Nvidia is in advanced talks to acquire artificial intelligence startup Inception for at least $500 million, in a potential deal that would bring the company’s Mercury diffusion language models and three prominent AI researchers into its expanding model development business.
The proposed transaction also includes performance-based payments tied to targets for employees joining Nvidia. No agreement has been signed or publicly announced, and neither company has commented on the talks.
The two-year-old startup has developed a different approach to AI-powered text generation, promising faster responses and lower operating costs than conventional language models.
Inception’s Mercury Technology at the Centre of Nvidia Talks
Inception’s Mercury family of diffusion language models is a key attraction in the proposed acquisition.
Traditional language models, including those powering ChatGPT and Claude, generate text sequentially, producing one token at a time. Inception’s technology instead refines multiple tokens in parallel, allowing the system to work on several parts of a response simultaneously.
Inception claims its diffusion-based approach can be up to 10 times faster and significantly cheaper to operate.
The technology could strengthen Nvidia’s Nemotron family of open AI models. One person familiar with the negotiations told The Information that Inception’s models and research team could contribute to Nvidia’s development efforts.
An acquisition would give Nvidia access to both Mercury’s diffusion technology and the researchers responsible for building it.
Stanford, UCLA and Cornell Researchers Expected to Join Nvidia
The proposed acquisition could also bring Inception’s three co-founders into Nvidia.
Stefano Ermon, Inception’s CEO and a professor at Stanford University, is expected to join alongside Aditya Grover of the University of California, Los Angeles, and Volodymyr Kuleshov of Cornell University.
Their expected arrival adds a significant research component to the potential transaction, alongside Inception’s existing technology.
The startup currently employs more than 30 people and has raised approximately $50 million from investors, including Menlo Ventures, M12, Mayfield and Nvidia itself.
Nvidia’s existing investment in Inception means the chipmaker already has a financial stake in the startup it is now reportedly seeking to acquire.
Microsoft Also Showed Interest in Inception
Microsoft had also expressed interest in a similar deal involving Inception,.
The reported interest from both Nvidia and Microsoft highlights the appeal of Inception’s work on diffusion-based language models, particularly as AI companies explore ways to improve response speeds and reduce computing costs.
Inception’s Mercury technology offers an alternative to the sequential text-generation methods widely used in generative AI systems.
For Nvidia, the acquisition could provide an additional route to developing its Nemotron models while bringing Inception’s academic-linked founding team into the company.
Acquisition Talks Began Before Nvidia’s Leadership Transition
The negotiations were said to have started before Jensen Huang left his position as CEO at Nvidia.
The timing makes the potential acquisition part of Nvidia’s initiatives to develop their own AI models, apart from their existing role as providers of hardware for AI computing.
The deal is valued at the bare minimum of $500 million, which may go up based on the performance of employees.
This acquisition is contingent upon the results of the negotiations and has not been made into any form of binding deal.
Also Read: Yuzvendra Chahal Becomes First Indian to Take 400 T20 Wickets
















