Groq is an AI hardware and cloud-inference company that designs custom Language Processing Unit (LPU) chips and operates an API for running open-source large language models at ultra-low latency. Developers rely on Groq’s platform to build cost-efficient RAG backends and other AI applications that demand fast, inexpensive inference.