Nvidia in Talks to Acquire or Deepen Investment in Open-Weight Model Startup Reflection AI

Nvidia is in early-stage discussions to acquire or deepen its investment in Reflection AI, a US developer of open-weight models. The talks could result in an acqui-hire, where Nvidia hires staff and licenses technology to avoid the regulatory scrutiny of a full acquisition, or involve further equity investment and chip supply agreements. A deal could be reached in the coming weeks, though discussions may still fall apart.

Strategic Context

Nvidia, the world’s most valuable company, has been aggressively investing in the AI ecosystem to bolster demand for its chips. While Nvidia has its own Nemotron open-weight model family, it lacks a frontier-scale model. Reflection AI is viewed as a strategic asset for the US in its AI competition with Chinese rivals like DeepSeek. The startup recently unveiled a customisable AI product running on Nvidia chips, claiming performance parity with Z.ai’s GLM-5.2.

Deal History and Valuation

• Valuation: Reflection AI was last valued at $25bn in a March funding round.
• Nvidia’s Stake: Nvidia has already invested $800mn in Reflection AI.
• Precedent: Nvidia previously used an acqui-hire structure for chip designer Groq in a $20bn deal in December.
• Recent Moves: In September, Nvidia agreed to acquire Hugging Face for $13bn.

Reflection AI’s founders, Misha Laskin and Ioannis Antonoglou, previously worked on Google’s Gemini model. The startup is backed by Sequoia Capital, Disruptive, and 1789 Capital (where Donald Trump Jr. is a partner) and has partnerships with the Pentagon and the Department of Energy.