AWS cannot restore its Bahrain region or one UAE zone six months after the strikes

Amazon Web Services has told customers it cannot restore access to its cloud facility in Bahrain or one of its three data hosting zones in the UAE, six months after drone strikes hit its sites in the Gulf.

The explanation matters because it shows the limits of the resilience built into cloud infrastructure.

“The damage to our infrastructure spanned multiple Availability Zones and exceeded what our regional and multi-AZ services are designed to withstand,” the company said.

Reuters reported the statement on Tuesday. Availability Zones are a key part of how cloud providers offer reliability. They are separate facilities with their own power and cooling systems, but close enough to operate as part of the same region.

The idea is that if one facility fails, the others keep running. Customers can also spread their systems across several zones to reduce the risk of an outage.

Amazon is now saying that the damage went beyond what this setup was designed to handle. The problem was not the failure of one facility, but damage affecting several parts of the same cloud region.

The problems began in early March, when two of Amazon’s UAE data centres were hit directly, and the Bahrain site was damaged by strikes nearby. Amazon reported structural damage, problems with power delivery and, in some areas, water damage from its own fire suppression systems.

Further disruption followed in Bahrain later that month. Amazon did not say whether the building itself had been hit. By the end of April, around 60 services in the region were still down.

In May, Amazon said restoration would take several months. It urged customers to “migrate all accessible resources to other Regions” and rebuild the rest from off-site backups. It also stopped billing for both regions.

Six months later, Amazon says it has helped Bahrain customers move their operations elsewhere. That is not the same as restoring the affected region.

There is still no public answer on whether any customer data has been permanently lost. Amazon has not said whether the data in the affected zones can be recovered, given a new restoration timetable, or explained whether the damaged buildings can be repaired or will need to be replaced.

For customers whose only backups were stored in those zones, that leaves an important question unanswered.

The Gulf was expected to become a major centre for the AI build-out, helped by cheap power and large amounts of sovereign investment. But that assumption has come under pressure this year.

We have followed the region’s paused investment decisions and rising insurance costs, the UAE’s decision to redesign Stargate underground after Iran named it as a target, and the argument that cheap cloud depends on stability as much as engineering.

Six months into the outage, this is no longer just a temporary incident. Cloud infrastructure is built around the assumption that a damaged facility or region can eventually be recovered.

In Bahrain and the UAE, customers now have a case where that recovery has not happened, raising questions for cloud providers, insurers and regulators about how much resilience these systems really provide when an entire region is affected.