Amazon Says Drone Strikes Damaged 3 AWS Sites in the UAE and Bahrain

Cloud outages can feel like a power flicker, until your bank app won't load and flights stop updating. On Sunday, March 2, 2026, Amazon Web Services (AWS) said drone strikes damaged three facilities tied to its UAE and Bahrain regions.

That matters because cloud services quietly run everyday life, from payments and retail checkouts to government portals. Below is a clear breakdown of what AWS says was hit, what customers experienced, and what this incident means for cloud security and reliability.

What happened to Amazon's UAE and Bahrain facilities, in plain language

Aerial view of a modern data center facility in a Middle Eastern desert landscape, showing structural damage from a drone strike with collapsed sections, rising smoke from fires, water puddles, and surrounding sand dunes under a clear blue sky with harsh sunlight. An artist's depiction of a damaged data center campus area, created with AI.

AWS says two data center sites in the UAE (its ME-CENTRAL-1 region) were directly struck by drones. AWS also said one facility in Bahrain (ME-SOUTH-1) was damaged by a nearby strike. In other words, three locations suffered physical impact or spillover damage tied to the same wave of attacks.

According to AWS, the incidents led to structural damage, power disruptions, and fires. Fire suppression efforts then added water damage inside parts of the buildings. AWS also warned that restoring service would take time because crews needed to stabilize facilities, re-establish power, and confirm network connectivity. A detailed public map of the affected buildings has not been released.

For the most widely reported summary of AWS's statement, see Reuters coverage of the damaged AWS facilities.

Where the strikes landed and what kind of damage was reported

ME-CENTRAL-1 (UAE) took the direct hits, while ME-SOUTH-1 (Bahrain) faced nearby impact. Even "nearby" damage can still hurt a data center, because power feeds, cooling, and fiber links often connect across rooms or adjacent structures. Once those shared links wobble, services can follow.

Which AWS services were disrupted, and why outages spread fast

In a modern Middle Eastern business office, four workers gather around a conference table with laptops and monitors displaying outage error icons, their frustrated expressions illuminated by dim lighting from power issues, with a city skyline visible through the windows in realistic style. Teams reacting to cloud service disruptions, created with AI.

Customers in the region reported elevated error rates and degraded availability, meaning apps worked sometimes, then failed, then partially recovered. AWS said many services saw issues, including familiar building blocks like EC2, S3, DynamoDB, Lambda, RDS, CloudWatch, and the Management Console.

Outages spread quickly because cloud services share the same physical backbone. If a facility loses stable power, cooling, or core networking, lots of "separate" products can stumble at once. Redundancy helps, but physical damage can limit what backup systems can do until repairs and inspections finish.

The knock-on effect for businesses in the Middle East

Banks, airlines, retailers, and government services can run on the same cloud region. As a result, concentration risk becomes real. Many unrelated organizations can feel the same outage at the same time, like a traffic jam caused by one blocked tunnel.

Amazon's response, customer next steps, and the bigger security question

AWS said it worked to protect staff, coordinate with local authorities, and restore power and connectivity. It also advised customers to rely on backups and shift workloads to other regions where possible. AWS indicated full recovery could take at least a day, since this wasn't a software bug, it was physical damage.

This disruption also lands amid escalating regional tensions, with news reporting attributing retaliatory strikes to Iranian forces. Whatever the wider context, the takeaway is simple: cloud infrastructure is still concrete, cables, and generators.

When a cloud region is hit physically, recovery depends on repairs, safety checks, and restored utilities, not just rerouting traffic.

What "resilience" looks like after a physical attack on cloud infrastructure

Multi-region backups for critical data and configs
Tested failover plans (not just written runbooks)
Offline copies of the most important records
Clear incident comms for customers, partners, and regulators

Conclusion

Three AWS sites in the UAE and Bahrain were damaged, and real service disruption followed. The cloud can feel invisible, yet it lives in buildings that can be targeted. Over the next few days, watch for official damage assessments, stability updates, and whether companies push for stronger regional redundancy. For most teams, the best move is calm planning, plus redundant design for rare days like this.