Amazon Web Services, the cloud computing arm of Amazon, suffered a significant outage on Thursday after overheating at a data centre in northern Virginia caused a power loss that damaged hardware and disrupted services for an unknown number of customers.
The incident struck the us-east-1 region's use1-az4 availability zone, the oldest and largest in AWS's global network and a facility that handles a substantial share of the world's internet traffic.
AWS said it had made "incremental progress" in restoring cooling systems but acknowledged that recovery was taking longer than anticipated.
Cryptocurrency exchange Coinbase, which relies on AWS infrastructure, warned users they could experience degraded performance as a result of the outage, though it said customer funds remained safe.
CME Group, the Chicago-based derivatives exchange, reported separate technical and latency problems on its CME Direct trading platform on the same day, later saying it had completed essential maintenance and restored access without identifying a specific cause.
AWS said it had brought additional cooling capacity online and shifted traffic away from the affected availability zone for most services.
An availability zone comprises one or more connected physical data centres designed to operate independently within an AWS region, a structure intended to prevent single points of failure.
The company advised customers to move workloads to alternative availability zones during the recovery period.
EC2 virtual server instances and EBS storage volumes hosted on affected hardware were knocked offline by the power loss during the thermal event, AWS said, adding that other services dependent on those resources could also experience impairments.
The incident is the third significant disruption to hit AWS's infrastructure in less than a year.
In October 2025, a DNS resolution failure in the same us-east-1 region took down thousands of websites and services worldwide, including Snapchat, Roblox and DoorDash, prompting renewed scrutiny of the risks created by concentrating critical digital infrastructure in a handful of cloud providers.
A further outage in March 2026, triggered by a physical incident at AWS facilities in the United Arab Emirates, cascaded across multiple regions and disrupted dozens of services.
Related reading
- Meta breaks ground on 32nd data centre as AI buildout accelerates across the United States
- Anthropic seeks European data centre deals, hires dealmaker
- Google Cloud launches Gemini Enterprise Agent Platform as centrepiece of Cloud Next overhaul
Northern Virginia is the most concentrated data centre hub in the world, hosting 663 facilities and routing a disproportionate share of global web traffic through a corridor of roughly 385 acres.
AWS did not disclose what caused the cooling failure or how many customers were directly affected.The recap
- AWS data center overheating disrupted services in northern Virginia.
- Traffic shifted away from the impacted Availability Zone.
- Coinbase reports degraded performance; working to re-enable trading.
