Back to Knowledge
Updated Sep 2026

Building with AWS

Already invested in AWS

AWS (Amazon Web Services)

Amazon's massive cloud computing platform. Offers everything from simple hosting to databases, AI, and more. Powers half the internet.

"Like a giant LEGO set for the cloud. Hundreds of services you can combine to build anything."

, or your company is? Here's how to build modern web apps on Amazon's cloud, how the free tier works now, and when it's worth the extra complexity over the Vibe Stack.

1AWS vs Vibe Stack: When to Choose What

AWS is a hardware store with every tool ever made. The Vibe Stack is a pre-assembled workbench. Both build real things; one asks you to pick every screw.

Choose AWS If...

  • Your company already runs on AWS (accounts, billing, security reviews)
  • You need custom networking: VPCs, private subnets, peering
  • Your data must stay in a specific AWS region or account for compliance
  • You need services the Vibe Stack doesn't have (queues at huge scale, data lakes, IoT)
  • Your team has AWS or infrastructure-as-code experience

Choose Vibe Stack If...

  • You want to ship this week with minimal config
  • You're building a web app, API, or AI feature
  • You want preview deployments on every PR out of the box
  • You're a solo developer or small team without DevOps help
  • You'd rather scale to zero than pay for idle servers

Our take: The Vibe Stack (Vercel + Neon) is the fastest path for most vibe coders. AWS earns its complexity when an organization, a compliance requirement, or a specific service forces the choice. You can also mix: a Vercel frontend calling AWS services is a very common setup.

2Recommended AWS Stack for Web Apps

AWS Amplify

AWS Amplify

Amazon's platform for building and deploying full-stack web and mobile apps. The current generation (Gen 2) is code-first: you define both frontend and backend in TypeScript, and it wires up AWS services like Cognito and DynamoDB for you.

"Like Vercel, but built by Amazon for the AWS family. If you're in AWS, it's a natural fit."

 Gen 2 (Hosting + Backend)

Git-based hosting for Next.js and React, plus a code-first backend

Why Use It

  • • Git-based deploys and branch previews (like Vercel)
  • • Gen 2 is code-first: backend defined in TypeScript, not console clicks
  • • Auth, data, and storage wired to real AWS services
  • • Per-developer cloud sandboxes for testing backend changes
  • • Custom domains with SSL

Gotchas

  • • Newest Next.js features can lag behind Vercel
  • • Old tutorials describe Gen 1 (CLI + console), which works differently
  • • You're still debugging AWS underneath when things break

Comparable to: Vercel for hosting, Supabase or Firebase for the backend pieces. Start a Gen 2 project from the templates in the Amplify docs so you get the current setup command.

Amazon RDS

AWS RDS (Relational Database Service)

Amazon's managed database service supporting PostgreSQL, MySQL, MariaDB, and more. Handles backups, updates, and scaling. Great for production workloads in AWS.

"Like hiring a database administrator who works 24/7. You use the database; AWS handles the maintenance."

 / Aurora (Database)

Managed PostgreSQL

PostgreSQL

A powerful, open-source relational database. Rock-solid, feature-rich, and the choice for serious production apps.

"Like the Toyota Camry of databases. Reliable, well-documented, handles anything you throw at it."

or MySQL

Why Use It

  • • Automatic backups and point-in-time recovery
  • • Multi-AZ high availability and read replicas
  • • Private networking inside your VPC
  • • Lives next to the rest of your AWS data

Gotchas

  • • A classic instance bills every hour it runs, used or not
  • • VPC and security-group setup before you connect
  • • Serverless functions need pooling (RDS Proxy) to avoid connection storms
  • • No Neon-style instant branching for dev environments

Comparable to: Neon or Supabase, without scale-to-zero branching by default. See Databases for the side-by-side.

AWS Lambda

AWS Lambda

Amazon's serverless compute service. Run code in response to events without managing servers. Pay only when your code runs. The original serverless platform.

