Particle.news
Download on the App Store

Nvidia Licenses Groq’s Inference Tech, Hires Founder in Deal Seen as an Acquihire

The move positions Nvidia to pair its training GPUs with low-latency LPU hardware for real-time AI inference.

Overview

  • Groq confirmed a non-exclusive licensing agreement with Nvidia for its inference technology and said founder Jonathan Ross, president Sunny Madra, and other team members will join Nvidia.
  • Groq will continue operating independently with CFO Simon Edwards becoming CEO, and GroqCloud will remain available.
  • Multiple outlets report the deal is valued at about $20 billion, though Nvidia and Groq have not disclosed financial terms.
  • Analysts and reporters characterize the structure as an acquihire designed to bring key talent and IP to Nvidia while reducing immediate antitrust scrutiny.
  • The agreement targets low-latency inference, adding Groq’s LPU architecture, compiler, and software stack to Nvidia’s offerings to complement its leadership in model training.