Cloudflare records 49 Internet disruptions in Q3 2026, with power failures and government shutdowns leading

Cloudflare has reported 49 Internet disruptions worldwide during the third quarter of 2026, highlighting how power failures, government-directed shutdowns, natural disasters and conflict continue to affect global connectivity.

The findings, published through Cloudflare Radar, cover incidents observed between July and September. Government-directed shutdowns accounted for 19 disruptions, including 18 associated with national examinations in Iraq. Power outages were the second-largest category, with 14 incidents across 10 countries, while natural disasters accounted for five.

The report illustrates how Internet connectivity depends on systems beyond telecommunications networks themselves. Electricity supplies, data centre cooling, physical cables, interconnection facilities and government decisions can all affect the availability of digital services.

Cloudflare said its global network spans more than 330 cities across more than 125 countries and handles over 20% of web traffic, giving it visibility into changes in connectivity through traffic observed across its network.

For organisations that rely on cloud platforms, digital applications and connected services, the findings offer a reminder that Internet resilience requires planning for failures across multiple layers of infrastructure.

Power failures expose critical infrastructure dependencies

Electricity remained one of the clearest sources of connectivity disruption during the quarter. When power systems fail, the impact can extend beyond homes and businesses to mobile base stations, routers, data centres and the facilities where networks exchange traffic.

Cuba experienced four power-related Internet disruptions during Q3, reflecting repeated instability in its electricity grid. The incidents occurred on July 6, July 14, August 3 and September 18 to 20. Cloudflare Radar observed traffic declines ranging from 18% to 59% across the reported events.

Honduras provided another example of how a local infrastructure failure can affect connectivity at a national scale. On September 10, a power failure at an interconnection facility affected at least 11 major providers. The outage lasted approximately three hours, during which HTTP traffic across the country fell by 63%.

The incident highlights the importance of interconnection points, where networks exchange traffic. If several providers depend on the same facility, a failure at that location can have consequences beyond a single operator.

Cloudflare also recorded power-related disruptions in countries including Eswatini, Botswana, Tajikistan, Kenya, Saint Lucia, Georgia and Gabon.

Together, these events show that network resilience depends not only on telecommunications infrastructure but also on the reliability of the electricity systems that support it.

Exam-related shutdowns dominate government-directed disruptions

Government-directed shutdowns were the largest category in Cloudflare's Q3 data, with Iraq accounting for 18 of the 19 incidents recorded.

The shutdowns were associated with national examinations and took place on a scheduled basis, generally lasting between 15 and 90 minutes. Most occurred between August 26 and September 28, with additional incidents recorded in early July.

Although individual shutdowns were relatively short, their repeated occurrence demonstrates how policy decisions can interrupt access to online services independently of technical network failures.

Internet access plays a role in education, commerce, communications and public services. Restrictions can therefore affect a broad range of activities, even when imposed for limited periods.

Cloudflare said Iraq's communications minister had discussed developing a digital system intended to eliminate the need for centralised exam distribution and the associated Internet shutdowns.

The report also noted that Syria and Sudan, both of which have experienced exam-related Internet restrictions in previous years, did not record similar disruptions during the quarter. Cloudflare cautioned that this may reflect differences in examination calendars rather than a change in policy.

Conflict and natural disasters disrupt connectivity

Physical damage and extreme weather added to the quarter's connectivity challenges.

In Ethiopia, renewed fighting in the Tigray region coincided with a government-directed Internet shutdown lasting 10 days. Cloudflare observed a complete loss of traffic in the affected area, with HTTP traffic falling by up to 30% at the country level.

In Ukraine, a power outage in Sevastopol on August 12 followed drone strikes targeting energy infrastructure. The resulting Internet disruption lasted approximately 10 hours, with traffic declining by up to 95%.

Natural disasters also affected networks in several regions.

