Nvidia nears $500 million deal for diffusion-model startup Inception

Nvidia is reportedly nearing a deal to acquire the AI startup Inception for at least $500 million, according to The Information. The purchase would strengthen Nvidia's open-source Nemotron models and its push into robotics.
Deal terms and the startup's workforce
According to three people familiar with the talks, as reported by The Information, Nvidia is close to buying Inception for at least $500 million. The agreement also includes performance bonuses for the more than 30 employees of the two-year-old startup, which is based in the San Francisco Bay Area.
The report rests on those sources, and the material available does not include confirmation of the details from the companies involved.
A different approach to language models
Inception builds language models using the diffusion process known from image generation. Rather than producing a reply token by token, as conventional models behind ChatGPT and Claude do, its models refine a complete response step by step. The startup most recently released Mercury 3.
Strengthening Nemotron and robotics
The acquisition is intended to bolster Nvidia's open-source Nemotron models and its physical models for robots. Nvidia is a semiconductor company based in Santa Clara. The move would also position it against OpenAI and Anthropic, which are both building their own inference chips.
Palantir and US government agencies already use Nemotron as an alternative to Chinese models, so an expanded Nemotron line would reach an existing customer base.
Open questions
- Will the deal be completed at the reported value of at least $500 million?
- How will Inception's diffusion approach be integrated into the Nemotron models?