AWS US-EAST-1 outage lasts three hours, but Fortnite stays online
Amazon's busiest cloud region failed on Sunday, yet the damage landed on staff sign-ins rather than the games people play.

Amazon Web Services' US-EAST-1 region, based in Northern Virginia, was disrupted for just over three hours on Sunday, 15 November. Unlike the big outage of October 2025, the major online games kept running. Epic Games' status page showed no Fortnite incident, and AWS did not list any games among the 11 services it said were affected.
Trouble began at 3:05 AM PST. AWS said it fixed the sign-in failures at 5:20 AM. A smaller group of customers kept losing connections until 6:21 AM, when AWS said it had fully mitigated the underlying network problem. That makes the full episode longer than the roughly two hours some online summaries gave. AWS has not yet explained what went wrong.
The hardest-hit service was IAM Identity Center, the tool companies use to sign their own employees into the AWS Console and other systems. When it failed, so did services that rely on it, including Amazon Connect and the AI coding tool Kiro. New customers also could not create AWS accounts. At one point AWS said affected customers had no workarounds.
This is why gamers were mostly spared. Players sign in to a game through the publisher's own account system, not through the staff login that broke. A studio's engineers may have been locked out of their dashboards for a while, but the servers hosting matches kept going unless they depended on the affected network pieces.
Those network pieces caused the longer tail. A subset of Network Load Balancers in one cell of a single Availability Zone, use1-az1, kept dropping connections after the sign-in fix. AWS's advice was to create a new load balancer with targets spread across several Availability Zones. Other services on the affected list included NAT Gateway, CloudWatch, Step Functions, OpenSearch Service and VPC Lattice.
The good news is that spreading systems across zones and keeping player logins separate from staff logins appears to have helped. The risk remains, though. Epic runs Fortnite's online services on AWS and Riot has moved to it, and Twitch is owned by Amazon. In October 2025, a DNS problem with DynamoDB in the same region caused a failure of about 15 hours that took down Fortnite and Roblox. Sunday was a far smaller event, but the cause is still unexplained. Details come from TalkEsport's report.