Nvidia is sharply escalating its bet on open-weight AI. According to the WSJ, the company will pay $6 billion to license technologies from Poolside and bring over 100 of its employees into the Nemotron team, plus invest another $1 billion directly in Poolside at a $12 billion pre-deal valuation.

The goal is to turn Nemotron into a model family that competes not only with DeepSeek and Kimi but also with closed models from OpenAI and Anthropic — a notable pivot for a company earning billions selling GPUs to the very labs whose closed models it now aims to rival.

Related: Nvidia Announces Nemotron 3 Ultra — Most Powerful Open Model from the U.S., Kimi K3 Could Be China’s DeepSeek Moment — Benchmarks Challenge US Frontier Models

WSJ: Nvidia Is Spending $6 Billion to Build a Powerful U.S. Alternative to Chinese AI