Singapore – Nvidia has entered into a non-exclusive licensing agreement with Groq to access the startup’s AI inference technology, as the US chipmaker continues to broaden its capabilities in serving large-scale and real-time AI workloads.
Alexander Davis, CEO of Disruptive, a firm that has invested more than half a billion US dollars in Groq, shared that the agreement between Nvidia and Groq is estimated at around US$20b.
The arrangement will see Groq founder Jonathan Ross, president Sunny Madra, and other members of the company’s engineering team join Nvidia to assist with the development and scaling of the licensed technology. The agreement does not involve the acquisition of Groq, which will continue to operate independently.
Groq has appointed Simon Edwards as its chief executive officer, while its cloud-based platform, GroqCloud, will continue to operate without interruption.
The licensing agreement reflects Nvidia’s wider strategy to reinforce its position in AI inference, an increasingly competitive segment of the semiconductor market as demand shifts beyond model training towards deployment and real-time execution.
An internal email from Jensen Huang, CEO of Nvidia, obtained by CNBC, highlighted how the company plans to integrate Groq’s low-latency processors into the Nvidia AI factory architecture to serve an even broader range of AI inference and real-time workloads.
Nvidia has previously pursued similar technology and talent-focused arrangements with emerging AI hardware companies, stopping short of full acquisitions. The company’s previous large investment took place in 2019, following the acquisition of Israeli chip designer Mellanox for US$7b.
Founded in 2016 by former Google engineers, Groq develops low-latency AI accelerator technology designed to improve inference performance for large-scale artificial intelligence applications.
