AI Infrastructure Essentials
This 2-3 hour Cloud Sparks session is a short, interactive introduction to the infrastructure behind AI on Google Cloud.
In AI Infrastructure Essentials, you will get a foundational overview of the hardware, software and networking needed to develop and manage AI models at scale. The session explores Google Cloud’s AI Hypercomputer architecture, compares compute accelerators such as GPUs and TPUs, and looks at the data pipelines and storage solutions that matter most for training performance.
You will learn to tell the three layers of the AI Hypercomputer apart, choose suitable accelerators for cost-effective AI workloads, evaluate storage and networking options to maximize training goodput, and compare deployment and consumption models for getting the most from your resources. The session includes an exercise and discussions along the way.
The session is aimed at IT decision-makers and infrastructure architects who want to understand the technical requirements and Google Cloud’s offerings for enterprise-grade AI deployment. Familiarity with cloud computing concepts and general data center infrastructure is expected.