Cloud Computing — Core Concepts
What Cloud Computing Actually Is
Before the cloud, if you wanted to run a website, you needed to:
- Buy a physical server (~$3,000–$20,000)
- Rent space in a data center
- Pay for power, cooling, and network
- Hire someone to maintain it
- Wait 2–4 weeks for delivery and setup
If your site got popular and you needed more capacity, you ordered more servers and waited again.
Cloud computing replaced this model with on-demand computing resources over the internet. You provision a virtual server in seconds, use it for an hour or a decade, and pay a per-hour rate. When you’re done, you release it.
This changed the fundamental economics of building software.
The Three Service Models
Cloud providers offer computing at different levels of abstraction. Most people have heard the acronyms; here’s what they actually mean:
IaaS — Infrastructure as a Service
You rent raw hardware resources: virtual machines, storage, networking. You’re responsible for everything above the hardware — operating system, runtime, your application.
Example: Renting a virtual machine on AWS EC2 or Google Compute Engine. You get a Linux box; you handle the rest.
Who uses it: Engineering teams who need control. Startups, enterprises running complex workloads.
PaaS — Platform as a Service
The provider manages the infrastructure and the runtime environment. You just deploy your code.
Example: Heroku, Google App Engine, Vercel. You push code; it runs. No server configuration.
Who uses it: Developers who want to ship fast without infrastructure headaches.
SaaS — Software as a Service
The entire application is cloud-hosted. You just use it through a browser.
Example: Gmail, Salesforce, Slack, Figma. No installation, no maintenance.
Who uses it: Everyone. This is the layer most consumers interact with daily.
The Big Three Providers
Three companies control roughly 65% of all cloud infrastructure worldwide:
| Provider | Market Share | Known For |
|---|---|---|
| Amazon Web Services (AWS) | ~31% | Broadest service catalog, first mover |
| Microsoft Azure | ~25% | Enterprise integration, Office 365 |
| Google Cloud (GCP) | ~11% | AI/ML tooling, BigQuery analytics |
AWS launched in 2006 — a full three years before the term “cloud computing” entered mainstream use. Their early head start is why Netflix, Airbnb, Pinterest, and thousands of major companies are still on AWS today.
How the Economics Changed Everything
The old model was CapEx (capital expenditure): buy equipment upfront, depreciate over 3–5 years, hope you guessed your capacity right.
The cloud model is OpEx (operational expenditure): pay for what you use, scale up or down in minutes, no stranded hardware if your startup fails.
For a company like Airbnb, which sees 10x more traffic over New Year’s Eve than a random Tuesday in February, this is transformative. They scale to demand, then scale back. No wasted hardware sitting idle 90% of the year.
Common Misconception: “The Cloud Is Less Secure”
Early cloud skeptics argued that storing data on someone else’s hardware was inherently risky. The reality turned out to be the opposite.
AWS, Azure, and GCP have security teams of thousands, hardware-level encryption, SOC 2 / ISO 27001 certifications, and physical data centers that look like military bunkers. Most companies — including hospitals, banks, and government agencies — are more secure on the cloud than they were running their own servers.
The risk isn’t “cloud vs. on-premise.” The risk is misconfiguration — which is a human problem, not a cloud problem.
One Thing to Remember
Cloud computing didn’t just move computers off-site — it turned infrastructure from a capital investment into a utility, like electricity. You use what you need, pay for what you used, and never think about the generator.
See Also
- Containerization Why does software that works on your computer break on everyone else's? Containers fix that — and they're why Netflix can deploy 100 updates a day without the site going down.
- Vector Databases Google finds web pages by keywords. Your brain finds memories by vibes. Vector databases are how AI does the brain thing — and it's weirder than you'd expect.