Nvidia Joins Big Tech Deal Spree, Licensing Groq Technology and Recruiting Executives

Nvidia has struck a deal to license chip technology from AI startup Groq while also hiring its CEO, a former Alphabet veteran, according to a Groq blog post on Wednesday. The move reflects a growing trend in Big Tech, where leading companies acquire technology and talent from promising startups without fully purchasing the firms.

Groq specializes in inference, the stage where AI models that have already been trained respond to user queries. While Nvidia dominates AI model training, it faces stiffer competition in inference, with rivals like AMD and startups such as Groq and Cerebras Systems entering the fray. Under the deal, Nvidia gains a non-exclusive license to Groq’s technology, and key personnel including Groq founder Jonathan Ross, Groq President Sunny Madra, and other engineering team members will join Nvidia.

A source close to Nvidia confirmed the licensing arrangement. Financial terms were not disclosed by either company. CNBC previously reported that Nvidia had agreed to acquire Groq for $20 billion, though neither Nvidia nor Groq commented on that report. Groq stressed that it will continue to operate independently under CEO Simon Edwards, maintaining its cloud business operations.

This deal echoes similar recent transactions in the AI sector. Microsoft acquired top AI talent from a startup through a $650 million licensing deal, and Meta reportedly paid $15 billion to hire Scale AI’s CEO without acquiring the company outright. Amazon has also recruited founders from Adept AI, while Nvidia executed a similar arrangement earlier this year. Although regulators are scrutinizing such deals, none have yet been reversed.

Bernstein analyst Stacy Rasgon noted in a client briefing that:

“Antitrust is the primary risk here, though structuring the deal as a non-exclusive license may preserve the appearance of competition, even as Groq’s leadership and technical talent move to Nvidia.”

Nvidia CEO Jensen Huang, Rasgon added, maintains one of the strongest tech-industry relationships with the U.S. government.

Groq’s valuation more than doubled to $6.9 billion from $2.8 billion in August 2024, following a $750 million funding round in September. The company, along with other startups, avoids using external high-bandwidth memory chips, sidestepping the global memory supply crunch. Groq relies on on-chip SRAM, which accelerates interactions with AI models and chatbots but limits the maximum model size it can serve.

Groq’s primary competitor in this SRAM-based approach is Cerebras Systems, which Reuters reports may go public as early as next year. Both companies have signed significant deals in the Middle East.

Nvidia’s Huang emphasized in his 2025 keynote that the company aims to maintain its leadership as the AI market shifts from training to inference a sector increasingly shaped by licensing deals, strategic hires, and emerging startups like Groq.