πΊπΈ USA Β· AWS (Amazon Web Services) overview
Status: π© COMPLETE π¦ LIVING Last updated: 2026-06-26 Plain-English tagline: The worldβs largest cloud platform. Hundreds of services for compute, storage, databases, networking, AI, ML, and almost any back-end need. Underpins a huge slice of the internet and is the default cloud for many AUS enterprises.
Front-matter facts
| Field | Value |
|---|---|
| Vendor | Amazon Web Services (Seattle, USA) β subsidiary of Amazon.com Inc |
| Country / origin | πΊπΈ USA |
| Recommended for Australian users? | β
Yes β AWS has two AUS regions (ap-southeast-2 Sydney, ap-southeast-4 Melbourne); strongest AUS-cloud ecosystem; IRAP-assessed for AUS government |
| Privacy summary | Enterprise-grade with no-training-by-default for AI services; AUS data residency via Sydney / Melbourne; SOC 2, ISO 27001, HIPAA, IRAP, GDPR |
| Free tier | Yes β AWS Free Tier with 12-month free + always-free + short-term trials across many services |
| Paid tiers | Pay-as-you-go usage-based; AUD billing native via AWS Australia; enterprise discounts via reserved instances / savings plans |
| First released | 2002 (early services); SQS / S3 / EC2 launched 2006 |
| Last reviewed | 2026-06-26 |
| Official site | https://aws.amazon.com |
What it is
AWS (Amazon Web Services) is the worldβs largest cloud platform β over 200 services covering compute, storage, databases, networking, security, AI / ML, IoT, analytics, developer tools, content delivery, and more. Itβs a separate division of Amazon (the e-commerce parent) and far larger than the e-commerce business in terms of profit.
AWS effectively defined the modern cloud-services category starting in 2006 with EC2 (rentable compute), S3 (storage), SQS (message queues). It has dominated the cloud market ever since, with ~30-33% global cloud market share as of mid-2026.
For AI specifically, AWS provides:
- AWS Bedrock β managed access to Claude, Llama, Mistral, Cohere, Amazon Nova models (see aws-bedrock.md)
- Amazon Q β Business (enterprise AI assistant) + Developer (coding AI)
- Amazon SageMaker β ML platform for training / deploying your own models
- Amazon Nova β Amazonβs own foundation model family
- AWS Trainium / Inferentia β custom AI chips
- Cloud GPUs via EC2 P-series, G-series (Nvidia hardware)
For Australian customers specifically, AWS has unique strengths:
- Two AUS regions: Sydney (
ap-southeast-2, established 2012) and Melbourne (ap-southeast-4, established 2023) β strong for data sovereignty + low latency - IRAP Protected assessment β assessed at PROTECTED level by ASD for AUS government use
- Strong AUS partner ecosystem β many AUS-based AWS consulting partners
- AUD billing native via AWS Australia Pty Ltd
- Used by: Atlassian, Canva, REA Group, Carsales, Telstra (many workloads), Westpac, ANZ, NAB (specific workloads), AUS government departments
What youβd use it for
Infrastructure
- Host web apps (EC2 / ECS / Fargate / Lambda)
- Store data (S3 for files, RDS for SQL, DynamoDB for NoSQL, OpenSearch, etc.)
- Network (VPC, Route 53 DNS, CloudFront CDN)
- Email (SES β Simple Email Service)
AI / ML
- Run inference on Bedrock (Claude / Llama / Nova / etc.)
- Train custom models (SageMaker)
- Add AI features to apps (Comprehend NLP, Rekognition vision, Polly TTS, Transcribe STT)
- Cloud GPUs for model training / fine-tuning
Data
- BigQuery-style analytics (Athena, Redshift)
- Data lakes (S3 + Glue + Lake Formation)
- Real-time streaming (Kinesis)
Developer tools
- Source control + CI/CD (CodeCommit, CodeBuild, CodePipeline, CodeCatalyst)
- Container registry + orchestration (ECR, EKS, Fargate)
- Observability (CloudWatch, X-Ray)
For Bible Quest-scale projects, AWS is overkill β Vercel + Supabase covers similar ground at much lower complexity. AWS shines for larger / enterprise / heavy infrastructure.
