AWS Suffers Total Loss in Bahrain, Abandons Recovery for Part of UAE

Amazon Web Services has announced that it cannot restore access to resources and data stored across multiple Availability Zones in the Middle East. The cause was physical damage sustained during the war between the United States and Iran. The affected scope is the entire Bahrain Region and one zone in the UAE Region. Reporting by Dan Robinson of The Register described the damage as exceeding the assumptions of regional redundancy.

The company explained the situation in an update to the AWS Health Dashboard. A determination of permanent loss is rare even among major cloud providers. The incident has once again highlighted the importance of multi-region design on the customer side.

Confirmation of Total Loss in the Bahrain Region

AWS has confirmed that it is unable to connect to resources and data stored exclusively in the Bahrain Region, me-south-1. Compute infrastructure and stored data that existed only in that Region remain unavailable. There is no prospect of recovery.

The company’s explanation emphasizes that the damage spanned multiple Availability Zones. It says the damage exceeded what regional and multi-AZ services are designed to withstand. As a result of its assessment, it decided to abandon recovery.

The update states as follows.

The damage to our infrastructure spanned multiple Availability Zones and exceeded what our regional and multi-AZ services are designed to withstand,

After a thorough assessment, we have determined that we are unable to restore access to the resources and data hosted exclusively in this Region.

The company says it reached the decision after a thorough assessment. The key point is that this was not a single equipment failure, but damage to the entire regional infrastructure. It is a case where the premises of cloud availability design have collapsed.

Timeline in Bahrain from March to April

The initial damage occurred in March. The first Availability Zone in Bahrain was damaged. AWS advised customers to migrate to other Regions. Many customers are said to have followed that advice and evacuated.

Further attacks occurred in April. The failure of a second Availability Zone rendered the entire Region unavailable. The gradual expansion of damage made recovery difficult. Customers who migrated at the time of the initial warning and those who remained saw starkly different outcomes.

According to the company, most customers had completed migration before the full outage. Those left behind were customers who had kept resources exclusively in that Region. The danger of architectures without replication or fallback destinations became reality.

The published timeline shows the prolonged nature of the damage. Damage accumulated from March into April. It was not a single outage, but destruction caused by sustained attacks. It illustrates the difficulty of maintaining infrastructure in wartime.

Damage in the UAE Region

Recovery has also been abandoned for one zone in the UAE Region. The affected zone is mec1-az2. Access to resources and data stored exclusively in that zone cannot be restored. The UAE Region consists of three zones, one of which has been permanently lost.

Two facilities in the country were reportedly hit by drone attacks in March. The damage occurred when Iran retaliated against attacks by the United States and Israel. Physical damage to the facilities was the direct cause. It is a case of military action extending to civilian computing infrastructure.

Recovery work continues in the other two zones, mec1-az1 and mec1-az3. AWS says it is working to recover resources within the Region. Some resources are still in the process of being restored. Rather than a total loss, the damage may remain a partial loss.

According to the company, most customers were able to resume operations in other Regions. They rebuilt through restoration from backup copies or by copying accessible data. Whether replication had been set up in advance determined the speed of recovery. Differences in preparedness at the design stage were reflected in the outcomes.

Limits of Redundancy and a Turning Point in

Design Philosophy In AWS terminology, a Region is a geographic area containing multiple isolated Availability Zones. Each zone consists of one or more independent data centers. Power, cooling, and networks are kept independent. The structure is designed to withstand the loss of a single site.

The company has long recommended distributing resources across multiple zones. That allows operations to continue despite the failure of a single site. This incident exceeded that assumption. Multi-AZ distribution cannot cope with the failure of an entire Region.

In Bahrain, damage to multiple zones occurred simultaneously. It could not be absorbed even by the resilience design of regional services. Redundancy designed for natural disasters or equipment failures was insufficient. Wartime destruction exceeded the design assumptions.

Customers have no choice but to move toward multi-region architectures. Critical compute infrastructure must also be replicated in separate Regions. Securing fallback destinations and pre-validating failover procedures are essential. Increased cost and operational load are unavoidable, but the cost of permanent loss outweighs them.

Current State of Recovery Support and Overall

Customer Impact AWS says its support teams are assisting with migration. It has a system in place to help move workloads to other Regions. It says it is also handling individual recovery consultations. The Register sought further comment, but the company declined to provide anything beyond the contents of the Health Dashboard.

What divided the impact was whether resources were stored exclusively in the affected areas. Customers with copies in other Regions were able to resume. Resources that existed only in Bahrain will not return. Data loss and loss of compute infrastructure occurred simultaneously. The thoroughness of distribution determined the success or failure of business continuity plans.

In development teams, automating migration work and preparing runbooks will become challenges. Some see the use of development assistance discussed in Complete Comparison of AI Code Assistants 2026: How to Choose Between Copilot, Cursor, and Tabnine as also useful for streamlining recovery work. The ability to switch infrastructure quickly is directly linked to competitiveness. Training in normal times will be important.

For enterprise customers, it is necessary to review contracts and the scope of responsibility. They need to distinguish between the provider’s scope of protection and the customer’s responsibility for replication. Compensation and support terms in the event of permanent loss will also become issues. Collaboration between procurement and engineering departments is essential.

Geopolitical Risks Facing Cloud Infrastructure

This incident marks the realization of geopolitical risk. Conventional threat models focused on power outages, fires, and earthquakes. The simultaneous destruction of multiple facilities by military attack was beyond expectations. It showed that computing infrastructure can become a strategic target.

The Middle East is highly important as a communications hub. Storing data within the region had the advantage of reducing latency. On the other hand, the physical danger in times of conflict is high. The balance between convenience and safety has collapsed. The very criteria for Region selection will be reconsidered.

The impact on regulated industries such as government, finance, and healthcare is significant. It is difficult to reconcile domestic storage requirements with ensuring safety. Whether overseas replication is permitted depends on regulations. Alignment between policy and technology is being called into question.

Providers also need to strengthen facility distribution and protection. Undergrounding and distributed placement alone have limits. Under continued attacks, maintenance personnel cannot easily access sites. Complete defense is difficult to achieve. Multi-region design on the customer side becomes the last line of defense.

Editorial Opinion

In the short term, migration to other Regions is expected to accelerate among customers with infrastructure in the Middle East. Designs that diversify backup storage locations are likely to be reviewed. Increased support requests could cause delays in migration work. Confirmation of pricing and compensation terms is also likely to proceed.

In the long term, geopolitical stability will carry greater weight in Region selection. Multi-region architectures that avoid dependence on a single Region are likely to become standard. For regulated industries such as government and finance, reconciling this with domestic storage requirements is likely to become a challenge. Recovery plans that assume physical destruction will be required.

The remaining question is how much destruction providers should anticipate. Contract terms premised on permanent loss appear to have been reasonable. The boundary between customers’ responsibility for replication and providers’ responsibility for protection is being questioned. The manner of verifiable disclosure also needs discussion.

References

• “AWS says wartime damage means some Middle East cloud resources are gone for good”, by Dan Robinson — The Register, 2026-09-16T12:11:56.000Z (ARR)
• Source URL: https://www.theregister.com/off-prem/2026/09/16/aws-says-wartime-damage-means-some-middle-east-cloud-resources-are-gone-for-good/5296830