Thinking Machines talent shift

Meta has pulled at least seven founding members out of Thinking Machines Lab, even as Mira Murati’s startup keeps hiring senior Meta researchers back the other way. (techcrunch.com) TechCrunch reported on April 24 that Meta had poached seven founding members from Thinking Machines, citing earlier reporting from Business Insider and a review of recent hires. The same review found Thinking Machines has been recruiting from Meta more than from any other single employer. (techcrunch.com) Among the most prominent arrivals at Thinking Machines are Chief Technology Officer Soumith Chintala, a Meta veteran who co-founded PyTorch, and Piotr Dollár, a former Meta research director who co-authored Segment Anything. Andrea Madotto and James Sun also moved from Meta to Thinking Machines, according to TechCrunch. (techcrunch.com) The departures have gone the other way, too. TechCrunch previously reported in October 2025 that Thinking Machines co-founder Andrew Tulloch left for Meta, and the outlet says the talent picture has stayed fluid into April 2026. (techcrunch.com 1) (techcrunch.com 2) That hiring fight is happening alongside a separate race for computing power, the specialized hardware and data-center capacity needed to train large artificial intelligence models. On March 10, Nvidia and Thinking Machines said they had signed a multiyear strategic partnership centered on that infrastructure. (thinkingmachines.ai) (nvidia.com) The deal gives Thinking Machines access to at least one gigawatt of next-generation Nvidia Vera Rubin systems, with deployment targeted for early 2027. Nvidia also said it made a significant investment in the startup as part of the arrangement. (thinkingmachines.ai) (nvidia.com) Vera Rubin is Nvidia’s next major artificial intelligence platform, built as a package of chips, networking and storage for giant model-training clusters. Nvidia said on March 16 that the platform was in full production with seven chips aimed at scaling what it calls AI factories. (nvidia.com 1) (nvidia.com 2) Thinking Machines has also expanded beyond Nvidia. TechCrunch reported on April 22 that the company signed a multibillion-dollar agreement to expand its use of Google Cloud infrastructure, including systems powered by Nvidia’s latest graphics processors. (techcrunch.com) Murati launched Thinking Machines in 2025 after leaving OpenAI, and the company has pitched itself as an artificial intelligence research and product lab focused on customizable systems. Its homepage says it is building tools so people can make AI work for their own needs and goals. (techcrunch.com) (thinkingmachines.ai) The result is a two-front contest: Meta is still trying to assemble research teams, while Thinking Machines is pairing its recruiting with reserved access to future Nvidia capacity and outside cloud contracts. (techcrunch.com) (thinkingmachines.ai) (techcrunch.com)