Meta has recruited five founders from Thinking Machines Lab, including an engineer reportedly valued at $1.5 billion.

      In brief: Meta has acquired five of the founding members of Thinking Machines Lab, the AI startup established by former OpenAI CTO Mira Murati, after she turned down an alleged $1 billion buyout offer. The highest-paid hire, co-founder Andrew Tulloch, reportedly received a compensation package of $1.5 billion over six years. This talent acquisition forms part of a larger restructuring within Meta's AI division, which includes a $14.3 billion investment in Scale AI, the appointment of Alexandr Wang as chief AI officer, Yann LeCun's departure after 12 years, cuts to 600 FAIR research positions, and the launch of Meta Superintelligence Labs, with its first closed-source model, Muse Spark, debuting on April 8.

      Meta has successfully recruited five founding members from Thinking Machines Lab, the AI startup founded by Mira Murati after her tenure as OpenAI's chief technology officer. The latest defection, founding engineer Joshua Gross, transitioned to Meta Superintelligence Labs in March after developing Tinker, the startup's main API product. His move follows that of co-founder Andrew Tulloch, who joined Meta in October with a reported compensation worth $1.5 billion over six years, which, if accurate, would mark it as the most expensive individual hire in the technology sector.

      The strategy is clear. Following Mark Zuckerberg's approximately $1 billion offer to buy Thinking Machines Lab, which was declined, Meta shifted to recruiting the founding members individually. Various sources have referred to this tactic as a "full-scale raid." This approach has proven successful; of the startup's original founding team, five have joined Meta, three returned to OpenAI, and one moved to Elon Musk’s xAI. Murati’s company, which raised $2 billion at a $12 billion valuation in a seed round led by Andreessen Horowitz in July 2025 and was reportedly negotiating for a new round at a $50 billion valuation by November, has seen the majority of its original team depart.

      Departures and their new roles

      Tulloch and Gross are the two departures with clearer trajectories. Tulloch, an AI researcher with a background at OpenAI, has joined Meta Superintelligence Labs, working under Alexandr Wang, the former Scale AI CEO appointed as Meta's first chief AI officer in June 2025. Gross, who previously worked at both OpenAI and Meta, now oversees engineering teams within the same division.

      The other exits from Thinking Machines took different directions. Barret Zoph and Luke Metz rejoined OpenAI in January 2026, along with Sam Schoenholz. Zoph was reportedly dismissed by Murati for "unethical conduct" before rejoining OpenAI immediately. Devendra Chaplot left to join xAI in March. As a result of these departures, Murati now has a significantly altered leadership team: she remains as CEO, Soumith Chintala, the PyTorch creator who joined from Meta’s FAIR lab, serves as CTO, while John Schulman continues as chief scientist.

      Meta's new AI structure

      These talent acquisitions are part of a larger restructuring that has transformed Meta's AI organization throughout the past year. In June 2025, Meta invested $14.3 billion for a 49% non-voting stake in Scale AI and appointed Wang to lead a new division, Meta Superintelligence Labs, alongside Nat Friedman, the former GitHub CEO. Zuckerberg remarked in an internal memo that Wang is "the most impressive founder of his generation."

      However, the restructuring has faced challenges. Yann LeCun, Meta’s chief AI scientist for 12 years and a pivotal figure in deep learning, departed in November 2025 after being instructed to report to Wang. LeCun expressed in an interview with the Financial Times in January that “you don’t tell a researcher what to do,” particularly someone of his stature, and labeled Wang as “young and inexperienced,” predicting further departures. Subsequently, LeCun raised $1 billion to establish AMI Labs in Paris, primarily comprising talent from Meta’s AI research division.

      By August 2025, Meta Superintelligence Labs was divided into four teams: the TBD Lab focused on large language models led by Wang; FAIR for fundamental research; a products and applied research division directed by Friedman; and an infrastructure unit managed by Aparna Ramani. The AGI Foundations team, previously responsible for the Llama model family, was disbanded after Llama 4 received lukewarm feedback. LeCun publicly stated that the AI team had "fudged" some results. In October 2025, approximately 600 roles were eliminated from FAIR and AI infrastructure units.

      The financial dynamics of talent acquisition

      Compensation figures in the AI talent market have escalated to a degree that skews normal recruitment practices. OpenAI’s CEO, Sam Altman, acknowledged that signing bonuses could reach up to $100 million to attract top talent. OpenAI’s chief