AWS Power and Cooling Failure Disrupts US-EAST-1 Cloud Workloads
A power and cooling failure in AWS’s US-EAST-1 region disrupted EC2 and EBS workloads, again exposing the blast radius of cloud concentration.
AWS Power and Cooling Failure Disrupts US-EAST-1 Cloud Workloads
A power and cooling incident in Northern Virginia knocked some Amazon Web Services customers into a familiar kind of cloud trouble: impaired EC2 instances and EBS volumes in US-EAST-1. The Register reported that AWS traced the disruption to the use1-az4 availability zone, where a thermal event followed a loss of power on impacted hardware. The outage matters because US-EAST-1 remains one of the cloud’s busiest and most dependency-heavy regions.
According to AWS’s service health updates, the company first flagged problems on May 7 at 5:25 PM Pacific time and later warned that EC2 instances and EBS volumes on affected hardware were hit by the power loss. The Register said AWS also cautioned that other services depending on those EC2 and EBS resources could see impairments, including elevated error rates and latency. That dependency chain is why a single availability-zone problem can still ripple into applications that look, from the outside, like they run across a whole cloud region.
The incident was not just a dashboard footnote. The Register quoted a reader saying servers in the region had become inaccessible while the AWS status page itself was misbehaving. AWS said in its updates, as cited by The Register, that it was shifting traffic away from the affected availability zone and advised customers to move workloads into other US-EAST-1 zones, though provisioning could take longer than usual. For operators under pressure, that warning cuts both ways: failover is possible, but only if spare capacity and automation are ready before the incident starts.
By 10:11 PM Pacific time, The Register reported that AWS had brought additional cooling capacity online and recovered some affected racks. AWS said it was working to recover more racks in a controlled way, a cautious pace that makes sense when the underlying problem is physical infrastructure rather than a software toggle. Cooling failures can create a slower recovery curve because providers have to protect equipment while restoring service, not merely restart a process or roll back a deployment.
US-EAST-1 matters because so many companies still treat it as the default place for critical cloud infrastructure. The Register noted that the same region was tied to major AWS outages in 2021 and October 2025, while AWS executives have argued that the region is not uniquely fragile but operates at unusual scale. That distinction won’t comfort teams whose failover plans depend on capacity appearing exactly when everyone else needs it too.
Photo by Albert Stoynov on Unsplash
