☁️ Module 15 — The Cloud: Renting Other People’s Computers

What you’ll learn: what “the cloud” physically is (hint: it’s buildings), why companies rent computers instead of buying them, what words like “serverless” and “SaaS” honestly mean, why your Netflix show loads fast, and why cloud bills make grown finance directors cry.

Why it matters: “the cloud” might be the most-used and least-understood phrase in modern life. Your photos are in it. Your email is in it. Your company probably pays for it. After this lesson, the fog lifts: you’ll know exactly where your data lives and why the whole world decided to rent instead of own.

🧠 The cloud is not a cloud

Have you ever wondered where your photos actually go when they’re “backed up to the cloud”? The word is doing a lot of marketing work. It sounds soft, weightless, everywhere and nowhere.

Here’s the honest version: the cloud is other people’s computers. That’s it. That’s the secret. When your phone says your photos are “in the cloud,” it means they’ve been copied onto a computer owned by Apple or Google, sitting in a very real building, in a very real place, plugged into a very real wall.

That building is called a data center — a warehouse-sized building packed with thousands of computers stacked in racks, running day and night. Picture an enormous supermarket, but instead of shelves of cereal, there are aisles and aisles of humming metal boxes with blinking lights. Data centers are usually built near cheap electricity (hydroelectric dams are popular neighbors) because all those computers are hungry. And because thousands of computers generate serious heat, the whole building is cooled like a giant refrigerator — some data centers spend nearly as much energy on cooling as on computing.

So the next time someone says “it’s in the cloud,” you can silently translate: “it’s on a specific metal box, in a specific warehouse, probably in Oregon or Ireland or Singapore.” Your photos have a street address. (Usually three street addresses — we’ll get to that.)

✅ Check yourself

  1. Physically speaking, what is “the cloud”?
  2. Why are data centers often built near dams?
Show answers 1. Other people's computers — thousands of them, in warehouse-sized buildings called data centers. 2. Cheap electricity. Data centers use enormous amounts of power, so being near a cheap source (like hydroelectric) saves millions.

🔌 Why rent computers when you could own them?

Fair question. You own your phone. Companies used to own their computers too — every business had a “server room,” a chilly closet full of machines that one stressed person kept alive. So why did nearly everyone switch to renting?

🌉 Analogy: Think about electricity. You could generate your own power at home — buy a generator, store fuel, maintain it, size it for your busiest day. Nobody does this. You plug into the grid and pay for exactly what you use. A century ago, factories really did run their own generators; then power plants made electricity so cheap and reliable that owning a generator became silly. The cloud did the same thing to computing. Amazon, Microsoft, and Google became the power plants; companies threw out their generators.

Where the analogy breaks down: electricity is identical everywhere — a volt is a volt. Computing isn’t. Each cloud company offers different tools with different names, which matters later when we talk about getting locked in.

Here’s a second picture, because this idea deserves two.

🌉 Analogy: Imagine you want to sell your grandmother’s famous samosas. Option one: build your own commercial kitchen — €80,000, six months, and you’re stuck with it if the business flops. Option two: rent a professional kitchen by the hour. Show up, cook, pay for the hours you used, walk away. If orders explode, rent more hours. If they don’t, you’ve lost almost nothing. The cloud is the rent-by-the-hour kitchen for computing.

Where the analogy breaks down: when you leave a rented kitchen, you take your pots home in one trip. Moving years of data and software out of a cloud is much slower and messier — more on that in the “downsides” section.

🍕 The three levels of renting

Cloud companies rent at three levels of “how much do you want done for you?” The industry has acronyms for these (IaaS, PaaS, SaaS), but forget the acronyms — remember the pizza.

Here’s the fun realization: you already use the cloud every day. Gmail is cloud software. Netflix is cloud software. You’ve been a cloud customer for years without signing anything that said “cloud.”

The three big landlords renting all of this out are Amazon Web Services (AWS), Microsoft Azure, and Google Cloud. Between them they run most of the internet you touch.

Level Kitchen version You handle Example
Bare computers Empty rented kitchen Almost everything A company running its own app on rented machines
Managed platform Kitchen with staff Only your recipe (code) A startup that only writes code, no machine-wrangling
Finished software Ordering the pizza Nothing — you’re the diner Gmail, Netflix, Google Docs

📈 Why everyone moved: the Black Friday problem

Three reasons dragged nearly every company into the cloud.

Reason 1: scaling. Imagine you run an online shop. On a normal Tuesday you need 10 computers to handle visitors. On Black Friday you need 1,000 — for one day. If you own your hardware, you have two terrible options: buy 1,000 computers that sit idle 364 days a year, or keep 10 and watch your site collapse at your most profitable moment. With the cloud, you rent 990 extra computers on Friday morning and hand them back on Monday. You pay for one weekend. This ability to grow and shrink on demand is called scaling, and it was flatly impossible in the owned-hardware world.

