Nvidia is in advanced talks to acquire Inception, a two-year-old Palo Alto AI startup, for more than $500 million, according to The…
Nvidia is in advanced talks to acquire Inception, a two-year-old Palo Alto AI startup, for more than $500 million, according to The Information. No definitive agreement has been signed yet, so this is still a report of talks.
Inception builds the Mercury family of diffusion-based language models. Most chatbots generate text one token at a time, in sequence. Diffusion models draft and revise multiple parts of a response in parallel, which can cut wait time, especially on longer responses.
The company has more than 30 employees and raised a $50 million seed round in November 2025 led by Menlo Ventures, with M12, Snowflake Ventures, Databricks and others participating. Nvidia's own NVentures was already a backer, and Andrej Karpathy and Andrew Ng invested as angels.
Under the deal, the three Stanford co-founders, Stefano Ermon, Aditya Grover and Volodymyr Kuleshov, would join Nvidia full-time, with earnouts for employees tied to performance targets.
My read: Nvidia researchers introduced a Nemotron model in May that can switch between diffusion and conventional sequential generation. Buying Inception looks like bringing in the people who built the approach they were already testing.
The Information says the technology would support further Nemotron development, including possible use in AI systems for robots. Latency matters a lot more when the thing waiting on an answer is a machine moving in the physical world, and I'm curious whether that is where this ends up first.
$NVDA
Sources:
Tech Funding News: https://techfundingnews.com/report-nvidia-in-advanced-talks-to-acquire-stanford-led-diffusion-ai-startup-inception-for-500m/
TradingView/GuruFocus: https://www.tradingview.com/news/gurufocus:35e9c9608094b:0-nvidia-eyes-500-million-deal-that-could-change-how-its-ai-answers/
Not financial advice. For informational and educational purposes only.
