Amazon Web Services said early Monday that it was recovering from a widespread disruption that knocked popular apps and websites offline and caused connectivity problems for companies across several sectors.
The incident centered on US-EAST-1 in northern Virginia and was linked to a DNS malfunction that prevented applications from reaching key cloud APIs.
By 6:35 a.m. Eastern on October 20, 2025, AWS said the “underlying DNS issue” had been fully mitigated, though some requests could still be throttled while engineers worked toward full resolution.
What went down and when
AWS reported “increased error rates and latencies” across multiple services at 3:11 a.m. ET, an alert that kicked off several hours of instability for customers dependent on its compute, database and identity products.
Engineers traced the problem to a DNS issue that disrupted access to the DynamoDB API, which in turn rippled through dependent services, Reuters reported.
By shortly after sunrise on the East Coast, AWS systems showed signs of recovery, with most service operations working normally and isolated throttling still possible while traffic backlogs cleared. The Associated Press noted there was no indication of a cyberattack behind the outage.
Outage tracking sites and company status pages reflected the impact as it spread through the early morning. In prior years, US-EAST-1 has been the source of high-profile events, including incidents in 2021 and 2023, which similarly cascaded to downstream platforms that rely on AWS for core functions.
Who was affected
The outage touched a broad swath of consumer and enterprise services. Gaming platforms such as Fortnite and Roblox, social apps including Snapchat, and creative tools like Canva reported disruptions.
Fintech firms also experienced problems, with Coinbase and Robinhood attributing issues to AWS, and Perplexity citing cloud instability for site trouble, according to Reuters. Amazon’s own devices and apps, including Ring and Alexa, also saw interruptions.
Regional effects varied by customer architecture. Some banks, retailers and transport services reported intermittent access or timeouts, while others were able to fail over to unaffected regions or cached content.
The US-EAST-1 region is heavily used because of its scale and pricing, which can magnify the impact of a localized event when dependencies converge there, Reuters noted in its technical recap.
