In a dramatic turn within the fiercely competitive artificial intelligence industry, OpenAI under CEO Sam Altman has successfully reclaimed several top researchers from Mira Murati’s startup, Thinking Machines Lab (TML). The talent exodus, which gained momentum in early 2026, underscores the intense rivalries, sky-high compensation battles, and shifting loyalties that define today’s AI landscape.
Mira Murati’s Ambitious Exit and Startup Launch
Mira Murati, who served as OpenAI’s Chief Technology Officer and played a pivotal role in the development and launch of ChatGPT, departed the company in late 2024. Her exit followed a tumultuous period that included her brief stint as interim CEO during the 2023 board crisis that temporarily ousted Sam Altman.
In early 2025, Murati founded Thinking Machines Lab with significant backing. The startup secured a massive $2 billion seed round, achieving a valuation of approximately $12 billion. It quickly attracted elite talent from OpenAI, Meta, Mistral, and other leading organizations. Key early hires included former OpenAI figures such as Barret Zoph (who became co-founder and CTO), Luke Metz, and several prominent researchers. The venture positioned itself as a bold new player, launching initiatives like the Tinker API for fine-tuning large language models and drawing investments from heavyweights including a16z, Nvidia, and AMD.
However, underlying tensions from OpenAI’s past leadership struggles lingered. Reports suggest Murati provided information, including Slack screenshots, that contributed to Altman’s short-lived 2023 removal, and she later testified in related legal proceedings, including Elon Musk’s lawsuit against OpenAI, regarding Altman’s management style.
The 2026 Talent Reclamation Wave
Starting in January 2026, a notable wave of departures hit Thinking Machines Lab. In a public announcement on X, Murati confirmed the departure of co-founder and CTO Barret Zoph. She promoted Soumith Chintala as the new CTO. Shortly afterward, Zoph, along with co-founder Luke Metz and researcher Sam Schoenholz, returned to OpenAI. OpenAI’s Fidji Simo, CEO of Applications, announced the high-profile homecoming.
The returns did not stop there. Additional researchers, including Trapit Bansal, Hongyu Ren, Shengjia Zhao, Jolene Parish (who had joined TML in April 2025 after previous roles at OpenAI and Apple), and Lia Guy, also made their way back to OpenAI or moved to other competitors like Meta. Reports indicate that OpenAI poached at least nine employees from TML in this period.
By February 2026, further exits were reported, highlighting growing instability at the young startup despite its strong financial foundation.
A Personal and Corporate Rivalry
The developments have been widely framed as a direct contest between Sam Altman and Mira Murati. Media coverage and industry analysts have dubbed it a high-profile “spat,” with OpenAI appearing to regain momentum by reclaiming its former talent. Some accounts point to internal challenges at TML as factors in the departures, though specifics remain limited.
This “boomerang” talent movement reflects broader realities in the AI sector: astronomical salaries and signing bonuses often reaching millions of dollars, rapid formation (and sometimes rapid turbulence) of well-funded startups, and the constant poaching wars among tech giants and newcomers alike. OpenAI continues to demonstrate its magnetic pull for top researchers, even as competitors like Meta actively raid talent pools.
What It Means for the Industry
While Thinking Machines Lab launched with considerable promise and resources, the recent talent losses represent a significant setback in its early journey. For OpenAI, the returns reinforce its position as the premier destination for elite AI minds amid accelerating competition in generative AI and beyond.
The situation remains fluid, with new developments possible as both organizations navigate the relentless pace of AI advancement. As the industry evolves, such talent battles are likely to intensify, shaping the future direction of innovation in artificial intelligence.
About The Author
