According to the news, Nvidia is negotiating to acquire/invest in Reflection AI, adding to the open source model消息称英伟达洽谈收购 / 追加投资 Reflection AI，加码开源模型

IT Home reported on October 11, according to the Financial Times, people familiar with the matter said that Nvidia is in advanced negotiations to acquire or invest in Reflection AI, a US startup. Reflection AI focuses on the development of open weight models, releasing its first open weight AI model, Beam, on October 5, local time. Negotiations are at an early stage and potential deals could take many forms, including the so-called “talent merger” model, people familiar with the matter added. Under this structure, Nvidia will absorb its core team and acquire technology licenses, thereby circumventing the lengthy antitrust reviews that typically accompany full-scale mergers and acquisitions. Nvidia is currently one of Reflection's major shareholders, and may choose to further deepen its ties with the company through a new round of equity investment or a partnership that provides more chips and computing resources. Nvidia has previously injected $800 million into Reflection (IT House note: the current exchange rate is about RMB 5.365 billion). According to several people familiar with the matter, the two sides are expected to finalize the plan in the coming weeks, but the parties also stressed that there is still a possibility of a breakdown in the negotiations. The terms of the deal are not yet clear, but Reflection was valued at $25 billion in its last round of funding in March (the current exchange rate is around RMB167.662 billion). Both Nvidia and Reflection declined to comment. In the face of the strict anti-monopoly scrutiny of AI mergers and acquisitions by regulators, "talent mergers" are increasingly becoming the preferred strategy for large technology companies. In December, Nvidia adopted a similar structure in a $20 billion deal involving chip design firm Groq, which just this week triggered legal action from former Groq employees who weren't absorbed. Nvidia has embarked on an aggressive investment expansion over the past two years aimed at boosting the overall AI industry ecosystem and consolidating its dominance at the heart of the computing economy. Although Nvidia has launched a family of open-weighted AI models called Nemotron, it still lacks its own cutting-edge, cutting-edge models. In July, Nvidia CEO Jen-Hsun Huang spearheaded a call for an open-weight “ecosystem” in the United States, emphasizing that it is essential to “spark innovation and prosperity opportunities across the country.” In September this year, the $5.5 trillion (current exchange rate of about 36.89 trillion yuan) technology giant added Hugging Face, an open-source model and dataset hosting platform, to its portfolio for nearly $13 billion (current exchange rate of about 87.184 billion yuan). Nvidia has funneled tens of billions of dollars to big-model labs such as OpenAI, emerging GPU cloud vendors such as Nebius, and numerous small and medium-sized teams. These investments have further pushed up the demand for their AI chips, which remain a key core material for any business that builds or runs large models. Related Reading: “Calling DeepSeek, Kimi, Nvidia Invested Reflection AI Releases Its First Open Weighted AI Model Beam”