Most major online outages in 2025 traced back to shared platforms

The largest outages of 2025 were driven by failures in shared cloud and network infrastructure, not isolated app-level issues.

Online services were no longer optional for most people in 2025. Work, entertainment, payments, and communication all depended on platforms that were expected to be available at all times. This year made clear how fragile that expectation can be.

The most disruptive outages were not small faults affecting one company or one market. They were failures deep inside shared systems that many services rely on at once. When those systems failed, the consequences spread rapidly across video platforms, games, cloud services, and social apps, frequently in several regions at the same time.

That pattern reflects how concentrated digital infrastructure has become. Many popular services now sit on the same cloud providers, content delivery networks, and core platforms. When something fails at that layer, the problem does not stay local. It moves outward, affecting businesses and users far beyond the original fault.

Using Downdetector data from 2025, Ookla reviewed millions of user reports to identify the largest website and service disruptions of the year. The data shows where outages had the widest reach, and which parts of the digital stack proved most exposed.

When cloud failures ripple outward

The most severe outages in 2025 were caused by infrastructure that serves as the foundation for consumer services. Users may have noticed problems with apps or websites, but the root cause often sat deeper in the stack.

The largest outage of the year happened on October 20, when AWS experienced a problem that resulted in over 17 million Downdetector reports across Amazon services and platforms that depend on them. The disruption lasted over 15 hours and was linked to a problem with automated DNS management for DynamoDB in the US-EAST-1 region. As the issue spread, services such as Snapchat, Netflix, and multiple e-commerce sites were affected, showing how a single fault can cascade across unrelated products.

Gaming accounted for the second-largest outage. On February 7, the PlayStation Network experienced a network-wide disruption that lasted more than 24 hours and resulted in over 3.9 million reports. Players were unable to access major titles, such as Call of Duty and Fortnite. Downdetector's Incident Attribution analysis revealed that the problem originated within PSN, without major involvement from cloud or internet service providers.

Cloud infrastructure was again involved in the third-largest incident. On November 18, a global Cloudflare outage lasted nearly five hours and led to more than 3.3 million reports. Websites, applications, and APIs around the world were affected, reflecting how widely Cloudflare's services are embedded across the internet.

Different regions, different points of failure

While global outages dominated total report volumes, regional data shows that the services people rely on most vary by location.

In the United States and Canada, every major outage exceeded one million reports. PlayStation Network recorded the highest number, with 1.6 million reports. YouTube followed with 1.5 million during a streaming issue in October. The same AWS outage drove 1.2 million reports in the region, while Snapchat saw nearly 945,000 reports as part of that event. Connectivity problems also stood out, including outages affecting Starlink and a major Verizon disruption in late August.

Europe saw a broader mix of affected services. PlayStation Network again generated the most reports, at 1.7 million. Snapchat followed with close to one million. A Vodafone outage in the UK produced more than 833,000 reports after a non-malicious software issue disrupted both broadband and mobile services. WhatsApp and Spotify also experienced large regional outages.

In Asia Pacific, social platforms and cloud services featured most often. X generated more than 645,000 reports in March, while Snapchat, YouTube, and AWS each experienced significant outages later in the year.

YouTube and AWS outages, as well as frequent troubles with WhatsApp, were among the most severe interruptions across Latin America. A major outage at Banco Itaú also stood out, highlighting how service failures can affect financial institutions and their customers.

The Middle East and Africa had the largest report volumes, which were attributed to telecom carrier du, as well as regional spillover from global outages involving Cloudflare and Snapchat.

Trends in Singapore

According to Cloudflare’s 2025 Internet Year-in-review , peak growth in Internet traffic surged 53% YoY, versus 19% globally. Google and Facebook were again the most popular Internet services for the fourth year in a row, while OpenAI’s ChatGPT held onto the number one spot for most popular Gen AI service for the third year in a row. Facebook and Instagram were the most popular social media platforms, followed by Tiktok.

Over 15% of Singapore traffic (vs 6% globally) was mitigated by Cloudflare's systems as being potentially malicious or for customer-defined reasons. Career and Education sites like educational institutions were the number one most targeted vertical while Computer & Electronics came in second, and People & Society came in third.

Across regions and services, the data points to a shared weakness. As digital platforms grow, they rely more heavily on common infrastructure. That dependence makes services easier to scale, but it also means failures are harder to contain.

The outages of 2025 show how closely connected today's online services have become. Problems that start in one system can spread across industries and borders within minutes. The result is not just inconvenient for users, but a clearer view of how much risk now sits inside a small number of shared platforms.