Reason 2: reliability. Cloud providers keep multiple copies of your data in multiple buildings — often three copies in three different cities. This is called redundancy: keeping spare copies so that no single failure can lose your data. If a data center in one city floods, catches fire, or loses power, your photos are still sitting safely in two other cities, and the switch-over happens so fast you never notice. Could your company’s server closet survive a flood? Exactly.

Reason 3: someone else patches the roof. Computers need constant care: security updates, broken parts replaced, software upgraded at 3 a.m. In the cloud, that’s the landlord’s problem. Your team stops changing lightbulbs and gets back to building the actual product.

🚕 “Serverless” — the most misleading word in tech

You may hear engineers say they built something “serverless.” Let’s be honest about this word: there are always servers. (Remember from Module 03: a server is a computer that stays on all the time so other computers can ask it for things.)

Serverless means you never think about the servers. You hand the cloud a small piece of code and say “run this whenever someone clicks the button,” and the cloud finds a computer, runs it, and bills you for the milliseconds used.

🌉 Analogy: Owning a car versus taking taxis. Taxis still involve cars — but you never buy one, insure one, park one, or change its oil. A car materializes when you need it and vanishes when you’re done. “Serverless” is taxis for computing.

Where the analogy breaks down: taxis cost more per trip than driving yourself. Serverless is often cheaper for occasional work — but like taxis, if you ride constantly, owning (or long-term renting) wins.

🌍 Why Netflix keeps a copy near your house

Data travels fast — but not instantly. A request from Sydney to a server in Virginia and back takes around 200 milliseconds even in the best case, because the signal physically crosses the Pacific through cables on the ocean floor. That delay is called latency — the waiting time between asking for something and getting the first response.

Cloud providers fight latency with regions: clusters of data centers spread around the world (Virginia, Frankfurt, Mumbai, São Paulo…), so companies can run their software near their users.

For heavily-watched content, there’s an extra trick. A CDN (content delivery network) is a web of small storage outposts scattered across hundreds of cities that keep copies of popular files close to viewers. When a show is hot, Netflix doesn’t stream it to you from a distant headquarters — a copy is already sitting in a facility near your city. It’s the same logic as a supermarket chain keeping a warehouse near every big town instead of shipping every yogurt from one national depot. One idea, one paragraph, and now you know why video rarely stutters.

✅ Check yourself

  1. What does “serverless” honestly mean?
  2. Why does Netflix keep copies of shows in many cities?
Show answers 1. There *are* servers, but you never manage or think about them — like taking taxis instead of owning a car. You hand over code; the cloud runs it and bills per use. 2. To reduce latency. Data crossing oceans takes real time, so keeping popular files physically near viewers (via a CDN) makes streaming fast and smooth.

💸 The meter is always running

Renting has a catch: the meter never stops. Cloud bills work like utility bills — you pay for every hour a computer runs, every gigabyte stored, every file sent across the internet.

⚠️ Watch out: a rented computer that nobody remembered to switch off keeps billing forever, like a tap left running in a house you moved out of. Companies routinely discover they’ve spent thousands per month on machines doing absolutely nothing. This is so common that “cloud cost engineer” is now a real, well-paid job: a person whose entire role is walking through the building turning off taps.

For a small experiment, cloud costs are close to zero. At big-company scale, cloud bills run to millions per month — which is still usually cheaper than the buildings, hardware, and staff it replaces. Usually.

⚠️ The honest downsides

The cloud won for good reasons, but let’s not write an advertisement.

Lock-in. Each cloud provider has its own special tools, and the more of them you use, the harder it is to leave — like a landlord whose apartment comes with custom furniture that fits no other building. Moving a large company from one cloud to another can take years. Providers know this, and it shapes their prices.

Outages. When millions of businesses rent from three landlords, one landlord’s bad day becomes everyone’s bad day. When a major cloud region fails — and it has: AWS’s US-East region has had famous stumbles that briefly broke apps, smart doorbells, and news sites across the internet at once — half the web catches a cold from a single sneeze. An outage is a period when a service is down and unreachable.

Privacy and sovereignty. Your data sitting in another company’s building — possibly in another country’s building — raises real questions. Whose laws apply to your medical records if they’re stored abroad? Many countries now require certain data (health, banking, government) to stay physically inside their borders — this idea is called data sovereignty. It’s a genuine, unresolved tension, and it’s why cloud providers keep opening data centers in more countries.

The environment. Data centers consume real electricity and real water (for cooling) — globally, data centers use roughly as much electricity as a mid-sized country. The honest picture: the big providers are among the world’s largest buyers of renewable energy and are racing to run on wind, solar, and cleverer cooling, partly from conscience and partly because power is their biggest bill. Progress is real; so is the footprint. Both things are true.

📝 Recap

🚪 Next up

You now know where modern software lives. Next, the most hyped question on Earth: what is AI, actually — and how does a machine “learn”? We’ll strip away the magic and the marketing in Module 16 — AI & Machine Learning Without the Hype.