AWS vs Azure vs Google Cloud Comparison

How the three major clouds differ in strengths, networking models, identity, data and AI platforms, with a service mapping and a structured way to choose.

Use the interactive Cloud Service Comparison tool for the full mapping of 75 capabilities.

Core service mapping

Capability AWS Azure Google Cloud
VMs EC2 Virtual Machines Compute Engine
Kubernetes EKS AKS GKE
Serverless containers Fargate / App Runner Container Apps Cloud Run
Functions Lambda Functions Cloud Run functions
Object storage S3 Blob Storage Cloud Storage
Relational RDS / Aurora Azure SQL / Database for PostgreSQL Cloud SQL / AlloyDB
Global strongly consistent SQL Aurora DSQL (Cosmos DB, NoSQL) Spanner
NoSQL DynamoDB Cosmos DB Firestore / Bigtable
Warehouse Redshift Fabric / Synapse BigQuery
Messaging SQS / SNS / EventBridge Service Bus / Event Grid Pub/Sub / Eventarc
Streaming Kinesis / MSK Event Hubs Pub/Sub / Managed Kafka
API management API Gateway API Management Apigee
Identity IAM / Identity Center Entra ID Cloud IAM / Cloud Identity
Keys KMS Key Vault Cloud KMS
Monitoring CloudWatch Azure Monitor Cloud Monitoring
GenAI Bedrock Azure OpenAI / Foundry Vertex AI

Meaningful differences (not just names)

Topic AWS Azure Google Cloud
Network scope VPC is regional VNet is regional VPC is global, subnets regional
Resource organization Accounts in Organizations (account = isolation boundary) Management groups → subscriptions → resource groups Organization → folders → projects
Identity center of gravity IAM per account + Identity Center Entra ID (shared with Microsoft 365) Google identity + IAM inheritance
Load balancing Regional ALB/NLB + CloudFront/Global Accelerator Front Door (global) + App Gateway Global anycast L7 LB native
Signature strengths Breadth and maturity of services, largest ecosystem Enterprise/Microsoft integration, hybrid (Arc), Windows/SQL licensing Data and analytics (BigQuery), Spanner, Kubernetes heritage, AI/TPUs

How to choose (decision framework)

  1. Existing estate and skills: Microsoft-heavy enterprises often favour Azure; teams with deep AWS skills move faster on AWS.
  2. Workload fit: global consistent databases, analytics at scale, specific AI models or hardware.
  3. Commercials: enterprise discount commitments, licence portability (Windows/SQL on Azure).
  4. Regions and compliance: available regions, sovereign cloud offerings, certifications.
  5. Ecosystem and partners: marketplace products, SI partners, managed services.
  6. Exit and concentration risk: regulated industries need exit plans.

[!CTO] A primary cloud with selective use of others for specific strengths is usually more effective than spreading every workload across clouds. Multi-cloud multiplies skills, tooling and security effort.

Interview questions and answers

A company asks you to pick a cloud for a new data platform. How do you decide?

Run a structured evaluation: requirements (data volumes, latency, governance, AI needs), proof-of-concept on representative workloads, total cost over 3 years including people, skills and hiring market, integration with existing identity and tools, and contractual leverage. Present a weighted decision matrix and an ADR.

What is the biggest conceptual difference when moving from AWS to GCP?

The resource hierarchy and IAM inheritance through folders/projects instead of account-centric isolation, and the global VPC networking model with Shared VPC.