Cable faults and fiber cuts shape Q1 connectivity
Cloudflare’s network spans more than 330 cities in over 125 countries, interconnecting with over 13,000 network providers. That breadth provides a unique vantage point on Internet resilience, letting us observe disruption impacts at local, national, and network levels. As always, this summary covers observed and confirmed disruptions — not an exhaustive list — with a larger set of traffic anomalies available in the Cloudflare Radar Outage Center. Both bytes-based and request-based traffic graphs illustrate impacts, with metric choice based on which better shows the disruption.
Q1 2025 saw notable disruptions from cable damage and power outages. Severe storms caused outages in Ireland and Réunion; an earthquake led to ongoing connectivity issues in Myanmar. Russian networks were taken offline by a reported cyberattack and purported technical problems, while a fire briefly took a Haitian telecom provider offline. Notably, no government-directed Internet shutdowns were observed this quarter — an unusual occurrence that has only happened previously in Q4 2023 and Q1 2022 in the three-year history of this report series.
Pakistan: AAE-1 cable fault degrades performance
Shortly after the new year, a fault in the AAE-1 submarine cable disrupted Internet connectivity in Pakistan. The Pakistan Telecommunications Authority published a January 2 alert stating the fault occurred near Qatar and would likely impact user experience nationwide. Because Pakistan connects to seven submarine cables for international traffic, the loss did not cause an observable outage. However, the impact appeared in bandwidth and latency graphs: on January 2 and 3, median latency peaked around 125 ms, up from approximately 80 ms pre-disruption, while media download speeds dropped to around 6 Mbps from roughly 9 Mbps.
Pakistan Telecom (PTCL, AS17557) also flagged potential "slow browsing" in an Instagram update, and their network quality graphs show similarly timed shifts in median bandwidth and latency. With two more cable connections expected in 2026, Pakistan's connection diversity means a single cable issue typically has minimal availability impact, as traffic re-routes across other paths.
Syria: Repeated fiber sabotage outages
A widespread outage spanning January 23-24 was caused by sabotage damaging two fiber optic cables along the highway between Damascus and Homs, according to the Syrian Ministry of Communications. HTTP and DNS request traffic from Syria dropped to near zero between 00:30 and 03:30 local time on January 24 (21:30 January 23 – 00:30 January 24 UTC). Traffic began recovering shortly after and returned to expected levels by 09:00 local time (06:00 UTC). Announced IPv4 address space, almost exclusively from Syria Telecom (AS29256), dropped approximately 90% during the period, indicating a significant portion of the network became unreachable.
Syria experienced another outage on March 25, again from sabotage damaging fiber optic cables. The Ministry of Communications reported damage in the Maaloula and Hasiya regions, causing a near-complete outage between 03:00–13:15 local time (00:00–10:15 UTC). The graphs again show near-complete loss of HTTP request traffic and significant loss of announced IPv4 address space. Paradoxically, DNS request volume from Syria was elevated during this outage — unlike the January event — and it remains unclear what drove the additional traffic to Cloudflare's 1.1.1.1 DNS resolver.
Nepal: Upstream provider outage ripples downstream
In early February, multiple Internet providers in Nepal saw services disrupted when Indian provider Bharti Airtel (AS9498) went offline. Nepal Telecom (AS23752), Worldlink Communications (AS17501), Dishhome Fibernet (AS139922), Vianet Communications (AS45650), and Ncell (AS38565) — all of which have Airtel as an upstream provider or peer — experienced traffic disruptions between 21:00–22:30 local time (15:15–16:45 UTC) on February 2.
Published reports offer conflicting explanations for the Airtel issue: one claims it was related to an ongoing payment dispute, while another attributes it to reported fiber cuts in Airtel's network.
Power grid failures take down connectivity across multiple regions
The first quarter of 2025 saw a series of widespread power outages across several countries, each with significant knock-on effects on Internet connectivity. In each case, Cloudflare Radar data showed clear dips in traffic that aligned with the reported timing of the underlying electrical failures.
In Angola, an interruption in the North and Center Interconnected System on January 6 caused a blackout affecting eleven provinces, according to the National Electricity Transmission Network (RNT). Internet traffic dropped between 14:45 and 22:00 local time (13:45–21:00 UTC). RNT said it was investigating the cause, though no further details were confirmed.
Sri Lanka experienced an island-wide outage on February 9 after a monkey came into contact with a transformer at the Pandura electrical substation. Traffic began declining around 11:30 local time (06:00 UTC) and recovered by roughly 21:00 local time (15:30 UTC). The impact was visible at the network level, particularly for major provider AS18001 (Dialog).
Chile saw one of the most severe grid failures of the quarter on February 25, with reports indicating that 98.5% of the country was affected. A disconnection of the 500 kV transmission system in the Norte Chico region interrupted power supply from Arica to Los Lagos. Traffic fell sharply at around 14:15 local time (18:15 UTC) and took approximately 12 hours to return to expected levels, with about 94% of customers having power restored about an hour after traffic recovered.
A ground fault at the 15 de Septiembre electrical substation in El Salvador was cited as the cause of a power outage that disrupted connectivity in Honduras on March 1. The Regional Operator Entity (OER) said the failure occurred at 09:22 local time (15:22 UTC), and traffic from the country dropped by roughly half. The disruption was short-lived, with traffic back to normal levels about two hours later.