How to sign up + first 5 minutes from Australia
- Go to
aws.amazon.com. Click Create Free Account. - Enter email, account name; verify
- Credit card required (AUS Visa / Mastercard / Amex accepted); wonβt be charged for Free Tier usage
- Identity verification (phone + maybe ID for some accounts)
- Pick support plan β start with Basic (free)
- Sign in to AWS Console (console.aws.amazon.com)
- Set up MFA immediately on root account
- Create an IAM user for day-to-day work (donβt use root after initial setup)
- Pick AUS region β
ap-southeast-2(Sydney) is the default for AUS customers - Try Free Tier services: spin up a t2.micro EC2 instance, S3 bucket, Lambda function
What it costs
Free Tier (12-month free, always-free, trials)
- EC2 t2.micro/t3.micro: 750 hours/month (~1 always-on instance)
- S3: 5GB storage, limited requests
- RDS: 750 hours of small instance + 20GB storage
- Lambda: 1M requests/month always-free
- DynamoDB: 25GB storage always-free
- Bedrock: some free quota for development
- Many smaller services: always-free at low usage
Pay-as-you-go (most customers)
- Per-second / per-hour / per-GB / per-request pricing
- Pricing varies massively by service + region + tier
- AUD billing via AWS Australia
- Sydney region is typically slightly more expensive than US-East (us-east-1) β pay for AUS data residency
Enterprise discounts
- Reserved Instances (commit to 1- or 3-year usage for ~30-70% off)
- Savings Plans (flexible commitments)
- Enterprise Discount Program (negotiated discounts for large customers)
- AWS Marketplace Private Offers (vendor-discounted bundles)
Hidden costs to know about
- Data egress (data leaving AWS) is the most common surprise β can be ~AUD $0.15/GB; large outflows add up
- NAT Gateway, Load Balancer, RDS can accumulate at low-medium usage
- Forgotten resources β leaving EC2 / RDS / NAT Gateway running unintentionally
- Cross-region replication has both data-egress + storage costs
- AWS Bills tab + Cost Explorer + Budgets β set alerts to catch surprises early
How it compares to alternatives
| Aspect | AWS | Microsoft Azure | Google Cloud (GCP) | Cloudflare |
|---|---|---|---|---|
| Market share | #1 (~30-33%) | #2 (~22-25%) | #3 (~10-12%) | Specialised |
| Services breadth | Most (200+) | Broad (200+) | Broad | Specialised (edge, security) |
| AI / ML services | Best Bedrock + SageMaker | Best Azure OpenAI Service | Best Vertex AI + Gemini | Workers AI + AI Gateway |
| Microsoft / Office integration | Limited | Best (native) | Limited | Limited |
| Google / Workspace integration | Limited | Limited | Best (native) | Limited |
| AUS regions | 2 (Sydney + Melbourne) | 2 (Australia East + Australia Central) | 2 (Sydney + Melbourne) | Global edge |
| AUS market presence | Strongest | Strong | Strong | Strong |
| Developer experience | Vast but complex | Strong | Cleanest APIs | Excellent (edge focus) |
| Best for AI | Multi-model via Bedrock | Microsoft-stack + OpenAI | Google models + open weights | Edge AI inference |
For most AUS organisations starting fresh: AWS if AI-heavy / multi-vendor; Azure if M365-heavy; GCP if Workspace-heavy / Google AI-heavy; Cloudflare if edge-AI-heavy or developer-velocity-focused.
Privacy / data handling
- Customer data is yours β AWS doesnβt access without explicit permission
- No training on customer data in Bedrock, SageMaker, Q (contractually)
- AUS data residency via Sydney / Melbourne regions
- IRAP Protected assessment for AUS government workloads
- HIPAA, GDPR, SOC 2 Type II, ISO 27001, ISO 27017, ISO 27018 all compliant
- Tenant-isolated by AWS account structure
- Encryption at rest + in transit by default for most services
- Audit logs via CloudTrail
- Customer-managed encryption keys (CMK) via KMS
Recent changes
- 2026: Nova family expanded; Amazon Q matured; new Sydney + Melbourne edge zones
- 2025: AWS Trainium 2 launched; expanded Bedrock model catalogue
- 2023: Melbourne region (
ap-southeast-4) launched - 2012: Sydney region (
ap-southeast-2) launched - 2006: S3, EC2, SQS launched
Gotchas
- AWS is huge β 200+ services is genuinely overwhelming; start small
- Pricing complexity is the #1 surprise β use Calculator (calculator.aws) before committing
- Egress costs trap many users β modelling them is essential
- Console UI is dense β the AWS Console is functional but not friendly; CLI / SDK / Terraform often easier for repeated tasks
- For small projects, Vercel + Supabase / Cloudflare offer simpler experience
- AWS Certified certifications are valuable for AUS IT careers β Solutions Architect Associate is the common starting point
- IAM (Identity and Access Management) is genuinely hard β most AWS security incidents trace to IAM misconfiguration
- Save your AWS Free Tier email β billing surprises if you exceed Free Tier without realising
- Australian Privacy Act: AWS publishes APP-compliant Data Processing Addendum; valid for AUS personal-information handling
- For tax-invoice handling (Australian businesses), enable AWS Australia billing entity properly
See also
- AWS Bedrock π₯ β AWSβs AI model platform
- AWS SageMaker π₯ β ML training / deployment
- Amazon Q Business π© π¦
- Amazon Q Developer π© π¦
- Amazon Nova π© π¦
- Microsoft Azure overview π₯ β main competitor
- Google Cloud overview π₯ β main competitor
- Cloudflare π© π¦
- Vercel π© π¦ β simpler alternative for many projects
- What is the cloud? π© π¦
- Containers and Docker π©
- Decision frameworks β Self-host vs SaaS π©