Groq is an American artificial intelligence company, founded in 2016, that builds custom silicon purpose-built for AI inference. Its core product is the Language Processing Unit (LPU), a chip designed from the ground up for inference workloads, delivering deterministic, low-latency performance at a fraction of the cost compared to general-purpose GPUs.
The company operates GroqCloud, a cloud platform that provides developers with fast, scalable access to openly available AI models such as Llama, GPT-OSS, Kimi K2, Qwen, and Whisper. The platform serves over 3 million developers and teams. For regulated industries requiring on-premise deployment, Groq also offers GroqRack.
Groq's technical scope spans custom silicon design, large language models, speech-to-text, text-to-speech, and computer vision. The company has raised $750 million to scale its inference capacity and entered into a $20 billion strategic agreement with NVIDIA in 2025.
