Groq Founder Joins Nvidia in Inference Tech Pact

AI chip startup Groq has confirmed a non-exclusive licensing agreement with Nvidia for its high-performance AI inference technology, clarifying reports that had suggested a full acquisition. Groq is a U.S.-based artificial intelligence company, with headquarters in Mountain View and San Jose, California.

Groq investor Alex Davis of Disruptive as saying Nvidia is acquiring Groq for $20 billion. While Groq did not confirm an acquisition, it stated that the agreement involves licensing its inference IP, with key leadership transitioning to Nvidia.

As part of the deal, Groq founder and CEO Jonathan Ross, President Sunny Madra, and other senior team members will join Nvidia to help scale the licensed technology. Groq will continue to operate as an independent company, with Simon Edwards appointed as the new CEO, and GroqCloud will remain fully operational.

Groq’s core innovation is its Language Processing Unit (LPU)—a processor purpose-built for AI inference. Unlike traditional accelerators, the LPU integrates large on-chip SRAM as primary weight storage, dramatically reducing latency and enabling efficient, scalable inference across chips.

The structure of the deal mirrors recent industry moves, such as Meta Platforms’ arrangement with Scale AI, where talent and technology were absorbed without a full acquisition. Regardless of terminology, the outcome is clear: Nvidia gains access to one of the most promising inference architectures in the market—marking a significant step in the ongoing consolidation of the AI chip landscape.