Amazon Web Services
Infra / DevOps · Usage-based
Overview
Amazon Web Services, or AWS, is the cloud computing arm of Amazon. It rents computing power, storage, databases, networking, and hundreds of other services over the internet, billed by usage, so a company can run its software without owning physical servers. It launched in 2006, starting with simple storage and rented virtual machines, and is now the largest cloud provider, with a market share above 30 percent. AWS says it offers more than 200 services and runs across 39 geographic regions and 124 availability zones, with more announced. Customers range from single developers to the largest banks, governments, and technology companies.
What people use it for
Companies run their applications and websites on AWS virtual machines and container services instead of their own hardware. They store files, backups, and data lakes in S3, its object storage. They run managed databases, both traditional SQL and other types, without maintaining the database servers. They build data pipelines and analytics on its data services, and train and run machine learning and AI models on its GPU capacity and the Bedrock service. Startups use it to launch fast and scale only when they have users. Large enterprises move existing data centers onto it. Teams also use it for content delivery through CloudFront, DNS through Route 53, email sending through SES, message queues through SQS and SNS, and the identity and networking services that connect all of these together. Many teams use it purely for S3 storage and backups even when the rest of their stack runs elsewhere.
Key capabilities
AWS lists more than 200 services. Core building blocks include EC2 for virtual machines, S3 for object storage, RDS and Aurora for managed databases, Lambda for running code without servers, and VPC for private networking. It has a content delivery network with more than 750 edge locations, plus local zones and dedicated connections for low latency needs. DynamoDB is its managed NoSQL database, and Redshift is its data warehouse. Managed services cover containers with ECS and EKS, serverless containers with Fargate, streaming, search, and machine learning, including SageMaker and Bedrock for building with foundation models. Access is governed by IAM, and infrastructure can be defined as code with CloudFormation or the CDK. Everything is controlled through a web console, a command line tool, and APIs. Pricing is pay as you go by the second or hour, with Reserved Instances, Savings Plans, and Spot capacity offering discounts for committed or flexible use, and a free tier for new accounts. AWS holds a wide set of compliance certifications and has been named a leader in Gartner's cloud platform ranking for 15 straight years.
Limitations
AWS is complex. It has hundreds of services with overlapping uses, a dense console, and documentation spread across many pages, and the learning curve is steep even for experienced engineers. Cost is the other major issue. Pricing is broken into hundreds of thousands of separate line items, which makes bills very hard to forecast, and charges for data leaving AWS or moving between regions add up quietly. Teams report bills that doubled or tripled overnight after a configuration change or a traffic spike. Getting real support costs extra on top of usage. Running AWS well usually needs dedicated staff or a consultant, which puts it out of reach for small teams that just want something simple.
Insight
AWS can run essentially anything, and for a company at real scale that breadth and reliability are hard to match. That is also the problem for everyone else. The service count is overwhelming, the console is dense, and the pricing is split into so many small pieces that even careful teams struggle to predict a bill, with data transfer charges a common surprise. Most cost blowups are not exotic workloads, they are misconfigured resources left running, oversized instances, or traffic patterns nobody modeled. Running it properly takes dedicated skill, whether an in house platform team or an outside consultant. For a large or fast growing company with that skill, AWS is a reasonable default. For a small team that wants to deploy an app and move on, a simpler platform costs less in both money and time, and AWS can wait until the scale actually demands it.
Pricing
Usage-based
Website
Last checked 2026-08-30