Barret Zoph has a new job. Again. The Thinking Machines co-founder, who was ousted from that startup before boomeranging back to OpenAI, has landed at Google as vice president of research, according to TechCrunch AI, which cited reporting from the Wall Street Journal.
If you’re losing track of Zoph’s moves, you’re not alone. His resume over the last two years reads like a game of musical chairs.
📍 The timeline
Here’s how fast this went, per TechCrunch AI:
- Zoph spent two years at OpenAI, then left in October 2024 to co-found Thinking Machines with Mira Murati, who had exited OpenAI the month before.
- In January, Zoph and fellow co-founder Luke Metz abruptly left Thinking Machines to rejoin OpenAI. It later came out that Zoph had been fired.
- The reunion didn’t last. He spent just five months at OpenAI running enterprise sales and left in June.
- Now he’s back at Google, a company he’d already worked at before, this time on the research side.
A Google spokesperson put it plainly to the Wall Street Journal: “We look forward to Barret returning to Google and bringing his RL and post-training expertise to Gemini.”
🔍 What the Google move actually signals
That quote is the part worth reading twice. RL means reinforcement learning, and post-training is the work that happens after a model’s initial training run: fine-tuning, alignment, and the reasoning tricks that separate a raw model from a useful one. It’s where a lot of the recent frontier gains have come from.
Google isn’t hiring Zoph for a sales seat. It’s putting him on Gemini’s core model quality. That’s a direct signal about where Google thinks the next round of competitive advantage lives, and it’s poaching that expertise straight out of OpenAI’s orbit.
💡 Why this matters beyond one executive
One person changing jobs four times isn’t the story. The pattern behind it is.
The AI industry’s turnover rate is high, and TechCrunch AI notes it’s been especially brutal at OpenAI. Over roughly the last eight months, the company has lost its COO, a top data center executive, and other critical staff, all while it readies itself for an IPO and holds one of the strongest positions in tech.
What stands out here is the contradiction. OpenAI is arguably the most powerful name in the field, yet senior people keep walking out the door. When a company that dominant can’t keep its bench, that’s a retention problem money alone clearly isn’t solving.
🧭 What to watch next
For practitioners and anyone tracking the frontier race, a few things to keep an eye on:
- Talent flow is a leading indicator. Where senior RL and post-training people land tells you which labs are betting hardest on reasoning models. Right now, some of that talent is flowing toward Google.
- OpenAI’s stability going into an IPO. Repeated high-level exits are the kind of thing investors ask about. Watch whether the departures slow or keep accelerating.
- Gemini’s roadmap. Zoph’s mandate is model quality. If Google’s post-training push shows up in future Gemini releases, his hire is part of that story.
The executive shuffle isn’t slowing down, and each move is a small tell about where the real leverage in AI is shifting. You can find the full reporting at the original source.