ChatGPT is also affected. Amazon's server malfunction causes half of the internet to crash.

When Amazon sneezes, half of the Internet catches a cold.

Due to the outage of Amazon's AWS servers, a large number of Internet services were forced to shut down, and ChatGPT was also affected.

The outage occurred in the us - east - 1 region in the eastern United States, which is the core part of AWS's global services.

According to data from the outage tracking platform Downdetector, more than 6.5 million user outage reports were received in total on that day.

Not a complete list

This AWS incident also caused the services of social platforms including Reddit to crash, and people almost had no way to vent their complaints.

Moreover, even AWS's own customer support ticketing system went down, so there was no way to report an error.

Fortunately, Musk's X doesn't use AWS, so it wasn't affected, which gave netizens the opportunity to discuss this AWS outage.

Some netizens used emojis to joke that Musk was the biggest winner in this incident.

But jokes aside, those affected by this incident may not be able to laugh at all...

Amazon's service outage affects all walks of life

How wide - reaching was Amazon's outage this time? Let's first look at the situation of developers.

Docker, an important application container engine, was found by developers to be completely paralyzed, with multiple core services interrupted.

In addition to Docker, another important development tool, npm, also had the same problem, and the popular AI programming tools Cursor and Vercel were not spared either.

Besides developers, other workers were also affected. The video conferencing software Zoom and the office platform Slack (similar to OpenAI's) both crashed.

Interestingly, it is said that Amazon started using Zoom for internal meetings this year. As a result, without Zoom, AWS couldn't hold online meetings, but Zoom couldn't be used without AWS being fixed...

Since work couldn't be done, could people just slack off?

No, because the game platforms Epic and Sony's PlayStation, as well as social platforms like Reddit and streaming video websites such as Netflix, Disney+, and Max also went down.

What about studying? Sorry, Duolingo crashed, and students couldn't log in to the homework platform Canvas.

In terms of AI tools, the most famous ChatGPT was also considered to be affected, and the well - known AI search tool Perplexity was also on the list.

Moreover, the impact spread from the online world to the offline world. Ride - hailing apps crashed, and McDonald's and Starbucks apps went down, making it impossible to hail a ride or order food in daily life.

Airlines were also affected. United Airlines and Delta Air Lines couldn't handle check - ins and baggage check - ins for passengers.

Worse than not being able to board a plane was not being able to get off a plane. Some netizens on Xiaohongshu shared their experiences, saying that after a Delta Air Lines flight landed, it couldn't dock, and all the passengers had to wait on the runway. The captain announced that the reason was Amazon's outage.

Even worse, smart lock users couldn't unlock their locks due to the network issue, but the alarm function of the locks worked, which attracted the police.

Of course, other Amazon smart home devices like Alexa went on a full - scale strike due to the lack of network services.

The incident even affected the UK on the other side of the Atlantic.

For example, the Premier League officially announced that due to the AWS outage, the semi - automatic offside system malfunctioned during the match between West Ham United and Brentford. If necessary, the traditional manual line - drawing method would be used to assist in the offside determination.

Of course, the list of "victims" is still very long...

Why is the Internet so vulnerable?

The problem this time occurred in AWS's us - east - 1 server, and the direct cause was a DNS (Domain Name System) resolution issue.

AWS stated that the problem originated from a subsystem within the EC2 internal network, which is used to monitor and manage the health status of network load balancers.

The abnormality of this monitoring system led to unstable network connections, database access delays, and failures of some API requests. Under the chain reaction, many core cloud services experienced large - scale outages.

us - east - 1 is the earliest - established region of AWS, with the longest - standing infrastructure and the most diverse service offerings. A large number of enterprises have deployed their core services in us - east - 1.

us - east - 1 is not only the base camp for computing/storage resources but also the centralized hosting area for many global control - plane services. This means that even if other regions are operating normally locally, control - layer operations such as deployment, authentication, and permission changes still rely on us - east - 1.

This is why us - east - 1 is different from other AWS servers, and its outage can affect the whole world.

Moreover, this is not the first time that us - east - 1 has crashed. Similar "large - scale paralysis" incidents have occurred in the past few years (2020, 2021, 2023), and each time, a large number of online services were affected.

Outages are inevitable, but the wide - reaching impact of this outage reveals the problems in the Internet -

Large - scale cloud services like Amazon's have indeed improved global network security and stability. However, the standardized and centralized nature of such services means that any small glitch could potentially lead to catastrophic consequences.

To avoid such situations, website developers need to set up a resilient mechanism.

Some netizens on Hacker News said that their static sites were connected to multiple regions through CloudFront and didn't rely solely on us - east - 1, so they were not affected by this incident.

Moreover, this native multi - region, fail - over deployment method is not technically complex, and the cost will not increase significantly.

It's time for network service developers to re - evaluate their deployment strategies.

Reference links:

[1]https://www.theverge.com/news/802486/aws-outage-alexa-fortnite-snapchat-offline#comments

[2]https://www.wired.com/story/what-that-huge-aws-outage-reveals-about-the-internet/

[3]https://www.businessinsider.com/ring-starbucks-alexa-app-down-aws-outage-issues-2025-10

[4]https://www.cnbc.com/2025/10/20/amazon-web-services-outage-hits-airline-websites-other-major-sites.html

[5]https://news.ycombinator.com/item?id=45640838

This article is from the WeChat official account "QbitAI", author: Keleixi. Republished by 36Kr with permission.