Nvidia has agreed to purchase assets from artificial intelligence (AI) chip startup Groq in a deal valued at nearly $20 billion, marking the largest transaction in Nvidia’s history.
Groq, established in 2016 by former Google engineers including TPU co-creator Jonathan Ross, particularly designs next-generation AI accelerator chips focused on low-latency inference.
In September, the company raised a funding round at a valuation of $6.9 billion that included investors such as BlackRock, Samsung, Cisco, and Neuberger Berman.
Groq described the agreement as a non-exclusive licensing deal for its inference technology instead of a full acquisition, as per CNBC.
Under the deal, Ross, company president Sunny Madra, and other leading players of the industry will join Nvidia, while Groq will continue its operations as an independent company under new CEO Simon Edwards.
GorqCloud, its cloud business, is not included in the transaction and will continue running separately.
Nvidia CEO Jensen Huang stated the agreement will allow Nvidia to incorporate Groq’s low-latency chipsets into its AI factory architecture.
He highlighted that Nvidia is licensing Groq’s intellectual property and hiring talent, not purchasing the company itself.
The transaction dwarfs Nvidia’s previous largest purchase, the $7 billion Mellanox agreement in 2019, and reflects the company’s increasing cash reserves that stood at $60.6 billion as of October.
This significant move underscores Nvidia’s push to strengthen its AI ecosystem amid increasing competition in the AI chip market.
