Google Cloud Run embraces Nvidia GPUs for serverless AI inference

August 21, 2024: Google Cloud Run Adds Nvidia GPUs for AI Inference - Google Cloud Run now supports Nvidia L4 GPUs, enabling serverless AI inference. This service aims to provide cost-effective, scalable GPU power only when needed, eliminating the need for always-on cloud instances or on-premises hardware. Developers can deploy real-time inference, generative AI models, and compute-intensive tasks with improved efficiency. Google assures minimal performance impact with cold start times between 11 to 35 seconds. Although model-agnostic, the platform recommends using models under 13B parameters for optimal performance.