"Like a vending machine for code. Insert event, get result, pay per use."

 (Backend / Events)

Serverless

Serverless

A cloud model where you don't manage servers: your code runs in response to requests or events, scales automatically, and you pay for usage. Modern platforms like Vercel's Fluid Compute reuse warm instances for many requests at once and bill for active CPU time, which makes serverless a much better fit for slow AI calls.

"Like renting a kitchen by the meal instead of buying a restaurant. Use it, pay for it, done."

functions that run on demand

Why Use It

  • • Pay per use, no servers to manage
  • • Triggered by almost any AWS event (uploads, queues, schedules)
  • • Scales automatically
  • • Many languages, including Node.js and Python

Gotchas

  • • Cold starts on idle functions
  • • Execution-time and payload limits (check current quotas)
  • • HTTP APIs need extra wiring (API Gateway or function URLs)
  • • Local testing takes extra tooling

Comparable to: Vercel Functions, which run on Fluid Compute

Fluid Compute

Vercel's default function runtime model (on for new projects since April 2025). Instead of one request per function instance, an instance can handle many requests at once, keep working after the response with `waitUntil`, and you're billed for active CPU time rather than time spent waiting on things like AI responses.

"Like a waiter who serves several tables at once instead of standing idle while one table's food cooks."

by default (in-function concurrency, fewer cold starts, billing on active CPU). Lambda gives you more knobs and deeper AWS event integration; Vercel gives you less to configure.

Amazon Cognito (Authentication)

User sign-up, sign-in, and tokens

Why Use It

  • • User pools with email/password and social logins
  • • MFA support
  • • JWTs

    JWT (JSON Web Token)

    A compact, secure way to transmit information between parties. Often used for authentication tokens after login.

    "Like a tamper-proof wristband at a concert. Shows you're allowed in without checking the list every time."

     that other AWS services understand
  • • Amplify Gen 2 auth is built on it

Gotchas

  • • IAM roles and settings are easy to misconfigure
  • • Hosted UI customization is limited
  • • Migrating users out later is painful

Comparable to: Clerk, Auth0, Better Auth, or Supabase Auth. See Auth for how they compare.

3The Free Tier & What Actually Costs Money

The AWS free tier is now credit-based. New accounts get $100 in credits at signup, plus up to $100 more as you explore services: up to $200 over 6 months.

Credits

Spend them on almost anything while you learn. When they run out (or the 6 months end), you're on normal pay-as-you-go pricing.

Always-free services

30+ services keep a monthly free allowance after that. Check aws.amazon.com/free for which ones and how much.

Always-on resources

A classic RDS instance, NAT gateways, load balancers

Bill every hour, even at zero traffic. Usually the biggest line.

Pay-per-use resources

Lambda, API Gateway, S3, Cognito

Cheap or free at hobby scale; grows with traffic.

Sneaky extras

Data transfer out, CloudWatch logs, multi-AZ copies

Small individually, surprising together.

Compare honestly: The Vibe Stack's free tiers (Vercel Hobby, Neon Free with scale-to-zero) cover most side projects indefinitely. On AWS, price out your design in the AWS Pricing Calculator before you build, and set a budget alert on day one.

4Building AWS Apps with AI

Claude Code

Claude Code

Anthropic's agentic coding tool. It lives in your terminal (and in VS Code, JetBrains, and on the web), reads your codebase, edits files, runs commands, and ships code. Install with the native installer (`curl -fsSL https://claude.ai/install.sh | bash` on macOS/Linux, `irm https://claude.ai/install.ps1 | iex` on Windows); it needs a Pro, Max, Team, Enterprise, or Console account.

"Like having a senior developer living in your terminal, ready to help 24/7."

 (Recommended)

Claude Code is strong at AWS: the SDKs, CDK, CloudFormation, IAM policies, and reading the error messages that come back. It can run the AWS CLI for you in your terminal, so it can inspect real resources instead of guessing.

