Reflection AI

Artificial Intelligence · Semiconductors · Open-Weight Models

Acquisition

Nvidia reportedly weighs acquiring Reflection AI, valued at $25B

The talks could let the chip giant pair its hardware dominance with a frontier-scale open-weight model it currently lacks.

• Nvidia is in early talks to acquire, acqui-hire, or deepen its existing $800 million investment in Reflection AI, the Financial Times reported, with a deal possibly landing within weeks.
• Reflection AI, founded in 2024 by former Google DeepMind researchers Misha Laskin and Ioannis Antonoglou, builds open-weight AI models for coding and agentic software tasks.
• Possible structures include an 'acqui-hire' where Nvidia hires staff and licenses technology to sidestep a lengthy antitrust review, a larger equity stake, or a deal supplying more chips.
• Reflection was last valued at $25 billion in a March funding round and has separately discussed raising $2.5 billion at that valuation, according to the Wall Street Journal.
• The talks follow Reflection's Oct. 5 launch of Beam, its first open-weight model, which it says matches China's GLM-5.2 on reasoning benchmarks while costing less to run.
• The move would echo Nvidia's $20 billion acqui-hire of chip designer Groq in December, a format Big Tech increasingly uses to avoid merger review — a deal that has since drawn a lawsuit from Groq employees excluded from it.
• Nvidia already dominates AI hardware; absorbing or licensing a frontier open-weight model would extend its leverage up the stack into the models that run on its chips.