Groq Inc

Groq is an AI infrastructure company focused on inference — the execution of trained AI models to serve users, products and agent tasks — rather than model training. The company designs and manufactures the LPU (Language Processing Unit), a chip purpose-built for AI inference, and operates GroqCloud, a platform for running inference workloads with low latency and cost efficiency across global data centers.

The company positions itself as an inference-focused cloud provider ("neocloud") and states that it pioneered the LPU. It has since introduced LPX, which the company describes as working alongside NVIDIA's next-generation GPUs to deliver inference capability at scale. Groq frames inference as an emerging capacity bottleneck as AI usage grows, and says it is building hundreds of megawatts of capacity with additional capacity planned.

Business model

Groq combines proprietary inference silicon with a cloud platform (GroqCloud) on which customers run inference workloads, operated out of data centers with capacity measured in megawatts.

Traction

Groq states it is building hundreds of megawatts of inference capacity with more planned, and announced the close of a $350 million Series A.

Latest developments

Groq announced the close of a $350 million Series A to build out its AI inference cloud, and introduced LPX as a complement to its LPU and to NVIDIA's next-generation GPUs.