Google DeepMind is bringing veteran artificial intelligence researcher Barret Zoph back into the Google fold, adding another twist to one of the most closely watched talent battles in the global AI industry.
Zoph, a co-founder and former chief technology officer of Mira Murati’s Thinking Machines Lab, said he is joining Google DeepMind as vice president of research, according to Reuters. The appointment puts him inside Alphabet’s flagship AI organization at a moment when Google is reshaping DeepMind’s leadership and competing aggressively with OpenAI, Anthropic and a growing field of heavily funded AI startups.
But for Zoph, the move is also something of a homecoming.
From Google Brain to OpenAI — and Back to Google
Before becoming one of the better-known names in the current generation of AI labs, Zoph spent roughly six years as a research scientist at Google Brain, the influential research organization that Google later combined with DeepMind in 2023 to create Google DeepMind.
His research at Google included work on neural architecture search, an approach that uses machine learning to help design neural networks automatically.
A 2017 paper by Zoph and researcher Quoc V. Le demonstrated how reinforcement learning could be used to generate high-performing neural network architectures. Zoph later contributed to work on NASNet, automated optimizer discovery, mixture-of-experts language models and research examining the “emergent abilities” that can appear as large language models scale.
That background makes his return significant beyond the executive title: Google is bringing back a researcher with deep familiarity with both its own AI research culture and several of its biggest competitors.
Zoph Became a Key OpenAI Research Leader
Zoph joined OpenAI after leaving Google and eventually served as a vice president of research, with a major focus on post-training — the crucial work that takes a pretrained AI model and improves how it follows instructions, reasons and behaves when deployed to users.
He left OpenAI in late 2024 as several senior figures departed the ChatGPT maker, including former CTO Mira Murati.
Murati subsequently launched Thinking Machines Lab in February 2025, assembling a team filled with veterans of OpenAI and other leading AI organizations.
Zoph became a co-founder and the startup’s CTO. Fortune reported at the time that Thinking Machines was established around a mission of making advanced AI systems more understandable, customizable and broadly capable.
The startup quickly became one of Silicon Valley’s most closely watched AI companies.
TechCrunch reported that Thinking Machines raised a massive $2 billion seed round at a $12 billion valuation, an extraordinary level of backing for such a young company.
A Short-Lived Return to OpenAI
Zoph’s tenure at Thinking Machines ended in January 2026.
The circumstances surrounding that split generated competing accounts at the time. WIRED reported that OpenAI subsequently brought Zoph and Thinking Machines co-founder Luke Metz back to the company, while noting that it could not independently verify allegations circulating about Zoph’s departure from the startup.
Zoph’s second period at OpenAI proved brief.
Fortune and The Verge reported that after returning in January, he took on responsibilities connected with OpenAI’s enterprise business before leaving the company again roughly five months later in June.
Now his next destination is confirmed: Google DeepMind.
Why the Timing Matters for Google DeepMind
The appointment lands during an unusually turbulent period in the competition for elite AI researchers.
Google announced a major DeepMind leadership restructuring earlier in August.
Demis Hassabis, the DeepMind co-founder and Nobel laureate who led the organization, transitioned from CEO to chairman of Google DeepMind and became Alphabet’s chief scientist. DeepMind CTO Koray Kavukcuoglu assumed day-to-day operational responsibilities while retaining his role as Alphabet’s chief AI architect.
The reorganization followed several high-profile departures.
Reuters reported that longtime Google figures Jeff Dean, Sanjay Ghemawat, Oriol Vinyals and Quoc Le left to create a new public-benefit AI company called Discovery Loop. Other prominent DeepMind researchers have also departed during 2026, adding to the intense movement of researchers among Google, OpenAI, Anthropic and newly created AI ventures.
Against that backdrop, recruiting Zoph gives DeepMind an experienced researcher with direct knowledge of Google Brain, OpenAI and one of the industry’s most prominent AI startups.
Google and Thinking Machines Already Had a Multibillion-Dollar Connection
There is another intriguing layer to the move.
Google and Thinking Machines Lab were already doing substantial business together.
In April, TechCrunch reported that Murati’s startup signed a multibillion-dollar agreement with Google Cloud to expand its access to AI infrastructure, including systems using Nvidia’s GB300 GPUs.
The agreement was reportedly worth several billion dollars and was designed to support model training, deployment and reinforcement-learning workloads.
Thinking Machines has continued developing its own AI technology. In July, it released Inkling, its first internally developed open-weight model, according to TechCrunch.
Zoph had already left Thinking Machines before those developments, but the cloud deal highlights how intertwined even rival AI companies have become: they may compete for researchers and customers while simultaneously relying on one another for computing infrastructure.
The Bigger Battle Is for People
Billions of dollars are pouring into GPUs, data centers and increasingly powerful AI models, but one of the industry’s scarcest resources remains human expertise.
A relatively small group of researchers has experience building frontier-scale models, designing training techniques and turning experimental systems into products used by millions of people.
That helps explain why the movement of a single prominent researcher can attract so much attention.
For Google, Zoph’s appointment is therefore more than another senior hire. It brings a former Google Brain scientist back into an organization facing one of the most consequential periods in its history.
For Zoph, the move completes an extraordinary professional loop:
Google Brain → OpenAI → Thinking Machines Lab → OpenAI → Google DeepMind.
And as the race toward increasingly capable AI intensifies, his next chapter may reveal just how aggressively Google intends to rebuild and reinforce the research bench behind Gemini and its broader artificial-intelligence ambitions.
Leave a Reply