For an industry built on the promise of decentralisation, Web3 continues to lean heavily on centralised internet plumbing. It’s a contradiction that builders acknowledge but rarely confront, until the system cracks in real time. That’s exactly what happened this week when a Cloudflare outage rippled across the internet, dragging multiple blockchain platforms into hours of downtime.
Blockchains may be distributed, but their interfaces, APIs, and data pipelines still rely on traditional Web2 giants for speed, security, and global reach. When one of those giants falters, the entire ecosystem is reminded that decentralisation is not just a technical goal, it’s operational insurance.
What really went wrong?
Cloudflare’s own post-mortem reveals that the outage began at 11:20 UTC, triggered by a permissions change in their ClickHouse database. This caused a core “feature” configuration file used by their Bot Management system to double in size. That oversized file broke through an internal limit (set at 200 features), making the proxy software panic and return large volumes of HTTP 5xx errors.
Engineers eventually rolled back to a previous version of the file, and by 14:30 UTC, core traffic began to recover , with full recovery reached by 17:06 UTC.
A Long-Standing Tension
Web3’s early builders prioritised speed and accessibility over infrastructural purity. Cloudflare, AWS, Google Cloud, and similar providers offered unmatched reliability and agility. But that convenience came at a cost: when these centralised layers fail, the “decentralised” user experience collapses instantly.
History has shown this repeatedly, EtherWallet’s DNS hijack in 2018, Infura and Alchemy outages in 2020, and several CDN disruptions that knocked explorers and wallets offline. These incidents reinforce a harsh truth: decentralisation is only as strong as the weakest centralised component in the stack.
The Cloudflare disruption exposed how ingrained these dependencies still are. Blockchain may be decentralised, but the pathways through which users interact with it remain surprisingly narrow. Most Web3 companies run front-end UIs, API gateways, analytics, and node proxies on a handful of large providers. Once Cloudflare’s routing layer choked, those single points of failure rippled across crypto platforms globally. It is a reminder that decentralisation is not just an on-chain design philosophy; it must be an operational and infrastructural discipline.
A Global Ripple Effect
The Cloudflare disruption underscored how dependent Web3 platforms still are. Though blockchains themselves remained online, front ends for platforms like Arbiscan, DefiLlama, and other explorers went down. This exposed a critical vulnerability: decentralised logic paired with centralised interfaces creates a structural bottleneck.
As analysts point out, relying on a single CDN or security provider becomes a systemic point of failure, not just a technical risk.
Why do these failures keep happening
This outage is neither unique nor unexpected. Web3’s reliance on centralised infrastructure for DNS, traffic routing, and content delivery introduces fragility. Most operators have not built redundancy off-chain, even though on-chain decentralisation is strong.
As Jimmy Song notes, subtle assumptions, like undocumented configuration file limits, can trigger cascading failures when not properly validated.
Cloudflare says it is tightening its systems with improved validation, global kill switches, and hardened configuration pipelines. But true resilience for Web3 requires more than vendor patches. It demands a shift from the “single cloud provider” mindset to multi-cloud operations, decentralised RPC networks, and self-hosted fallback infrastructure.
Way Forward
Web3 builders are increasingly aware that decentralisation must extend beyond consensus mechanisms. Redundancy across multiple cloud providers, decentralised RPC networks, peer-to-peer content delivery, and self-hosted infrastructure are emerging as credible strategies.
Some teams are already experimenting with hybrid setups, mixing centralised performance with decentralised fallback layers, to limit disruption when global outages strike. It is not about abandoning cloud services entirely, but about designing systems that don’t collapse when a single provider fails.
Closing Thoughts
This outage is a stress test for Web3’s maturity. Blockchain cannot fulfil its decentralisation promise if the surrounding infrastructure remains centralised and fragile. As the industry scales, eliminating single points of failure will be essential. The next evolution of Web3 will belong to teams that treat infrastructure with the same seriousness as consensus.
True decentralisation won’t happen by accident; it will happen by deliberate design.
