Groq

Groq delivers ultra-fast AI inference using its custom-built Language Processing Unit (LPU) to accelerate Large Language Models (LLMs) at scale.

Groq specializes in high-speed AI inference, leveraging its proprietary Language Processing Unit (LPU) Inference Engine: a chip specifically architected for generative AI and LLMs. The LPU's unique dataflow architecture bypasses the memory and compute bottlenecks of traditional GPUs, delivering consistent, ultra-low-latency performance and superior energy efficiency. This technology, accessible via the GroqCloud platform or on-premise GroqRack clusters, enables real-time application deployment for demanding enterprise customers. Founded in 2016 by former Google engineers (including a lead designer of the TPU), Groq is setting the new standard for real-time AI compute.

https://groq.com