Amazon Web Services (AWS)

Operator's take

AWS is the default infrastructure layer for teams that need serious cloud compute and don't want to manage physical servers. It slots into a no-code/low-code stack at the infrastructure tier: backend hosting, file storage (S3), managed databases (RDS, DynamoDB), serverless functions (Lambda), and API Gateway — the building blocks that no-code platforms either sit on top of or connect to via HTTP. Operators reach for it directly when a platform abstraction isn't enough, when they need fine-grained control over cost and performance, or when a client environment is already AWS-native.

The AI story at AWS has moved fast. Amazon Bedrock is the managed foundation-model layer (Claude, Llama, and others), Bedrock AgentCore is the purpose-built serverless runtime for deploying and scaling production agents, Amazon Q remains the developer-focused AI assistant (Q Developer), and Kiro is AWS's agentic IDE — now with CLI, web, and mobile surfaces alongside the IDE. Amazon Quick, announced at the "What's Next with AWS" event in April 2026, is the newer AI assistant for general work: it connects to your apps and data and acts on your behalf, and is positioned as the successor to Amazon Q Business. This is no longer a tack-on — it's central to how AWS is positioned today.

The free tier restructured: new accounts get $100 in credits upfront, can earn up to $100 more by exploring key services (up to $200 total), and get access to 90+ services for up to 6 months. The account closes automatically at 6 months or when credits run out — no auto-upgrade to billing unless you convert to a Paid plan. A separate "Always free" layer (30+ services free within monthly usage limits) still runs on both the Free and Paid plans. Cost management still becomes real work at scale without dedicated monitoring tooling in place.

What it's good at

• Serverless compute with Lambda — deploy functions that scale from zero to thousands of concurrent requests with no server provisioning; pay only for compute time consumed.
• Breadth of managed services — 240+ services covering compute, storage, networking, AI/ML, databases, and security; reduces the expertise surface needed for complex infrastructure.
• Agentic AI suite — Amazon Bedrock for managed foundation models, Bedrock AgentCore for production agent deployment, Amazon Q Developer for coding assistance, Amazon Quick as the work AI assistant (positioned as Q Business's successor), and Kiro for AI-native development; not a tack-on, it's now central to the platform.
• Global reach — multiple dozen regions and hundreds of availability zones for low-latency deployments and geographically distributed disaster recovery (exact numbers increase; check aws.amazon.com/about-aws/global-infrastructure/ for current count).
• Pay-as-you-go pricing — no upfront commitments; costs scale with usage. New accounts get up to $200 in credits over 6 months on the Free plan (no charges unless you convert to Paid).
• Security and compliance — IAM, encryption, and a large library of compliance certifications (SOC 2, HIPAA, FedRAMP); relevant for regulated industries.

What it's not

• Not approachable for non-technical operators — the console is vast and the pricing model non-trivial; without someone comfortable in cloud infrastructure, misconfiguration and unexpected bills are real risks.
• Not cost-predictable without active monitoring — pay-as-you-go flexibility cuts both ways; resource sprawl and idle services routinely surprise teams who don't have billing alerts configured.
• Not the right fit if a no-code platform already abstracts it away — Supabase, Railway, and Render run on AWS under the hood and are the right choice when you don't need direct AWS control.
• Not a quick-start environment — setup, IAM policy design, and service wiring have meaningful ramp time compared to higher-abstraction alternatives.