Tip: Name the exact services and regions you want ("Lambda behind API Gateway in us-east-1, Postgres on RDS"). AWS has hundreds of services and clarity prevents creative detours. Keep it in plan mode before anything that creates or deletes resources.

Company requires AWS billing? Claude Code can run through Amazon Bedrock with your AWS credentials. Run claude, choose the 3rd-party platform option, then Amazon Bedrock, or set it manually:

Terminal

export CLAUDE_CODE_USE_BEDROCK=1
export AWS_REGION=us-east-1   # optional if your AWS profile sets a region
claude

Details and required IAM permissions: Claude Code on Amazon Bedrock.

AWS's own AI coding tools

Amazon offers its own AI coding assistants and an agentic IDE, plus AI help inside the AWS Console. They're tuned for AWS services and best practices, and they're a natural fit if your organization already standardizes on AWS tooling. Product names and plans change often, so check AWS's current developer-tools page rather than trusting a blog post (or an AI's memory).

Other AI tools

Cursor, GitHub Copilot, and Gemini-based tools all know AWS basics. Whatever you use, the risk is the same: models trained on old docs suggest deprecated patterns. Paste in current docs or connect a docs server via MCP when accuracy matters.

5Quick Start: Host a Next.js App on Amplify

1

Create an AWS account and lock it down

Sign up at aws.amazon.com and claim the new-account credits. Turn on MFA for the root user, create a separate day-to-day user or SSO login, and set a budget alert before anything else.

2

Push your app to Git

Amplify deploys from Git providers such as GitHub, GitLab, and Bitbucket. Make sure your Next.js app builds locally with npm run build first.

3

Open Amplify in the AWS Console

Search for "Amplify", choose to deploy an app, and connect your Git provider.

4

Pick the repo and branch

Amplify detects Next.js and proposes build settings. Add your environment variables (DATABASE_URL, API keys) here, not in the code.

5

Deploy

Save and deploy. Every push to that branch redeploys, and you can turn on previews for pull requests.

Live! Your app gets an amplifyapp.com URL; add a custom domain in the Amplify settings. Want auth or a database managed by Amplify too? Add a Gen 2 backend following the Amplify docs' Next.js quickstart. Before real users arrive, run through the deployment checklist.

6Common AWS Gotchas

IAM permissions hell

IAM (Identity and Access Management) controls who can do what. It's powerful and confusing, and you will meet "AccessDenied" often. The tempting fix (give everything admin) is how accounts get breached.

Fix: Use AWS-managed policies to start, then tighten. Paste the exact error into Claude and ask for the minimal policy that allows that one action. Never commit access keys; prefer SSO or short-lived credentials.

VPC networking complexity

RDS lives inside a VPC. Getting Lambda (or your laptop) to reach it means subnets, security groups, and sometimes a NAT gateway, which is an always-on cost.

Fix: Use RDS Proxy for serverless callers, or consider Aurora Serverless. Or skip the VPC work entirely with Neon or Supabase over a pooled connection.

Lambda cold starts

Functions that haven't run recently take extra time to start, which can make an API feel sluggish, especially with large bundles or VPC access.

Fix: Keep bundles small, pay for provisioned concurrency on hot paths, or host the API on Vercel Functions where Fluid Compute reuses warm instances across requests.

Surprise bills

Forgotten test databases, NAT gateways, data transfer, and verbose logs keep billing quietly. Credits hide this for a few months, then the real bill arrives.

Fix: Set AWS Budgets alerts immediately. Review Cost Explorer weekly while learning. Tag resources per project and delete what you're not using. Infrastructure as code (CDK, or Amplify Gen 2) makes teardown one command instead of an archaeology dig.

Ready to build on AWS?

AWS is powerful. Just make sure it's the right fit, and compare with Google Cloud before you commit.

AWS Amplify Docs Compare: Google Cloud The Vercel Playbook