Cloud platforms are where most data engineering now runs. Rather than buying servers, teams rent storage, compute, and managed data services from a provider and pay for what they use. Each major cloud offers the same building blocks under different names - object storage for the lake, a managed warehouse, a Spark/Hadoop service, streaming ingestion, and orchestration - so the patterns transfer even when the product names don't.

The practical mapping across providers is worth memorizing: object storage is S3 (AWS), ADLS / Blob (Azure), GCS (Google); the managed warehouse is Redshift, Synapse / Fabric, and BigQuery; managed Spark is EMR, HDInsight / Databricks, and Dataproc; streaming is Kinesis / MSK, Event Hubs, and Pub/Sub. Once you know one cloud's data stack, the others are largely a translation exercise.

The choice of cloud is usually dictated by where the rest of the organization already is - identity, networking, billing, and existing data gravity matter more than feature-by-feature comparison. The pages here cover the providers and the specific services data engineers reach for on each.