Nvidia is dropping $6 billion to license the "Model Factory" AI platform from startup Poolside while moving 109 of its engineers onto the Nvidia payroll. According to an investor letter first reported by Newcomer, Jensen Huang's firm is pairing the licensing fee with a direct $1 billion equity check at a $12 billion pre-money valuation. The incoming cohort comprises the engineering backbone behind Poolside's Laguna model, leaving the startup's three founders behind to steer what remains of the corporate shell.

The maneuver underscores Nvidia's calculated transition from merchant chip supplier to full-stack platform. By ingesting Poolside's model-building harness, Nvidia significantly expands the automated training pipelines powering its own Nemotron family. That push places Nvidia in direct architectural competition with the very frontier lab customers buying its Blackwell systems, reinforcing its ambition to dominate enterprise foundation layers.

Predictably, the investor memo takes pains to declare that the deal is "neither an acquisition nor an acquihire." In reality, it is the latest masterclass in regulatory evasion: Big Tech cutting massive licensing cheques and raiding elite engineering rosters to swallow tech assets without triggering antitrust M&A scrutiny. Nvidia has already deployed variations of this playbook with Groq and Enfabrica. Poolside expects to return the $6 billion windfall to its investors by late next year, cementing yet another pseudo-merger that bypasses regulatory oversight while solidifying Nvidia's vertical grip on AI infrastructure.

NVIDIAAI InvestmentLarge Language ModelsAI RegulationPoolside