Mira Murati’s Thinking Machines Talent Drain: Co-founders Return to OpenAI
Mira Murati’s Thinking Machines faces a major talent drain as CTO Barret Zoph and co-founder Luke Metz return to OpenAI. Analysis of the $12B startup's crisis.
They’re shaking hands, but it’s clear the gloves are off. Mira Murati’s high-profile AI startup, Thinking Machines Lab, is facing a significant leadership crisis just months after its explosive debut. Two of its co-founders and a key technical staffer are deserting the $12 billion venture to return to their former employer: OpenAI.
On Wednesday, January 14, 2026, Murati announced that CTO Barret Zoph had parted ways with the company. While she quickly named Soumith Chintala as the new CTO, the narrative took a sharp turn when OpenAI’s Fidji Simo confirmed just 58 minutes later that Zoph, along with co-founder Luke Metz and Sam Schoenholz, were officially back on the OpenAI roster.
Thinking Machines Talent Drain: A Massive Setback
The departure of two co-founders in a single sweep is a massive blow for a company that closed a $2 billion seed round only six months ago. Led by Andreessen Horowitz with participation from Nvidia and AMD, the funding was predicated on the strength of its founding team—many of whom are now exiting. Reports from Wired suggest the split wasn't amicable, hinting at deep-seated friction within the leadership.
This isn't the first time Thinking Machines has lost talent. In October, co-founder Andrew Tulloch left for Meta. While the startup still boasts John Schulman as Chief Scientist, the rapid churn of founding members raises questions about the long-term stability of OpenAI spin-offs in an increasingly volatile talent market.
This content is AI-generated based on source articles. While we strive for accuracy, errors may occur. We recommend verifying with the original source.
Related Articles
Pentagon cancels Anthropic's $200M contract over military AI control disputes, chooses OpenAI instead. ChatGPT uninstalls surge 295% as ethical concerns mount.
OpenAI's GPT-5.4 can now control mouse and keyboard directly. Is this the end of office work as we know it?
OpenAI's GPT-5.4 introduces native computer control, marking a shift from AI creators to AI operators. What happens when machines start clicking our mice?
OpenAI's GPT-5.4 shifts AI competition from pure performance to token efficiency. With 1M context window and 33% fewer errors, what changes for businesses and developers?
Thoughts
Share your thoughts on this article
Sign in to join the conversation