Thinking Machines Lab Recruits Meta Talent as Cognition Eyes $25 Billion Valuation
- Sources
- 3
- Claims verified
- 3 of 4
What happened
Thinking Machines Lab (TML) has emerged as a major destination for senior AI talent, specifically recruiting heavily from Meta's research divisions. The startup, currently valued at $12 billion, has grown its headcount to approximately 140 employees and recently secured a multibillion-dollar cloud agreement with Google Cloud to access Nvidia's latest GB300 chips. This infrastructure deal places TML in a similar compute tier to industry leaders like Anthropic.
Simultaneously, the AI coding sector is seeing massive valuation spikes. Cognition AI is reportedly in early-stage discussions to raise hundreds of millions of dollars at a $25 billion valuation, more than doubling its previous mark. These developments underscore a broader trend of capital and talent flowing toward specialized AI startups that can secure both top-tier researchers and the massive compute resources required for next-generation model training.
Key points
- Thinking Machines Lab has aggressively recruited senior researchers from Meta, including Weiyao Wang and Soumith Chintala, bringing its total headcount to 140.1
- TML secured a multibillion-dollar cloud deal with Google for Nvidia GB300 chips and a separate gigawatt-scale processor supply agreement with Nvidia.13
- Cognition AI is in early talks to raise hundreds of millions of dollars at a $25 billion valuation, more than doubling its previous valuation.2
- Thinking Machines Lab is currently valued at $12 billion following a $2 billion seed round led by Andreessen Horowitz.13
Why it matters
The 'Compute-Talent' Flywheel: Startups are using massive infrastructure deals (like TML's Google/Nvidia partnerships) as a primary recruiting tool to lure researchers away from Big Tech who need high-end hardware for their work.
Valuation Divergence: The jump in Cognition's valuation to $25 billion suggests that 'AI for Software Engineering' is currently viewed as one of the most lucrative and defensible verticals in the AI ecosystem.
Meta's Open-Source Vulnerability: As Meta continues to lose key researchers to startups, its ability to maintain its lead in open-source AI (like Llama and SAM) may be challenged by more nimble, specialized labs.
Fact-check
- Thinking Machines Lab is hiring top talent from Meta and OpenAI.TechCrunch and Business Insider confirm multiple named hires from Meta and OpenAI joining Thinking Machines Lab, though talent flow is bidirectional.Verified
- Thinking Machines Lab has secured major infrastructure/cloud deals (Google/Nvidia) to scale compute.Reuters and TechCrunch document multi-year partnerships with Nvidia and Google Cloud for significant chip supply and infrastructure.Verified
- Cognition is in funding talks at a $2.5 billion valuation.Bloomberg and other outlets report the valuation in talks is approximately $25 billion, contradicting the $2.5 billion figure in the cluster.Disputed
- The combined cluster constitutes a concrete, dated news event that is material and newsworthy.The cluster describes specific, dated developments regarding personnel and funding reported by multiple news organizations in April 2026.Verified
Sources
- [1]TechCrunch (opens in a new tab)Meta's loss is Thinking Machines' gainThinking Machines Lab is aggressively hiring senior researchers from Meta while securing high-end Nvidia compute through a new Google Cloud deal.High factualityBias 0 from −6 (left) to +6 (right)
- [2]Reuters (opens in a new tab)AI startup Thinking Machines clinches capital, major chip supply deal with NvidiaThinking Machines Lab has secured a massive multi-year partnership with Nvidia for gigawatt-scale compute and next-generation processors.High factualityBias 0 from −6 (left) to +6 (right)
- [3]Bloomberg (opens in a new tab)AI Coding Firm Cognition in Funding Talks at $25 Billion ValueCognition AI is seeking to raise hundreds of millions in new funding that would value the coding startup at $25 billion.High factualityBias 0 from −6 (left) to +6 (right)
L–R: source bias
