NexusNation
NexusNationSYSTEMSDEV
← All incidents

Production Incident Atlas

AWS UAE outage (2026): two availability zones damaged

Physical damage and power loss impaired two availability zones in AWS’s UAE region. AWS reported cascading errors in services including S3, DynamoDB, EC2, Lambda, and RDS; recovery remains incomplete.

AWS Middle East (UAE) Region (ME-CENTRAL-1)Software & infrastructureOther failure modeCritical severityAvailability

What happened

Incident timeline

10 documented events

  1. AWS later reported that objects struck a data center in mec1-az2, causing sparks and fire. The fire department shut off power to the facility and generators while responding.

  2. AWS said it was investigating issues with services in the UAE region.

  3. AWS identified a localized power issue in mec1-az2, said existing instances there would be affected, and began routing requests away from the zone.

  4. AWS described the facility impact and power shutdown. It said other UAE zones were functioning normally at that time and recommended using unaffected zones or restoring from backups.

  5. AWS reported a localized power issue in a second UAE Availability Zone, mec1-az3. It also reported EC2 launch errors and increased errors and latency in S3 and DynamoDB.

  6. With two zones impaired, AWS reported significant S3 and DynamoDB errors, EC2 launch failures, and high failure rates for S3 data transfers.

  7. AWS said physical impacts from drone strikes affected two UAE facilities and one Bahrain facility. It reported structural damage, disrupted power, and additional water damage in some cases.

  8. AWS said S3 and DynamoDB recovery was still in progress, dependent services remained affected, and EC2 launches were throttled. It recommended moving workloads and restoring from remote backups in another region.

  9. AWS said the UAE region could not reliably support customer applications, recommended migrating accessible resources and restoring inaccessible resources from remote backups, and said restoration was expected to take several months.

  10. AWS said it could not restore access to resources and data hosted exclusively in mec1-az2. Recovery work continued for resources in the other affected zones.

Detection

AWS began investigating service issues in ME-CENTRAL-1 at 4:51 AM PST on March 1. At 5:19 AM, it reported a localized power issue affecting one Availability Zone and said it was routing requests away from the zone.

Recovery

AWS reported some services recovering after traffic was shifted away from affected zones, but customer resources in the affected zones had a longer recovery path. By March 3, S3 and DynamoDB remained impaired and EC2 launches were throttled. AWS later advised migration to other regions. In its September 15 update, AWS said access to resources and data hosted exclusively in mec1-az2 could not be restored; recovery work continued for other affected resources.

Lessons afterward

AWS advised customers to use disaster-recovery plans, migrate workloads to another region, and restore inaccessible resources from remote backups. Its status updates also show that one zone’s failure and two zones’ impairment had materially different effects on regional services. Treat this as a documented AWS event, not a guarantee that every service or customer was affected identically.

Editorial analysis

What happened, in plain language

AWS’s Middle East (UAE) Region is divided into Availability Zones: separate infrastructure groups within the region. AWS says its regional S3 service is designed to withstand the loss of one zone. In this event, one zone was hit first; a second zone was later impaired. AWS then reported rising errors in S3 and DynamoDB and impacts to services that depended on them.

AWS initially described a localized power problem. In a later update, it said objects struck a data center, causing sparks and fire, and that the fire department shut off power while responding. AWS subsequently attributed physical impacts in the UAE and Bahrain to drone strikes amid the regional conflict. This timeline uses AWS’s own wording and does not independently attribute responsibility.

AWS recommended that customers move workloads to other regions and restore inaccessible resources from remote backups. In its September 15 update, AWS said it could not restore access to resources and data hosted exclusively in UAE Availability Zone mec1-az2, while work continued on other affected infrastructure. That statement does not say all UAE-region data was lost.

Editorial severity: Critical, based on the prolonged, multi-service regional impact and AWS’s September statement about resources isolated to one zone. AWS did not assign this editorial label.

Last reviewed 10/9/2026 by Nexus Lead Architect.

Change notes

  • 10/9/2026: Initial publication using AWS public status reports and official Amazon updates.

Corrections or additional primary evidence? Contact the editorial team.