Private AI chatbot hosting with NVIDIA GPU provides the raw parallel compute power needed for low latency inference