Groq has declared that it has entered into a non-exclusive licensing pact with NVIDIA concerning inference technology. Both entities are set on collaboratively broadening the reach of high-performance yet cost-effective inference technologies. As per the deal, Groq's founder along with team members will integrate into NVIDIA to aid in its technological advancements and widespread implementation. Meanwhile, Groq will persist in its independent operations, with its GroqCloud service remaining untouched by this partnership.

Earlier, whispers in the industry suggested that NVIDIA might snap up Groq for a staggering $20 billion. However, NVIDIA swiftly refuted these rumors, clarifying that the two companies have merely inked a technology licensing agreement.

Established in 2016, Groq's pride lies in its self-developed LPU inference chip, which stands as the linchpin of this collaboration. This chip is meticulously tailored for AI inference scenarios, boasting substantial benefits when executing mainstream large language models. For NVIDIA, this alliance serves to plug any loopholes in its AI inference prowess, solidify its full-stack technological edge, and diminish potential acquisition hazards, thereby equipping it to tackle market competition more effectively.