Thinking Machines Lab has hired Weiyao Wang, an 8-year Meta veteran who helped build multimodal perception systems including SAM3D, the latest in a steady stream of Meta researchers defecting to the $12 billion startup. Wang's final day at Meta was last week. His arrival caps a month in which Thinking Machines Lab also signed a multibillion-dollar Google Cloud deal and grew headcount to roughly 140.
The Google Cloud agreement, announced Tuesday at Google Cloud Next, gives TML access to Nvidia's GB300 chips and makes it one of the first startups running on the new hardware. The deal, which builds on an earlier Nvidia partnership, puts TML in the same infrastructure tier as Anthropic and Meta itself.
The talent flow runs in both directions, but it is increasingly lopsided. Business Insider reported last week that Meta has poached 7 of TML's founding members. A review of recent LinkedIn hires suggests TML has hired more researchers from Meta than from any other single employer, even as Meta reportedly held acquisition talks with the startup around this time last year.
Key facts
- 01Weiyao Wang left Meta after 8 years and joined Thinking Machines Lab last week.
- 02TML now counts roughly 140 employees and is valued at $12 billion despite shipping just one product.
- 03Business Insider reported Meta has poached 7 of TML's founding members; TML has hired more researchers from Meta than any other employer.
- 04TML signed a multibillion-dollar Google Cloud deal Tuesday for access to Nvidia's GB300 chips.
- 05Soumith Chintala, an 11-year Meta veteran and PyTorch co-creator, is now TML's CTO.
The most prominent defection is Soumith Chintala, who spent 11 years at Meta and co-founded PyTorch, the open source framework that underpins most of the world's AI research. Chintala left Meta in late 2025 and was appointed TML's CTO earlier this year. Piotr Dollár, another 11-year Meta veteran and co-author of the Segment Anything model, is now on TML's technical staff.
“Thinking Machines Lab is valued at $12 billion despite shipping just one product, and its 140-person headcount is increasingly stocked with Meta defectors including 11-year veteran Soumith Chintala.”— Jaeden Schafer
Andrea Madotto, a research scientist in Meta's FAIR division, joined TML in December. James Sun, a software engineer with nearly 9 years at Meta working on LLM pre- and post-training, also made the move. Kenneth Li, a Harvard PhD, spent only 10 months at Meta before jumping to TML this month.
TML has pulled talent from across the industry, not just Meta. Neal Wu, a three-time gold medalist at the International Olympiad in Informatics and a founding member of Cognition, joined early this year. Jeffrey Tao arrived via Waymo, Windsurf, and OpenAI. Muhammad Maaz previously held a research fellowship at Anthropic, Erik Wijmans came from Apple, and Liliang Ren joined in March after 2.5 years on Microsoft's AI Superintelligence team pre-training OpenAI models for code.
Meta's pitch to researchers is well documented: seven-figure pay packages with no strings attached. The counter-pitch from Thinking Machines Lab is equity in a company already worth $12 billion despite having shipped only one product. For researchers comparing offers, that valuation is small relative to OpenAI and Anthropic, which suggests room for the upside to compound.
The Google Cloud deal sharpens that calculus. As we covered earlier this week, Google's three-layer AI strategy now includes TML as an anchor compute customer for its TPU and GB300 capacity. Pairing premium hardware access with founder-tier equity is a combination Meta cannot easily replicate, even with cash.
The risk for TML is concentration. A 140-person company that has released a single product cannot afford to lose its most senior people to Meta's checkbook, and the seven founding-member departures show the door swings both ways. A spokesperson for TML declined to comment Friday morning.
The broader read is that the AI talent market has bifurcated. Meta is winning short-term with cash, while Thinking Machines Lab is winning the researchers who would rather own a slice of a $12 billion bet than cash a guaranteed check. With Nvidia's newest silicon now wired into its stack, TML has the compute to convert that talent into product. Whether it ships a second one fast enough to justify the valuation is the only question that matters.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