In Guam, Super Typhoon Bavi followed Super Typhoon Sinlaku by less than three months, contributing to a second major storm-related connectivity disruption in the same year. Damage to electricity infrastructure and transmission lines prolonged recovery, with Internet traffic taking several days to return to typical levels.

In Colombia, a magnitude 7.4 earthquake in August damaged infrastructure and was associated with a four-and-a-half-day disruption affecting traffic from Colombia Telecomunicaciones.

Nepal experienced a shorter disruption after flash floods on August 26 damaged local infrastructure and essential services, including electricity. Cloudflare Radar recorded a two-hour outage during which HTTP traffic fell by up to 18%.

These incidents demonstrate that connectivity disruptions can emerge when natural hazards damage the physical infrastructure supporting digital networks, including power distribution and fibre-optic links.

Data centres and individual networks remain vulnerable

The report also identified disruptions affecting individual providers and infrastructure facilities, including incidents in countries with established telecommunications networks.

On August 13, a cooling system failure at a data centre in Phoenix, Arizona, caused a near-total drop in Internet traffic associated with Namecheap's network. The incident illustrates how cooling systems, which help maintain safe operating temperatures for computing equipment, can become critical points of failure.

In Portugal, Internet provider MEO experienced an outage following what was described as a “mass attack”. Cloudflare also observed disruptions affecting networks and providers in Guyana, Haiti, Turkey and Iran.

These events varied in scope and cause. Some affected individual networks rather than entire countries, while others involved wider regional connectivity.

The distinction matters because a decline in traffic observed on a particular network does not necessarily mean that all Internet access in a city or country has been lost. Cloudflare Radar tracks traffic patterns and annotates identified disruptions, offering a view of their observed impact rather than a universal measure of every user's connectivity.

What the findings mean for enterprise resilience

For businesses that depend on cloud services, digital applications and connected operations, the report underscores the need to consider a wider range of failure scenarios when planning for continuity.

Network redundancy remains important, but separate network connections may not provide full protection if they depend on the same electricity supply, data centre, interconnection facility or physical fibre route.

Organisations can use these findings as a prompt to review several areas:

Infrastructure diversity: Assess whether backup connections and service providers rely on genuinely separate physical routes, facilities and power sources.

Power and cooling resilience: Review backup electricity, cooling arrangements and recovery procedures for critical infrastructure and technology services.

Geographic redundancy: Evaluate whether essential workloads and services can continue operating if a particular region or facility becomes unavailable.

Incident response and recovery: Test continuity plans against short, repeated interruptions as well as longer outages caused by severe weather, infrastructure damage or government restrictions.

Visibility and monitoring: Combine network traffic observations with application performance, service availability and customer-impact monitoring to understand how disruptions affect business operations.

These measures cannot eliminate every cause of disruption. However, they can help organisations identify shared dependencies, reduce single points of failure and prepare for a broader range of incidents.

Internet resilience depends on more than network design

Cloudflare Radar recorded 49 disruptions during Q3 2026, with government-directed shutdowns and power failures accounting for much of the total. Natural disasters, conflict and failures affecting individual networks added further examples of how connectivity can be interrupted.

Recovery times varied. Cloudflare reported that traffic returned to typical levels in under an hour for 14 disruptions, while 16 incidents recovered within three hours. Even short interruptions, however, can have consequences when organisations depend on continuous access to digital services.

The wider lesson is that Internet resilience depends on the systems surrounding the network as much as on the network itself. Reliable power, diverse infrastructure, physical protection, effective monitoring and well-tested recovery plans all contribute to keeping digital services available.

As businesses become more dependent on connected operations, understanding these dependencies will remain an important part of managing technology risk and business continuity.

---

About Cloudflare

Cloudflare is a connectivity cloud company that provides services designed to help organisations connect, protect and accelerate their applications, networks and digital infrastructure. Its offerings include network performance, cybersecurity, application services and developer tools. Cloudflare Radar provides insights into global Internet traffic, connectivity and disruptions using data from Cloudflare's network and other aggregated sources.

Source: Cloudflare Blog