Fire at Amazon Middle East data center causes Claude and ChatGPT downtime

A fire at an Amazon data center affected by tensions in the Middle East has disrupted operations for major technological products like Claude and ChatGPT, hosted by Amazon Web Services (AWS). 

Multiple Availability Zones in the Middle East (ME-CENTRAL-1 and ME-SOUTH-1) are facing severe disruptions after a fire caused by external objects striking a data center led to a total power shutdown.

Amazon Web Services says it will take a while to recover. 

Why is the AWS restoration taking so long?

Amazon Web Services (AWS) is currently facing a significant service disruption in the ME-CENTRAL-1 (UAE) and ME-SOUTH-1 (Bahrain) regions. Official health reports show that the issue began early on Sunday, March 1, 2026, after a localized power issue affected the mec1-az2 and mec1-az3 Availability Zones.

AWS engineers explained that while Amazon S3 is designed to survive the loss of a single zone, the failure of two zones simultaneously has led to “high failure rates for data ingest and egress.” 

AWS is now advising customers to move their data and applications to alternate regions entirely. Major technological infrastructure like Claude, ChatGPT and AWS Bedrock have experienced significant network disruption because they depend heavily on AWS GPU Clusters and API Networking that route global traffic through affected data centers in the Middle East.

The power issue is due to the physical damage caused by external objects on a data center facility in the mec1-az2 zone. The impact occurred on March 1 around 4:30 am and created sparks and a subsequent fire. 

In response, the local fire department ordered a total power shutoff for the facility and its backup generators to safely extinguish the flames.

Restoration, however, will take far longer. AWS has stated that they must wait for formal permission from fire officials to re-energize the building. Once power is restored, the technical teams must perform a “careful assessment of data health.” 

Because the shutdown was abrupt and involved fire, there is a risk of physical damage to storage hardware. AWS has warned that even after connectivity returns, the complete repair of its storage systems will take a while.

AWS confirmed that customers are still seeing increased EC2 API errors and instance launch failures. It is currently impossible to launch new instances in the affected regions, and while existing instances in the unaffected mec1-az1 zone remain running, the overall regional stability is compromised. 

Cryptopolitan mentioned in an earlier report that the US used Anthropic’s Claude during Iran strikes despite tensions that came to a head on Friday when Trump ordered to cut off the tech from government operations completely. 

How have investors reacted to the news?

The Middle East is currently experiencing a period of intense geopolitical instability after reports that at least three people, who were nationals from Pakistan, Nepal, and Bangladesh, were killed in the United Arab Emirates during retaliatory strikes. These attacks followed weekend military operations by the United States and Israel against high-ranking officials in Iran.

President Donald Trump stated in a Truth Social post that U.S. combat operations in Iran are ahead of schedule but vowed to avenge the deaths of U.S. service members. 

The uncertainty has been reflected in the market as a surge in oil prices and a sharp increase in investments into “safe assets” like gold, which jumped by 2%.

Cryptopolitan reported that the market reaction was severe in Asia, with Hong Kong’s Hang Seng index dropping 2.05% and Japan’s Nikkei falling 1.53%. 

Despite the chaos, Bitcoin has returned into the green as US markets opened for the day, up about 4% in the last hour to almost $68,500. Other major tokens have followed with gains too, with Ethereum (ETH), Solana (SOL) and Ripple (XRP) all posting over 3% rallies in the last hour.