss of one of its top data center execs, the churn has left onlookers scratching their heads.

What This Says About the Industry

Zoph's path — OpenAI, Thinking Machines, back to OpenAI, now Google — is less a career arc than a portrait of the current AI labor market: a small pool of in-demand researchers circulating between the same handful of frontier labs, with few staying put long enough to build institutional depth. For Google, hiring someone with Zoph's RL and post-training pedigree is a direct bid to strengthen Gemini at the exact layer that matters most right now — making models reason and behave correctly after pretraining. For the industry, it is another reminder that in the race for talent, the labs are competing less for fresh bodies and more for a tiny number of proven names.

Based on reporting by Lucas Ropek at TechCrunch.