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 generators | Power grid | |
|---|---|---|
| Buy servers, install them, hire people to maintain them | Rent capacity from AWS/Google/Microsoft | |
| Pay for peak capacity even at 3 AM | Pay only for what you use, per minute | |
| Adding capacity takes months | Adding 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:
| Model | Meaning | Pizza version | Example |
|---|---|---|---|
| IaaS (Infrastructure as a Service) | Rent raw machines; you install everything | Buy ingredients, cook at home | AWS EC2 |
| PaaS (Platform as a Service) | Push your code; the platform runs it | Order delivery | Vercel, Heroku |
| SaaS (Software as a Service) | Just use finished software in a browser | Eat at the restaurant | Gmail, 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
- 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.
- Classify: renting a virtual machine and installing Node.js on it; pushing this site to Vercel; using Google Docs. Which is IaaS/PaaS/SaaS?
- 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.