A race condition in Amazon's DynamoDB DNS management caused a major outage, disrupting services and estimated damage in the hundreds of billions. The error began on October 19, 2025, when the system left an empty DNS record due to a timing conflict between DNS planners and enactors. This led to widespread failures, impacting EC2 launches and other AWS services. Amazon has suspended the affected automation until fixes are implemented.
https://www.theregister.com/2025/10/23/amazon_outage_postmortem/