Barret Zoph Joins Google After Thinking Machines Lab and OpenAI Stint

TL;DR

• Barret Zoph has joined Google, most recently Google DeepMind, after a brief return to OpenAI following his departure as co-founder and CTO of Mira Murati's Thinking Machines Lab.
• His exit after less than a year at Thinking Machines Lab marks a major early leadership shakeup for the high-profile startup, which launched in early 2025.
• The move underscores the escalating AI talent war, with Google, OpenAI, and Meta aggressively poaching top researchers with massive compensation packages and frontier model leadership opportunities.

From OpenAI Star to Startup CTO and Back Again

Barret Zoph's career has become a case study in just how fluid and fiercely competitive the top tier of AI talent has become. Zoph, widely recognized as one of the key architects behind ChatGPT and OpenAI's post-training efforts, is now headed to Google.

The hire caps a whirlwind year of moves for the researcher. Zoph left OpenAI in late 2024 to co-found Thinking Machines Lab with former OpenAI CTO Mira Murati, serving as its Chief Technology Officer. The startup, which officially emerged from stealth in February 2025, was instantly one of the most hyped new entrants in AI, attracting a roster of ex-OpenAI talent and a reported multibillion-dollar valuation before even unveiling a product.

Zoph's departure from Thinking Machines Lab was first reported in early August 2025, less than six months after the lab's public launch. At the time, his exit was seen as a surprising blow to the fledgling company, given his central technical role and close partnership with Murati.

A Brief Return to OpenAI

Rather than heading directly to a competitor, Zoph initially made a high-profile return to OpenAI. His comeback was viewed as a major win for CEO Sam Altman, bringing back a leader who had overseen critical advances in reinforcement learning from human feedback (RLHF), reasoning models, and the training pipelines that powered GPT-4o and subsequent models.

During his second stint at OpenAI, Zoph was reportedly involved in post-training and agentic model development as the company raced to ship its next-generation models. His return, however, proved to be short-lived, lasting roughly a year before the latest move to Google.

Why Google, and Why Now

Zoph is expected to join Google DeepMind, Google's flagship AI division led by Demis Hassabis, where he will work on frontier model development. While neither Google nor Zoph has publicly detailed his exact title or research mandate, the hire is consistent with Google's aggressive push over the past 12 months to consolidate top research talent as it accelerates work on its Gemini models.

For Google, landing Zoph is a significant coup. The company has been locked in an intense battle with OpenAI, Anthropic, and Meta to retain and recruit the small pool of researchers capable of leading frontier model training. Zoph brings not only deep expertise in model post-training and productization but also direct insight into the technical roadmaps of both OpenAI and Thinking Machines Lab.

What It Means for Thinking Machines Lab

For Thinking Machines Lab, Zoph's departure raises fresh questions about its long-term stability and technical direction. The startup was founded on the premise of building more understandable, customizable, and capable AI systems, with Murati assembling an all-star team including former OpenAI co-founder John Schulman and former VP of Research Barret Zoph.

Losing its CTO so early is an undeniable setback, though the company has continued to hire and has downplayed any single point of failure. Murati remains CEO, and the lab still retains a deep bench of researchers from OpenAI, Character AI, and Meta. The company has yet to release its first major model, and its ability to execute without one of its core technical founders will be closely watched by investors and competitors alike.

Inside the Escalating AI Talent War

Zoph's circuit from OpenAI to Thinking Machines Lab and back to OpenAI before landing at Google is emblematic of a broader industry reality in 2026: elite AI talent has never been more mobile, or more expensive.

Over the past year, compensation packages for top researchers have reportedly soared into the tens of millions annually, with Meta, Google, and OpenAI all offering unprecedented equity and compute resources to lure leaders. Meta's formation of its Superintelligence Labs and OpenAI's aggressive retention efforts have only intensified the bidding war.

Unlike previous tech talent cycles, these moves are not just about prestige. A single top researcher can influence the trajectory of a multibillion-dollar model program, making hires like Zoph strategic assets that directly shape product timelines, safety approaches, and competitive positioning.

As frontier labs race toward artificial general intelligence, the battle for the people who can actually build it is becoming just as important as the models themselves.

AndroGuider Team

Articles written by the AndroGuider team. We try to make them thorough and informational while being easy to read.