Meta’s loss is Thinking Machines’ gain.

Weiyao Wang spent eight years at Meta, his first job after college, helping build multimodal recognition systems and contributing to open segmentation projects, including SAM3D. My last day at Meta was last week and I have since joined the Thinking Machines Lab.

His move to Thinking Machines Lab (TML) comes as the AI ​​startup expands into multiple fields. TML just signed a multibillion-dollar cloud deal with Google, giving it access to Nvidia’s latest GB300 chip and making it one of the first startups to run on the hardware.

The deal, announced at Google Cloud Next on Tuesday, follows a previous partnership with Nvidia and places TML on the same infrastructure layer as Anthropic and Meta. (Meta is known to have been negotiating the acquisition of Thinking Machines around this time last year, and has recently been selecting TML founders one by one.)

The talent pool is still in flux. Wang and Kenneth Li, a Harvard PhD who spent 10 months at Meta before joining TML this month, are the latest example of talent acquisition going both ways. Business Insider reported last week that Meta had recruited seven of TML’s founding members. A review of recent hires shows that Thinking Machines are attacking Meta again. After reviewing LinkedIn profiles, it appears that TML has hired more researchers from Meta than any other single employer.

The most notable one is Soumith Chintala, CTO of TML. He spent 11 years at Meta and co-founded PyTorch, an open source deep learning framework that now powers most of the world’s AI research. He will leave Meta at the end of 2025 and was appointed CTO earlier this year. Piotr Dollár, another 11-year Meta veteran who served as Director of Research and co-authored the influential Segment Anything model, now serves on TML’s technical staff. Andrea Madotto, a research scientist specializing in multimodal language models in Meta’s FAIR department, joined TML in December. James Sun, a software engineer who has been teaching pre- and post-LLM training at Meta for about nine years, also took the leap.

TML has attracted talent from beyond Meta as well. Neal Wu, a three-time gold medalist at the International Information Olympiad and a founding member of popular coding startup Cognition, joined us earlier this year. Jeffrey Tao comes from Waymo, Windsurf, and OpenAI. Muhammad Maaz previously worked as a researcher at Anthropic. Erik Wijmans is from Apple. Before joining in March, Liliang Ren spent two and a half years on Microsoft’s AI Superintelligence team pre-training OpenAI models for code.

The startup currently has about 140 employees.

Meta’s salary package (no strings, 7 numbers) is now well known. For researchers considering other options, the calculation can be as simple as: Thinking Machines Lab is currently valued at $12 billion. While these numbers would have been unthinkable for the company at this stage in its previous technology cycle (it has only launched one product to date), there is still plenty of financial upside compared to the record valuations of OpenAI and Anthropic.

As of Friday morning, a TML spokesperson declined to comment for this story.