When AWS sneezes, the internet catches a cold ⛈️

Cliquez ici pour lire en français

Monday, October 20th, 2025, 9 AM. You try to open Snapchat to share your morning coffee. Nothing. You attempt to launch Fortnite. Black screen. Your Alexa voice assistant refuses to respond. Even your online banking displays an error message. You’re not alone: millions of people worldwide experienced the same frustration. The culprit? A massive outage at AWS, a name you may have never heard of, yet one that powers a significant chunk of your digital life.

How can a single company simultaneously knock out Snapchat, Netflix, British banks, small Quebec businesses, and your favorite video game? Welcome to the behind-the-scenes reality of the modern web, where everything rests on invisible and surprisingly fragile foundations.

So what exactly is AWS ? 🤔

Imagine you want to open a restaurant. You’d need an equipped kitchen, refrigerators, ovens, a location, electricity… In short, an entire costly infrastructure before even serving your first dish. Now imagine that instead of buying all of that, you could rent exactly what you need, when you need it, and pay only for what you use.

That’s precisely what AWS (Amazon Web Services) does for the internet. Launched in 2006 by Amazon, AWS rents out what are called « servers » and computing services to businesses worldwide. But what exactly is a server?

Image générée avec une IA / Image generated with AI

A server is simply a very powerful computer that runs continuously to store information and operate applications. When you watch a show on Netflix, your vacation photos are stored somewhere on servers. When you play games online, a server coordinates the match. When you send a message on WhatsApp, it passes through servers.

Traditionally, every company had to buy and maintain its own servers in its own offices. It was expensive, complicated, and inflexible. Then came the « cloud, » a marketing term for storing your data and running your applications on someone else’s servers, accessible via the internet.

AWS has become the undisputed king of this market. Today, the company controls roughly one-third of the global cloud market—twice as much as its nearest competitor, Microsoft Azure. That’s massive.

How AWS became the internet’s invisible backbone  🏗️

AWS’s story begins almost accidentally. In the early 2000s, Amazon was « just » an e-commerce giant. To handle its traffic spikes (particularly during sales periods), the company had developed a super-powerful and ultra-flexible computing infrastructure.

Amazon’s engineers had a revelation: what if we sold this excess computing capacity to other businesses? In 2006, AWS was born with a simple but revolutionary promise: instead of investing millions in your own servers, rent ours on demand.

Image générée avec une IA / Image generated with AI

The success was meteoric. Cash-strapped startups could suddenly launch with infrastructure worthy of a multinational corporation. Large companies could abandon their expensive data centers. Even government agencies made the leap.

Today, AWS hosts thousands of services you use daily without realizing it. Netflix? On AWS. Spotify? AWS. Huge swaths of the U.S. government? AWS. Banks, hospitals, schools, video games, dating apps… The list is staggering.

But this concentration creates a major problem: AWS has become a single point of failure. When it goes down, it’s a domino effect.

Black Monday: anatomy of a digital disaster 💥

Let’s return to that fateful Monday, October 20th, 2025. At 3:11 AM Eastern Time, something breaks in a region AWS calls « US-EAST-1, » located in Northern Virginia. This region is AWS’s Mecca: it’s the oldest, the most important, the one handling the most global traffic.

The culprit? A malfunction in a load balancer monitoring system. To understand what that is, imagine a massive shopping mall on a Saturday afternoon. Load balancers are like security guards directing customer flows to different entrances to prevent bottlenecks. When this system fails, everyone gets stuck at the same door.

In AWS’s case, this malfunction triggered a spectacular domino effect. AWS services constantly communicate with each other. When one stops responding properly, it contaminates the others. It’s like a house of cards: remove one and everything collapses.

Image générée avec une IA / Image generated with AI

The consequences were immediate and global. Snapchat became unusable. Fortnite kicked out its players. Millions of people couldn’t access their bank accounts. Quebec roofing companies couldn’t manage their orders. Candy makers in Lanaudière couldn’t ship their products.

Even more concerning: critical services like British banks Lloyds and HSBC experienced disruptions. Imagine the consequences if such an outage had lasted several days instead of a few hours.

Why the entire web relies on so few players  🎯

You might be wondering: « How did we end up putting all our eggs in one basket? » The answer is both simple and complex.

First, it’s a matter of economics. Building and maintaining your own servers costs a fortune. You need to buy hardware, rent secure facilities, pay technical teams 24/7, manage outages, replace obsolete equipment… For a startup or SME, it’s simply out of reach. AWS, Microsoft Azure, and Google Cloud let you start with just a few dollars per month.

Image générée avec une IA / Image generated with AI

Second, it’s a matter of expertise. Running modern computing infrastructure requires rare and expensive talent. AWS employs thousands of engineers among the world’s best. What company can compete with that?

Finally, it’s a matter of reliability. Paradoxically, despite this outage, AWS remains generally very reliable. Services run 99.9% of the time. Building an equally robust infrastructure yourself is extremely difficult.

Result: three companies (Amazon, Microsoft, and Google) share two-thirds of the global cloud market. This concentration obviously raises geopolitical questions. What happens if the U.S. decides to cut off access to non U.S. companies?  What happens in the event of a massive cyberattack? What happens if a simple technical error, like Monday’s, lasts not a few hours but several days?

The fragile balance of the modern web ⚖️

This outage reminds us of an uncomfortable truth: the internet, which we imagine as a decentralized and resilient network, actually rests on a few critical chokepoints. It’s as if all the roads in a country converged on a single bridge. As long as the bridge holds, everything’s fine. But if it collapses…

Solutions exist in theory. Companies can distribute their services across multiple cloud providers (called « multi-cloud »). They can also design their applications to keep functioning even if part of the infrastructure fails. But these approaches are expensive and require specialized technical expertise.

Image générée avec une IA / Image generated with AI

In Europe, the question of « digital sovereignty » resurfaces with every major incident. Should we develop our own European alternatives to AWS? The project exists (with initiatives like Gaia-X), but the gap is enormous and resources are limited.

Regulation is also evolving. DORA (Digital Operational Resilience Act), entering into force in 2025 in Europe, requires financial institutions to better assess their critical service providers and prepare failure scenarios. It’s a start, but the road is long.

Living with the risk 🥲

The October 20th, 2025 outage will go down in history as a brutal reminder of our collective dependence on a few tech giants. Within hours, millions of people and businesses were paralyzed by a technical problem in a data center in Virginia.

This fragility is the price we pay for the convenience of the cloud. We’ve traded the complexity of managing our own servers for the simplicity of entrusting everything to specialists. But like all centralization, it creates systemic vulnerabilities.

Image générée avec une IA / Image generated with AI

The question is no longer whether another major outage will occur, but when. And what if next time, instead of lasting a few hours, it lasted several days? Could our hyperconnected world survive?

One thing is certain: until we diversify the foundations of the internet, we’ll remain at the mercy of a simple technical error in a corner of Virginia. Food for thought before posting your next Snapchat selfie, isn’t it?

What about you—did you notice this outage on Monday? How did it affect you? Do you think we should develop alternatives to American cloud giants? Share your experience and thoughts in the comments!