Groq

Groq is an inference infrastructure provider positioned as "the premier neocloud for fast inference." The company developed proprietary LPU (Language Processing Unit) technology designed specifically for language model inference, claiming performance advantages over general-purpose GPU solutions. Groq recently announced expansion to LPX systems working alongside NVIDIA's next-generation GPUs to "deliver unparalleled inference capability." The platform's positioning centres on solving the inference bottleneck in AI applications. Groq argues that as AI workloads grow more complex, inference becomes the performance constraint, and their LPU architecture directly addresses this. The company claims users can achieve both speed and affordability without choosing between them—typically viewed as a trade-off in the industry. Groq has raised $650 million in funding and is "building hundreds of megawatts of capacity" for inference workloads. The platform serves customers through their console interface at console.groq.com, though specific pricing, model selection, and tier details are not published publicly. The company appears to operate on a sales-driven model with custom pricing based on usage volume and workload requirements.