Nvidia Spends $6 Billion to License AI Coding Startup Poolside's Technology and Recruit Its Employees
Nashnova编辑部
Nvidia will pay AI coding startup Poolside $6 billion to license its model-training system and has extended offers to 109 key employees — the largest deal yet in Nvidia's pattern of acquiring startup capabilities through licensing rather than outright acquisition.
What does $6 billion actually buy?
Nvidia is licensing Poolside's Model Factory — a complete pipeline for training AI coding models.
It has also extended job offers to 109 employees who worked on Poolside's coding model, Laguna.
This means → Nvidia is not buying a finished product. It is buying the capability to build models and the people who built them.
Why is Poolside willing to sell?
In a letter to investors, Poolside said the Nvidia hardware it needs to stay competitive in AI models now exceeds what it can independently obtain.
In plain terms = Poolside wants to keep training large models but cannot secure enough Nvidia chips on its own — partnering with Nvidia is the only path to the compute it needs.
The $6 billion will be distributed to investors by the end of next year. Nvidia will also invest an additional $1 billion at a $12 billion pre-deal valuation. All three co-founders will stay.
Is this an acquisition?
Poolside's letter explicitly states the deal is not an acquisition and not an "acquihire" — a transaction where a company is bought primarily for its talent.
In practice, however, with its core product licensed away and 109 key staff recruited, Poolside's room for independent operation has narrowed sharply.
This reflects a pattern: Nvidia is using deals that look nothing like acquisitions to absorb startup capabilities while avoiding the scrutiny and integration costs of traditional M&A.
Why does Nvidia keep using this playbook?
This is not the first time. Nvidia previously struck a $20 billion similar deal with AI inference chip designer Groq and a $900 million deal with networking hardware designer Enfabrica.
Nvidia is itself a major developer of open-source AI models, competing directly with many of its own customers — yet it is simultaneously paying these startups massive licensing fees.
This means → Nvidia's strategy is to trade licensing fees for technology assets, gaining capability while sidestepping antitrust risk. Why it is willing to pay this much, however, remains unclear.
Content is for reference only, not financial advice.