Nvidia Pays $6B for Poolside's Model Factory and Its Team

Is it an acquisition if the company insists it isn't one? Nvidia is paying $6 billion to license Poolside's "Model Factory," the internal system the coding-model startup uses to train and evaluate models end to end. As part of the deal, 109 Poolside employees, the team behind the Laguna model, are receiving job offers from Nvidia. Poolside's three founders stay put, and Nvidia is adding a $1 billion investment at a $12 billion pre-money valuation. The story was first reported by Newcomer, based on a letter sent to investors.

Follow the money, not the label

Poolside says this is neither an acquisition nor an acqui-hire. Look at the mechanics, though: the $6 billion license fee gets distributed to Poolside's investors by the end of the year, and the people who built the technology move to Nvidia's payroll. Investors get paid, the team changes badges, and what remains of Poolside keeps its name and its founders. If that structure sounds familiar, it should. Microsoft used a version of it to absorb Inflection's team, and Google did the same with Windsurf's. Because no merger technically occurs, these deals tend to slide past the review process that a formal takeover would trigger. Regulators in the US have examined similar arrangements before; whether this one draws attention is an open question.

What Nvidia actually gets

The interesting asset here is the Model Factory itself: a production line for building custom models, covering data preparation, training runs, and evaluation. Nvidia already develops its own Nemotron model family, and owning a proven model-building pipeline lets it offer chip customers something beyond hardware, closer to "bring us your data, leave with your own model." There is friction in that ambition: some of those customers are model developers themselves, so Nvidia's move up the stack puts it in more direct competition with the people buying its GPUs. It also fits a spending pattern. Last week Nvidia agreed to backstop OpenAI's Ohio data center lease with up to $105 billion; this week it is buying its way deeper into the model layer.

The consolidation problem for everyone else

Independent model developers keep disappearing into the balance sheets of a few giants, and each exit shrinks the pool of suppliers with any incentive to compete on price or terms. For companies building on AI today, the practical defense is architectural: route your model calls through an abstraction layer so switching providers stays a configuration change rather than a rewrite. It costs a few days of engineering now. When the consolidation bill arrives as a price increase or a discontinued API, those few days become your leverage.

Sources: The Decoder