Nvidia is in talks to deepen its investment in Reflection AI or offer the US startup computing power — a move that could potentially lead to an acquisition offer, the Financial Times reported on Saturday, October 10, citing people with direct knowledge of the matter. Bloomberg confirmed the discussions on October 11, citing a person familiar with the matter.
Possible deal structures include an "acqui-hire," in which Nvidia would hire staff and license the technology rather than pursue a full acquisition; an agreement to provide the startup with more chips; or an increase in Nvidia's equity investment. The discussions are at an early stage, no terms have been reached, and a deal could still fall apart. Reflection declined to comment, and Nvidia did not immediately respond to a request for comment.
Reflection, founded in 2024 by former Google DeepMind researchers Misha Laskin and Ioannis Antonoglou, debuted its first open-weight model, Beam, earlier in October — pushing into coding and AI-agent tasks at a lower cost than many competitors. The company said Beam scores comparably on advanced reasoning benchmarks to GLM-5.2, the flagship open model from China's Z.AI.
Nvidia is already a major backer and strategic investor: it put $800 million into the startup, the FT reported. The Wall Street Journal previously reported Reflection had held talks to raise $2.5 billion at a $25 billion valuation, and CEO Laskin told CNBC in April the startup was raising at a $25 billion pre-money valuation. Reflection has also struck billion-dollar deals with SpaceX and cloud provider Nebius Group.
The talks underscore how open-weight models have become strategic ground: Nvidia wants a deeper foothold in the open ecosystem it sells chips into, and Reflection — with coding agents as its wedge — is one of the few Western startups positioned to challenge the Chinese open-model dominance of DeepSeek and Kimi.