AI & Enterprise
Nvidia unveils Groq 3 LPU inference chip for multi-agent workloads
Nvidia unveiled its Groq 3 LPU, a chip designed for AI inference, at its GTC 2026 developer conference, SiliconANGLE reported on March 17. Unlike conventional GPUs, the chip is optimized to run AI models and is designed to complement GPUs with ultra-fast memory and low-latency performance for multi-agent systems. Nvidia plans to deploy it in Groq 3 LPX racks and operate it with Vera Rubin NVL72 racks.
