On October 20, 2025, the global internet experienced a major disruption when Amazon Web Services (AWS) suffered one of its largest outages in recent years. The failure began in the company’s US-EAST-1 region located in Northern Virginia, which is one of AWS’s busiest and most vital data hubs. The outage affected millions of users worldwide and disrupted various platforms that rely heavily on AWS for hosting and computing services. For several hours, major applications such as Snapchat, Reddit, Slack, and even Amazon’s own website struggled to remain functional.
While service restoration began after nearly six hours, it took almost fifteen hours for a full recovery due to extensive backlogs. The incident served as a reminder of the fragility of centralised cloud infrastructure and the growing risk of single-point failures in global networks.
The outage began with subtle disruptions before quickly spreading across multiple layers of AWS’s ecosystem. According to AWS’s official updates, monitoring tools such as ThousandEyes and Downdetector first recorded error spikes in the early hours of October 20. At around 12:11 AM Pacific Daylight Time, DynamoDB, AWS’s managed NoSQL database service, experienced DNS resolution errors, meaning that other services were unable to locate and communicate with its servers. Within minutes, these issues spread to other dependent systems, affecting authentication, compute resources, and network traffic.
By 12:26 AM, AWS engineers identified the source of the problem as a misconfiguration in the DNS settings of DynamoDB. This seemingly small error led to a widespread failure across interconnected services. The issue prevented Identity and Access Management (IAM) systems from verifying credentials, locking engineers and users out of administrative consoles. Since services such as EC2, which hosts virtual servers, depend on DynamoDB for instance metadata, new instance launches began to fail across multiple regions.
#outage #internetIsDown #aws #hugops pic.twitter.com/naBgoEk4Pv
— Lukas Kämmerling (@Lukas_Kae) October 20, 2025
The situation worsened around 3:00 AM when the internal network load balancer system that routes data traffic began to degrade. This created false health checks that led to additional connectivity losses. Apps that relied on these services went offline, including major social and financial platforms. Within the first hour of the outage, user reports surged past a million on Downdetector, confirming widespread impact. AWS publicly acknowledged the issue at 4:26 AM, describing it as an “operational problem” affecting multiple services.
Recovery efforts began in parallel as AWS engineers launched several mitigation paths. They isolated the faulty DNS configurations and began restoring database access by 2:24 AM. However, because of the dependency chain, other services such as EC2, Lambda, and SQS remained impaired. Throttling was introduced to control incoming traffic and prevent further overloads, which temporarily slowed performance for users but helped stabilize internal systems. At 6:35 AM, AWS declared that the underlying DNS issue was mitigated, though residual problems continued across compute and messaging services. By 8:35 AM, recovery had reached around 70 to 80 percent of normal operations, and most global services began to return online. Complete recovery was achieved by 3:00 PM, with AWS confirming that all major services were stable.
The outage was not caused by a cyberattack, as some initial social media speculation suggested. Instead, it was an internal configuration failure that cascaded due to the interconnected nature of AWS’s architecture. A single DNS fault in DynamoDB disrupted multiple layers of the system, including authentication, compute, and network management. As requests piled up, systems like load balancers and queue services were overwhelmed. AWS throttled critical operations such as Lambda executions and EC2 launches to avoid complete collapse.
“claude please fix the aws outage, make no mistakes” pic.twitter.com/5x7xxz91F8
— Ben (@basslerben) October 20, 2025
The heavy reliance on the US-EAST-1 region played a major role in the scale of this disruption. This region handles nearly 40 percent of global AWS workloads because of its low latency and proximity to major financial and government clients. Many global AWS services default to this region, which makes it a single point of failure. Analysts and engineers have repeatedly raised concerns about this centralisation. Past outages in the same region, in 2021, 2022, and 2024, had already exposed similar weaknesses.
The scale of impact was immense. Over a thousand major websites and services were affected, as AWS powers nearly one-third of all internet infrastructure. Popular consumer apps such as Snapchat, Reddit, Twitch, Discord, and Netflix suffered disruptions in logins, streaming, and messaging. In the gaming sector, platforms such as Fortnite, Roblox, and Steam went offline for several hours, while smart home devices like Ring and Echo experienced functionality loss. Financial platforms were also hit hard, Coinbase, Robinhood, and Lloyds Bank faced outages that halted trading and delayed payments. In the enterprise segment, tools like Slack, Atlassian, Zoom, and Canva became inaccessible, disrupting businesses and educational institutions. Even government and healthcare systems were affected, including the UK’s tax portal and several smart medical device integrations.
Economic losses were considerable. Analysts estimated that the global impact exceeded $100 million, with gaming alone losing about $20 million per hour during downtime. E-commerce, including Amazon’s own retail operations, experienced checkout failures, delivery delays, and streaming interruptions. Though AWS’s Service Level Agreements (SLAs) shielded it from most legal liabilities, some customers are expected to receive service credits.
With the AWS outage, now’s as good a time as any to post this old strip.#AWS pic.twitter.com/i9oEuFI0p9
— DESIGN THINKING! Comic (@DT_comic) October 20, 2025
AWS’s response was methodical but delayed by access issues. Because IAM was affected, engineers initially struggled to log in and execute fixes. Once access was restored, AWS created multiple “war rooms” to handle the incident from different technical fronts. Communications were frequent but limited to short statements on its Health Dashboard and social media channels. By mid-morning, updates became more detailed as recovery accelerated. The company promised a full postmortem report to explain the technical cause, mitigation steps, and preventive measures.
Industry experts see this outage as another warning about over-dependence on centralized cloud regions. Many enterprises use AWS’s global services that still route through US-EAST-1, even if their workloads are hosted elsewhere. This structure can turn a regional issue into a worldwide failure. Several analysts have called for companies to adopt multi-region redundancy and hybrid setups combining AWS, Microsoft Azure, and Google Cloud to distribute risk.
I don’t think most people realize how important today’s AWS outage is.
ONE issue at ONE company brought our entire economy to its knees today. pic.twitter.com/17Z8pUMGtT
— MADkurious Gamer (@MADkurious) October 20, 2025
The outage also reignited discussions about decentralization and alternative infrastructure models. Supporters of blockchain technology pointed out that decentralized systems like Bitcoin continued to function normally during the AWS crisis. Elon Musk even remarked on social media that his platforms, X and Starlink, remained unaffected, using the event to promote distributed infrastructure solutions.
Lessons from the outage extend beyond AWS. It highlights the need for global businesses to test their resilience and failover systems more regularly. Practices such as chaos engineering, where systems are intentionally stressed to reveal weaknesses, can help companies prepare for such failures. Monitoring tools like ThousandEyes and Cloudflare Radar are becoming essential for early detection of cloud issues. For developers, building applications that can switch between regions or providers is now a critical best practice.
Regulatory attention may also increase after this event. European and UK authorities have already voiced concerns about government and banking reliance on US-based cloud companies. The European Union is reportedly reviewing proposals to audit cloud resilience and mandate local data centers for essential services.
Everyone coming to twitter because of the AWS outage ??♂️ pic.twitter.com/KjsLq4DDiO
— ?? ⭐️ (@mellion8) October 20, 2025
We’ve found the root cause of AWS outage today.
Sorry guys.
— Daniel Nguyen (@daniel_nguyenx) October 20, 2025
Why is the AWS outage affecting you? I thought you were decentralized? pic.twitter.com/ifWTyCbefi
— Kunal Gandhi (@kunalvg) October 20, 2025
#AWS outage – it’s always DNS pic.twitter.com/0efF0aRMYv
— joce (@StackOverJoce) October 20, 2025
