What the AWS outage means for web3

The AWS outage is a reminder of how fragile modern digital systems have become when so much depends on a single provider. A fault in a domain name system (DNS) record triggered cascading failures across Amazon Web Services infrastructure, taking down over 14,000 websites and causing more than $1 billion in losses within just two hours, including Coinbase, MetaMask, and Robinhood among the casualties. When the functions came back online, data synchronisation caused a domino effect of further disruption.

AWS tried to safeguard this with multiple, localised points of failure, but failed to account for a regional DNS disruption. As the cloud provider for more than 90% of Fortune 100 companies, this event showed that even the largest centralised providers cannot fully guard against these risks of centralisation. Exposing what happens when we over-rely on essential infrastructure that lacks the correct safeguards in place.

While private companies scrambled to restore access, governments are building vital public infrastructure like AI and national digital ID on the same flawed assumptions and often hosted on the same providers. When those fail, verification, payments, and access to essential services can all halt at once.  The incident exposed a simple truth: When essential infrastructure is concentrated in one place, resilience disappears, and even the largest cloud providers cannot fully guard against the risks of centralization.

The question should not be whether another major outage will happen again, but when, and how to design for such concentrated risk. Distributed and decentralized infrastructure, verifiable credentials, and trust registries are some technologies deployed across the world to ensure the lowest possible probability of centralized failure. Allowing data to stay in departmental silos while verification can be decoupled from centralised databases to make identity private by default.

Centralization is a structural risk, not a technical error

The AWS outage revealed the deeper flaw in the foundations upon which the modern internet was built. Much of the internet’s backbone now depends on a concentrated handful of cloud, CDN & DNS providers. Three hyperscalers, AWS, Microsoft Azure, and Google Cloud, now control almost 70% of the global cloud infrastructure market. When one of these fails, just as seen recently, the effects cascade far beyond their own networks. To put it simply, our systems were designed for uninterrupted continuity like lean manufacturing plants — streamlined for outputs but brittle when the manufacturing line stops.

This problem stretches across all industries; public institutions also use the same cloud providers to run ever-increasing national digital identity frameworks, payments, and AI models. What started out as a pursuit for seamless scalability has evolved into a dependency that binds critical systems together in an ever-increasing manner.

When so much of daily life rests on a few platforms, failures can no longer be treated as isolated events. They become systemic risks that ripple through everything connected to them.

The false comfort of scale

Centralization works best when it’s easy. Promising speed, efficiency, and lower costs — who wouldn’t choose these, but they also have their hidden costs lurking. When everything goes through the same backbone, it’s only a matter of time until resilience gives way to fragility.

Centralized systems, even with redundancy built in like AWS has, share the same weak points. One wrong setting, security breach, or other problem, like an all too common DNS fault, can bring whole sectors down at once.

When automation is used, that risk goes up, and when these systems break down, people get locked out of the internet and accounts at best, and their payments, public services, or national identity at worst.

The irony is that the same systems designed to eliminate downtime through automated efficiency have also created an environment where even a moment of failure is no longer acceptable. Each layer of automation adds convenience but removes human oversight, compounding the risk when things go wrong. Until resilience becomes as valued as efficiency, every outage will remind us how little control we truly have.

Distributed systems as the foundation of resilience

Real resilience does not come from adding more backup servers. It comes from removing the need for a single point of control. Decentralized systems do this by design.

In a decentralized network or system, independent nodes share decision-making and verification independently. Making this practical in real systems needs mechanisms that let you verify information without holding the underlying data themselves. Technologies such as verifiable credentials, blockchain-based trust registries, and self-sovereign identity already enforce these principles. Instead of requiring data to be stored and checked in a single system, verification can be performed cryptographically across a distributed network.

This means that data can stay where it already lives, inside departmental or organizational systems that operate it, while verification can be decoupled from its centralization. Verification can happen without pulling or exposing the underlying data that’s centralized in one place, removing the biggest single point of failure while maintaining data where it belongs, with the individual or entity that owns it. A restructuring of our entire infrastructure is not necessary; rather, it is to lessen the magnitude and effect of our failures.

Applied to AI, this means models could draw verified data from multiple trusted sources without relying on one central hub. Compliance checks in finance could run across distributed systems that verify transactions without passing through every single provider. Overall, resulting in an infrastructure that can fail safely instead of catastrophically

Critical infrastructure doesn’t have to be this centralized

Critical infrastructure does not have to live inside a big centralized database that expands what corporations and governments know while creating an ever-enticing honeypot for fraudsters and critical outages. We do not need one digital backbone to make essential systems efficiently work together. Interoperability can be achieved through distributed systems, open standards, and verifiable data that coordinate data without concentrating control.

We already have the ability — blockchain and web3 technologies allow sectors like finance, identity, and AI to coordinate securely across independent networks instead of depending on one central provider.

Infrastructure designed this way enforces boundaries automatically, by limiting what data is shared, how it is stored, and who can access it, while still accommodating collaboration between systems that were never meant to operate together. Efficiency comes from distribution, not dependence.

The real cost is not just in case critical centralized services go offline once again, but whether the next generation of digital infrastructure for AI, National Identity, and cloud-based servers can be built to withstand failure. By adopting distributed verification and blockchain-based trust frameworks, we can create systems that strengthen resilience instead of replacing its current fragility.