AWS Outage Reignites Debate Over Internet Centralization

Monday October 20 was marked by a memorable Amazon Web Services outageaka AWS, a pillar of web infrastructure whose absence has had far-reaching repercussions on a global scale. Amazon provided some information on the origin of this incident, explaining that it was a problem of DNS resolution.

The DNS, you Domain Name Systemis a crucial link in the chain that allows users to access websites; it can be thought of as a large Internet directory that maps each domain name, such as journaldugeek.com, to its IP address.

When an Internet user types an address into their browser, a request is sent to resolve the domain name: the DNS system then translates this readable name into a unique IP address, allowing the browser to locate the corresponding server so that the client can access its content. According to Amazon, it was this translation system that stopped working on Monday, preventing users from accessing the servers hosting the content even if they were working perfectly.

A few hours after the first reports, Amazon managed to correct the problem. But this malfunction still reminded all operators and Internet users to what extent the DNS and the protocols that allow it to function are essential to the proper functioning of the Internet. Without them, even the most robust infrastructures become inaccessible, paralyzing a large part of the web.

A colossus with feet of clay

At the same time, this outage also highlighted the risks associated with centralization of networks on a global scale. As of the first quarter of 2025, AWS hosted or provided infrastructure for a considerable portion of the global web, with approximately a third of the cloud infrastructure market. And when one of these behemoths falls ill, entire sections of the Internet collapse with it, as we saw on Monday.

When dozens of very popular platforms, such as communications services (Signal, Zoom…), artificial intelligence services like Perplexity, video games like Fortnite and money transfer systems like Venmo suddenly become inaccessible, it becomes difficult to ignore the implications of this centralization for the resilience of the Internet, which plays a vital role in our society. It is for this reason in particular that many specialists are calling for diversification of cloud providers and increased decentralization of critical services, in order to limit the impact of an isolated outage.

But in practice, this ideal unfortunately risks being difficult to reconcile with the ruthless reality of the market. Cloud giants like AWS, Microsoft Azure or Google Cloud, which invest colossal sums in their infrastructures, are obviously seeking to increase their market share. It is not only an indicator of commercial success, but also a very important strategic lever for these entities which have everything to gain from making themselves indispensable, in particular to attract industry giants and sign extremely profitable contracts.

In this context, will it even be possible to reduce this systemic dependence? This seems relatively unlikely. It will therefore be necessary to knock on wood so that the major players in the industry manage to make all this infrastructure as solid as possible, in order to prevent these incidents from multiplying in the future.