Nvidia
Nvidia in advanced talks to buy diffusion-LLM startup Inception
Inception's diffusion models could strengthen Nvidia's open-source Nemotron line, according to the report.
The answer
Nvidia is reportedly near a deal for Inception, at $500 million or more.
What happened: The Information reported on 19–20 November that Nvidia is in advanced talks to acquire Inception, a startup building diffusion-based large language models. The reported price is at least $500 million, with performance-linked earnout awards for employees possible. (OODA Loop, citing The Information)
The numbers: Inception is about two years old, is based in Palo Alto and has more than 30 employees. One source says its seed round was $50 million, led by Menlo Ventures, with NVentures, Nvidia's venture arm, among the participants. The reported price is about ten times that seed.
The context: Stefano Ermon, a Stanford professor, is CEO. Aditya Grover (UCLA) and Volodymyr Kuleshov (Cornell) are also named as co-founders. Talks reportedly began before Jensen Huang stepped down as Nvidia CEO.
Why it matters: The reported rationale is to use Inception's diffusion technology to strengthen Nvidia's open-source Nemotron models. Some reports link the deal to robotics and physical AI. Inception claims its models run up to 10x faster than comparable models, a claim the company makes itself.
The rivals: Microsoft was also said to have shown interest in buying Inception.
The catch: Neither Nvidia nor Inception has confirmed the talks. A digest from The Neuron reported that Inception's API shuts down on 28 December. That has not been verified with Inception.
What's next: The price, the closing timeline and whether Mercury stays available as a standalone product are all unresolved.
Sources
- Nvidia nears deal to acquire diffusion LLM startup Inception for over $500 millionOODA Loop (citing The Information) · 20 November 2026
Free · no sign-up · no spam
Never miss a FeaturedDaily story
Add our free feed to any news reader and every new story lands there the moment we publish it.