Many of those nodes do not run on hobbyists’ home machines. They run in commercial data centers.
As of November 2026, between 35% and 60% of Ethereum execution-layer nodes are hosted on cloud providers. AWS alone accounts for approximately 35%.
When a cloud provider has a bad day, the effects tend to show up in user-facing products. Wallets and exchanges that depend on cloud-hosted nodes have historically seen outages during cloud failures. Blocks keep getting produced, but people trying to check balances or submit transactions may find the lights off.
There is also some murk around the scope of this event. Remote procedure call (RPC) providers, the services that let apps send requests to the blockchain, reported minor performance issues around mid-November 2026. Those issues have not been tied to widespread Ethereum disruptions. Research covering late October through mid-November 2026 also turned up no verified report of a cloud hardware malfunction affecting Ethereum during that window.
So the reported disruption should be treated as a developing situation. Its full reach has not been established.
The network has been here before
In October 2025, failures in AWS’s DynamoDB service disrupted key Ethereum services for hours. Consensus operations stayed functional throughout.
Each incident has fed a long-running debate within the Ethereum community about infrastructure diversification. A network that markets itself on resilience should not have a large chunk of its access points concentrated in a few corporate data centers.
What this means for Ethereum users and investors
The most important point for traders is the gap between protocol health and practical access. Ethereum’s core consensus has held through past cloud outages. Yet during those windows, users can lose the ability to interact with their assets through popular apps.
There is a practical response available to users. Running a self-hosted node, or keeping access to more than one RPC endpoint, reduces dependence on any single provider.
Three things are worth tracking in the days ahead. The first is whether the affected cloud provider is identified. The second is how long the access problems last and which services were hit. The third is whether this episode, like October 2025 before it, pushes node operators to spread their infrastructure across more providers, or whether the industry returns to business as usual once services come back.
Leave a Reply