Nvidia is paying $6 billion to license Poolside’s Model Factory — the training pipeline behind its Laguna coding models — plus another $1 billion investment at a $12 billion valuation. Job offers go out to 109 Poolside engineers. What’s missing: an acquisition. That’s the point.
What Nvidia actually bought
Poolside, founded in 2023 by ex-GitHub CTO Jason Warner and Eiso Kant, builds open-weight coding models. Laguna S 2.1, released in July, is a 118B-parameter MoE with a 1M-token context that beats models 10x its size on agentic coding. The $6B license covers the factory that produces them — and it’s non-exclusive, so Poolside can sell the same tech again. The founders stay, the company keeps running.
The acqui-license playbook
License fee plus investment plus hiring the core team gets Nvidia everything an acquisition would — without an antitrust review. This is becoming 2026’s default deal structure for big-tech AI shopping. The bigger signal: Nvidia sells chips, and it just bought its way into the AI coding software layer sitting on top of them. Poolside was worth $3 billion in late 2024. Quadrupled in under two years, without ever being sold.
You Might Also Like
- Laguna s 2 1 Poolside 118b Params 8b Active Trained From Scratch in Nine Weeks
- Cloudrouter Gives Your ai Coding Agent its own Cloud Machine and Thats a big Deal
- Code Arena Finally Gives Developers a Fair way to Judge ai Coding Models
- Ricursive Intelligence Just Raised 335m in 4 Months and i Think Nvidia Should be Paying Attention
- Nvidia Personaplex Just Made Every Voice ai Pipeline Feel Obsolete
Top AI Product 
Leave a comment