- Published on
Report: Nvidia in Talks to Acquire or Further Invest in Reflection AI, Doubling Down on Open-Source Models
- Authors
- Name
- DeepHorus
According to the Financial Times, people familiar with the matter said Nvidia is in in-depth negotiations to acquire U.S. startup Reflection AI or make an additional investment in it. The chip giant is already one of Reflection's major shareholders, having previously injected $800 million into the company. The people added that the talks are still at an early stage and specific deal terms have not yet been clarified; both Nvidia and Reflection declined to comment.
Reflection AI focuses on developing open-weight models and on October 5 local time released Beam, its first open-weight AI model. In its previous funding round in March this year, the startup was already valued at $25 billion. Its technology route aligns closely with the direction of Nvidia's layout in the open-source field in recent years.
People familiar with the matter said the potential deal could take multiple forms, including a so-called 'acqui-hire' model. Under this structure, Nvidia would absorb its core team and obtain technology licensing, thereby bypassing the lengthy antitrust review that usually accompanies a full acquisition. In addition, Nvidia might also choose to further deepen its ties with the company through a new round of equity investment, or through a partnership plan that provides more chips and computing resources.
Multiple people familiar with the matter said the two sides are expected to finalize a plan within the coming weeks, but all parties also stressed that the talks could still fall apart. This means the final form and scale of the deal remain highly uncertain.
Facing regulators' strict antitrust scrutiny of M&A in the AI sector, 'acqui-hires' are increasingly becoming the preferred strategy of large tech companies. Last December, Nvidia adopted a similar structure in a $20 billion deal involving chip design company Groq. That deal this week just triggered a lawsuit by former Groq employees who were not absorbed, adding uncertainty to the prospects of this model.
Nvidia has launched an aggressive investment expansion over the past two years, aiming to boost the overall AI industry ecosystem and consolidate its dominant position at the core of the compute economy. Although it has already launched a family of open-weight AI models called Nemotron, Nvidia itself still lacks a top-tier frontier large model developed independently. In July this year, Nvidia CEO Jensen Huang led calls to establish a U.S.-based open-weight 'ecosystem,' stressing that this is crucial for 'sparking innovation and prosperity opportunities across the United States.'
In September this year, this tech giant with a market value of $5.5 trillion acquired Hugging Face, an open-source model and dataset hosting platform, for nearly $13 billion, showing its clear intention to continue increasing its bets in the open-source arena.
Nvidia has funneled tens of billions of dollars in total to leading large-model labs such as OpenAI, emerging GPU cloud providers such as Nebius, and many small and medium-sized teams. These investments in turn further boosted market demand for its AI chipsāfor any company building or running large models, these chips have always been the critical core commodity that determines the lifeline of computing power. If the deal with Reflection is ultimately finalized, Nvidia's layout in the field of open-weight models will be further strengthened.
