Recap for Monday, 20 October 2025 It's a normal Monday morning. You just want to check your Duolingo street, quickly gamble a round of Fortnite or pay at Starbucks with the app. But then: Nothing works anymore. Welcome to what is probably the biggest internet outage of the year!
What actually happened?
Around 3 o'clock in the morning (ET) for us in Europe it was already bright day, it has caught Amazon Web Services (AWS) violently. And when we say ‘heavy’, we mean: More than 1,000 services offline at the same time. It's not a joke. The internet had a collective heart attack.
The epicenter? The AWS data center in Northern Virginia, affectionately known as the US-EAST-1 region. Sounds unspectacular, but it is one of the most important digital hubs in the world. When the DNS resolution (i.e. the telephone book of the Internet) gave up the spirit there, it went all at once: DynamoDB – one of AWS’s central databases – closed, followed by EC2 (virtual servers) and S3 (memory).
The Hall of Shame: Who was all affected?
The list reads like a who’s who of the digital world:
Social media messaging:
- Snapchat (up, there go the Streaks flute!)
- Signal (of all things the ‘safe’ alternative)
Gaming:
- Fortnite
- Roblox
- Pokémon GO
- Duolingo (RIP your learning streaks?)
Financial service providers:
- Robinhood
- Venmo
- Coinbase (crypto trader in panic mode)
And much more:
- Amazon itself (the irony!)
- Alexa & Ring
- United and Delta Airlines
- Canva
- ChatGPT
- Starbucks app
- Adobe Creative Cloud
- The New York Times Games
Even universities like Rutgers reported problems with Canvas, Zoom and Grammarly. In short: Pure chaos.
Why is a failure so dramatic?
Here's where it gets interesting: AWS is not just one cloud provider among many. At about 30% Amazon's market share is the invisible framework that carries much of the Internet. If AWS sneezes, half the internet gets a cold.
The problem is the central dependency. Thousands of companies rent from AWS using their computing power, storage and databases. Sounds practical, it is. Until the moment when exactly this central point fails. Then everything stops.
Experts estimate that the financial damage of this failure in the Hundreds of billions of dollars could go. Not because AWS itself loses so much, but because all the affected companies couldn't do business for hours.
What was the cause?
Amazon has stated that an Software update for DynamoDB It contained an error. This error then triggered a chain reaction. After all: It was not a cyberattack or hacker attack, but ‘only’ human error or a technical error.
A security expert put it in a nutshell: “Mostly it is not a military operation or espionage. Most of the time, it is simply human error.”
The funny side: Elon had to add his mustard, of course.
While half the internet was on the ground, Elon Musk could not hold back again. His comment on a post about the failure? “Not us.”
X (formerly Twitter) actually continued to run – presumably because Musk reduced reliance on AWS after the takeover. Of course, he took the opportunity to promote X's new chat function as an alternative to Signal. Marketing at its finest!
Timeline: From failure to (partial) solution
- ~3:00 ET: The First Mistakes Arrive
- ~7:00 GMT: The failure reaches its peak
- In the morning: Amazon is working on the solution in “multiple parallel ways”
- ~13:00 ET: The main problems have been fixed
- In the evening: Most services are running again, but some users still report delays
The failure lasted about 6 hours in its full intensity. For many of us, it probably felt longer.
What do we learn from this?
This failure has made us all painfully aware again: The Internet is more fragile than we think. The entire digital infrastructure depends on a few major providers – AWS, Microsoft Azure, Google Cloud. When one of these giants stumbles, a huge part of the Internet crashes with it.
The solution? diversification. Experts have long called for companies to distribute their services across multiple cloud providers. This is called multi-cloud strategies in technical jargon. Sounds expensive and complicated? It is, too. But probably still cheaper than hours of outages.
The big question remains: Why is there no immediate redundancy? Why can such a small, localized problem have such a global impact? These are questions that the tech industry urgently needs to address.
TL:DR
The October 20, 2025 AWS outage will go down in the history books of the biggest Internet breakdowns. He has shown how dependent we are on a few tech giants and how quickly our digital lives can come to a standstill.
For most of us, it was an annoying Monday with Duolingo strike loss and missed Fortnite sessions. For the tech world, it was another wake-up call that we urgently need to think about alternatives to the current centralized cloud infrastructure.
Until then, it reads: Keep your fingers crossed that AWS will be more careful with the next software update. And maybe we should all consider whether we still need some analog backup plans. You know, just in case.
By the way, your duolingo strikes haven't been reset. Duolingo was gracious. Unlike AWS. ⁇
EN 
DE 
ES 
FR 
IT