What is the Cloud?

The cloud is other people's computers, rented by the minute — data centers, AWS, and why nobody buys servers anymore.

basicscloudawsinfrastructure

The one-sentence version

The cloud is computers you rent instead of own — millions of servers in giant buildings (data centers), hired out over the internet by the minute.

There is no fluff in the sky. "Your photos are in the cloud" means: your photos are on a company's computers in a warehouse, probably in Virginia, Dublin or Mumbai.

The electricity analogy

A century ago, factories generated their own power on-site — expensive machinery, dedicated staff, wasted capacity. Then power grids arrived: plug in, pay for what you use, let specialists run the generators.

The cloud did the same thing to computing:

Own generatorsPower grid
Buy servers, install them, hire people to maintain themRent capacity from AWS/Google/Microsoft
Pay for peak capacity even at 3 AMPay only for what you use, per minute
Adding capacity takes monthsAdding capacity takes seconds

Before the cloud (roughly pre-2006), launching a web app meant buying physical servers, renting rack space, and guessing your future traffic. Guess low and you crash on launch day; guess high and you've burned your funding on idle metal. AWS (Amazon Web Services) changed this in 2006 by renting Amazon's spare computers via an API — and accidentally created what is now Amazon's most profitable business.

What a data center is

A data center is a warehouse-sized building filled with racks of servers — stripped-down computers (no screens, no keyboards) optimized to run 24/7:

  • Hundreds of thousands of machines per building
  • Industrial cooling (those machines produce serious heat)
  • Redundant power: grid + batteries + diesel generators
  • Multiple fat internet connections
  • Security somewhere between a bank and an airport

The big three cloud providers — AWS, Microsoft Azure, Google Cloud (GCP) — each run dozens of these around the planet, grouped into regions (e.g. us-east-1 in Virginia, ap-south-1 in Mumbai). You pick regions close to your users, because physics: every 1,000 km adds milliseconds (see The Internet).

What you can actually rent

Log into AWS and the "menu" (it's all APIs, of course) looks like:

  • Virtual servers — a computer of any size, ready in 60 seconds, billed per second (AWS calls this EC2)
  • File storage — effectively unlimited, pay per gigabyte (S3)
  • Managed databases — a PostgreSQL database where AWS handles the backups, updates and failover (RDS)
  • Plus ~200 more: AI models, video streaming, email sending, queues…

"Managed" is the key word that justifies cloud prices: you get the database; they get woken at 3 AM when its disk fails.

The killer feature: elasticity

Elasticity means growing and shrinking on demand — and it's the real reason the cloud won.

Think of a food-delivery app: traffic spikes 10× every day at lunch, and 50× on New Year's Eve. Owning 50× the servers for one night a year is absurd. In the cloud, you rent 50× for six hours and hand it back. Modern setups do this automatically (auto-scaling): more users arrive → more servers spin up → traffic drops → they vanish, along with their cost.

This is also why startups can exist as they do: two students with a credit card get, for a few dollars, infrastructure that would have required millions in hardware twenty years ago. Instagram served 30 million users with ~13 engineers and zero owned servers — then sold for a billion dollars.

IaaS, PaaS, SaaS — three depths of renting

Three acronyms you'll hear constantly, ordered by how much you manage yourself — the pizza analogy is the standard way to remember them:

ModelMeaningPizza versionExample
IaaS (Infrastructure as a Service)Rent raw machines; you install everythingBuy ingredients, cook at homeAWS EC2
PaaS (Platform as a Service)Push your code; the platform runs itOrder deliveryVercel, Heroku
SaaS (Software as a Service)Just use finished software in a browserEat at the restaurantGmail, Notion, Figma

This very study site is the stack in action: a static site (this code) deployed on Vercel (PaaS), which itself runs on AWS (IaaS), edited with tools like GitHub (SaaS). Turtles all the way down — every layer renting from the one below.

Industry perspective

  • Netflix runs entirely on AWS — they own essentially no servers, while competing with Amazon in video. The economics of the cloud beat the awkwardness.
  • Cloud skills are a hiring filter: Level 10 (Cloud & DevOps) — AWS, Docker, Kubernetes, CI/CD — is where this roadmap makes you employable on this front.
  • The monthly cloud bill is a real engineering concern. "We optimized the service and cut the bill 40%" is a resume bullet that gets interviews; accidentally leaving servers running is a rite of passage.
  • Interview vocabulary from this page: region, managed service, auto-scaling, IaaS/PaaS/SaaS — all reappear in Level 6 system design discussions.

Common beginner mistakes

  • "The cloud is safer/vaguer than a real place." It's a physical building. Data is subject to that country's laws — which is why companies choose regions carefully (and why "data residency" is a thing).
  • "Cloud is always cheaper." Per-hour, renting costs more than owning — like taxis vs owning a car. It wins on flexibility, zero maintenance staff and elasticity. At huge steady scale, some companies (Dropbox famously) move off the cloud to save money.
  • "Serverless means no servers." It means you don't think about the servers — they're still there, managed by the provider.
  • Confusing "the cloud" with "the internet." The internet is the road network; the cloud is rentable buildings you reach via those roads.

Interview perspective

Check yourself

  1. Your side project gets featured on the news and traffic jumps 100× for one afternoon. Walk through what happens if you own one server vs run auto-scaled cloud servers.
  2. Classify: renting a virtual machine and installing Node.js on it; pushing this site to Vercel; using Google Docs. Which is IaaS/PaaS/SaaS?
  3. Where, physically, are your WhatsApp photos? Be as specific as the concepts on this page allow.

🎉 That's Level 0 complete. You now have the full vocabulary: computers run software; the internet connects them; websites are files browsers render; frontends talk to backends through APIs; data lives in databases; and it all runs on rented computers in the cloud. Next: Level 1 — Choosing a Language, where you start writing software yourself.