Amidst announcing its most robust single-quarter financial performance to date, NVIDIA made a series of three major investment announcements. Firstly, it acquired the open-source AI model community Hugging Face for a staggering $12.9 billion. Secondly, it secured model licensing, employees, and additional investments from AI programming model startup Poolside, amounting to approximately $7 billion. Lastly, it negotiated a new round of financing for AI search engine application company Perplexity. All three investments are centered around the open-source domain. Previously, NVIDIA CEO Jensen Huang had publicly expressed his support for open-source large models. Now, the company is no longer content with merely being an AI chip supplier but is actively positioning itself within the open-source large model sector.

Since 2023, NVIDIA has embarked on investing in AI model layer enterprises. In 2024, it launched multiple open-source large model products, and by 2025, it introduced the fully self-trained flagship open-source model series, Nemotron 3. This year, NVIDIA has further expedited its strategic layout (or positioning) by acquiring multiple large model companies, with the aim of creating its own high-quality open-source large models.

This strategic shift is driven by changes in market demand. Closed-source giants in the U.S. large model sector are developing their own chips, while rumors have circulated under the Trump administration about restricting Chinese open-source models, putting pressure on NVIDIA's demand for computing power. Huang, along with 25 tech companies, once co-authored an open letter in support of open-source models. His rationale for positioning NVIDIA within the open-source large model ecosystem mirrors the company's CUDA layout (or positioning) two decades ago—creating controllable computing power demand by constructing a robust open-source model ecosystem, which, in turn, strengthens the chip business.

Although NVIDIA's open-source large model, Nemotron 3 Ultra, currently ranks only tenth in intelligence index rankings, trailing behind models such as Kimi and DeepSeek, NVIDIA's determination and investment should not be underestimated.