Nvidia has secured a licensing agreement to use AI chip acceleration technology developed by hardware startup Groq, while also recruiting the company’s CEO and key executives, Groq confirmed on Wednesday. The CEO, previously a senior contributor to Google’s AI chip initiative under Alphabet, will now transition into a leadership role at Nvidia.
The partnership reflects a trend seen across the tech industry, where major companies seek access to breakthrough innovations and top-tier talent without fully acquiring emerging firms.
Groq focuses on AI inference, a computing process that enables pre-trained models to generate responses to user prompts. Although Nvidia continues to lead the market for training AI systems, the inference hardware segment has grown far more competitive. Rivals such as AMD, along with other chip startups including Cerebras Systems, have accelerated efforts to capture share in the space.
Groq said the licensing arrangement is non-exclusive, meaning its technology can still be used outside Nvidia’s ecosystem. The company also confirmed that founder Jonathan Ross—who previously played a central role in launching Google’s AI chip development program—will join Nvidia alongside Groq President Sunny Madra and senior members of its engineering division.
A source familiar with the agreement verified the licensing deal has been finalized. Financial terms have not been made public.
Earlier reports from CNBC suggested Nvidia might be pursuing a multi-billion dollar acquisition of Groq, but both companies declined to comment on those claims. In its official announcement, Groq stated it will continue operating independently under new CEO Simon Edwards, and its cloud services business will remain active.
Similar talent-focused licensing and recruitment deals have recently emerged across Big Tech. Microsoft brought senior AI leaders into its ranks through structured licensing agreements, Meta recruited AI industry founders through incentive-backed executive roles, and Amazon onboarded key AI founders without acquiring entire firms. Several of these arrangements have attracted regulatory attention, though none has been reversed so far.
Bernstein analyst Stacy Rasgon noted that antitrust concerns are the most significant regulatory risk tied to these agreements. He also said that structuring the Groq partnership as non-exclusive could help preserve competitive optics, even as leadership and engineering talent shifts to Nvidia.
Rasgon also highlighted Nvidia CEO Jensen Huang’s strong political relationships, describing the company’s government alignment as among the most influential within U.S. technology circles.
Industry Outlook: AI Inference Competition Intensifies
Groq more than doubled its company valuation last year, rising to $6.9 billion from $2.8 billion following a $750 million funding round in September. The startup’s chip design relies on SRAM (on-chip memory), avoiding the need for external high-bandwidth memory modules. This architecture supports faster chatbot response speeds and smoother model interactions, but may restrict the size of AI models it can serve.
Cerebras Systems, a direct competitor using a similar memory strategy, is reportedly preparing for a public listing next year. Both companies have also expanded into Middle Eastern markets through large AI hardware agreements.
In a keynote earlier this year, Jensen Huang emphasized that Nvidia plans to remain a long-term leader even as the AI chip market shifts from model training toward inference acceleration.