Cuba's National Electric System (SEN) failed on March 14 following "a failure at the Diezmero substation," according to a post from the state energy agency on X. That post noted the failure "caused a significant loss of generation in the west of #Cuba and with it the failure of the National Electric System, SEN." Request traffic from the country dropped immediately. Updates from the Cuban Electric Union and the Cuban Telecommunications Company kept subscribers informed over the following days. Traffic returned to expected levels around 20:00 local time on March 16 — two full days after the incident began.
An explosion and fire at the La Chorrera Thermoelectric Power Plant in Panama caused a nationwide blackout starting at 23:40 local time on March 15 (04:40 on March 16 UTC). Traffic dropped immediately, but recovery was relatively quick: the electric system reached 75% recovery by 03:00 local time (08:00 UTC) and full restoration was completed at 06:08 local time (11:08 UTC). Traffic volumes began rising once power came back.
Severe weather and seismic events
Storms hit Ireland and Réunion
Storm Éowyn battered Ireland in late January, knocking out power and water, causing property damage, and limiting air and train travel. The storm also disrupted Internet connectivity: traffic from the Connacht and Ulster provinces fell by 75% compared to the previous week at 06:30 local time (06:30 UTC) on January 24. Recovery was gradual, with traffic in those provinces approaching prior-week levels by mid-day on January 28.
Cyclone Garance made landfall over Réunion at roughly 10:00 local time (06:00 UTC) on February 28. Winds exceeding 100 miles per hour (160 km/h) caused power outages and infrastructure damage that disrupted connectivity. The most significant traffic impacts came in the hours after landfall, but it took several days for traffic to return to expected levels, which occurred around 08:00 local time (04:00 UTC) on March 4.
Earthquake in Myanmar
A magnitude 7.7 earthquake struck Myanmar on March 28 at 12:50 local time (06:20 UTC), triggering power outages and fuel shortages. Country-level traffic dropped by about 40% almost immediately. Regional impacts were far more severe: traffic from Nay Pyi Taw fell 97% compared to the previous week, Mandalay dropped 90%, Ayeyarwady lost 88%, Bago 50%, and Shan State was down 38%.
While regular traffic patterns and volumes bounced back within days at a country level, recovery at the network level was mixed. Both AS134840 (MCCL) and AS136442 (Oceanwave) saw significant drops in traffic after the earthquake, and both remained disrupted through the end of the first quarter. Peak traffic on MCCL increased slightly but stayed well below pre-earthquake levels nearly two weeks on. Oceanwave saw steady growth after the initial disruption and was approaching pre-earthquake peaks at the time of writing. (The cause of a significant spike in request traffic from Oceanwave on April 3–4 was unclear.) AS163255 (Mytel) and AS135300 (Myanmar Broadband Telecom) fared better, with smaller disruptions and significantly faster recoveries.
Cyberattack and fire damage
Russian ISP Nodex hit by cyberattack
Russian Internet provider Nodex (AS29329) said on January 7 that its network had been attacked by Ukrainian hackers, resulting in "its complete failure." The Ukrainian Cyber Alliance, a pro-Ukraine activist community formed in 2016, claimed responsibility in a Telegram post. Nodex said restoration efforts were ongoing and that the process was "painstaking and long."
The impact was visible in traffic data: traffic from the network began dropping after 03:00 local time (00:00 UTC) on January 7 and reached zero around 05:30 local time (02:30 UTC). Traffic remained essentially non-existent until around 14:00 local time (11:00 UTC) the following day, then recovered fairly quickly. Announced IPv4 address space fell by two-thirds at the same time traffic dropped to zero, recovering at 21:20 local time (18:20 UTC).
Los Angeles wildfires
During the early days of the 2025 Southern California wildfires, between January 7–9, there were clear Internet disruptions in at least 13 Los Angeles neighborhoods. Traffic drops of over 50% compared to the previous week were especially noticeable in Pacific Palisades, Altadena, Malibu, Temple City, and Monrovia. In the weeks that followed, traffic remained significantly lower than before the fires, particularly in Pacific Palisades and Altadena, reflecting the devastation in those areas. Malibu, Temple City, and Monrovia recovered their traffic significantly sooner, though peak levels remained somewhat below pre-fire levels.
Haiti fiber cuts
The Director General of Digicel Haiti (AS27653) reported on January 15 that fires in the metropolitan area had damaged two international fiber optic cables the previous night, and a third outage at 10:30 am affected all international services, including Internet and Moncash. The fires caused two complete Internet outages for Digicel Haiti's customers. Both traffic and announced IP address space (IPv4 and IPv6) dropped to zero between 20:30–21:45 local time on January 14 (01:30–02:45 on January 15 UTC) and again between 10:15–11:00 local time on January 15 (15:15–16:00 UTC).
Connectivity Failures Across Three Countries
Russia: A Brief and Suspicious Drop
On January 14, several major Russian networks—including AS8359 (MTS), AS12389 (Rostelecom), AS16345 (Beeline), AS31133 (MegaFon), and AS203451 (K-telecom)—experienced a short but sharp connectivity disruption. Traffic from these networks fell by roughly 80% between 14:00 and 14:30 UTC.
Roskomnadzor issued a brief statement acknowledging a “connectivity issue” that was “promptly restored.” However, independent observers speculated the cause may have been a failed update to Russia’s “firewall” system that was quickly rolled back. No official confirmation of that theory was provided.
Georgia: Cable Damage in Turkey and Georgia
Magticom (AS16010), one of Georgia’s largest ISPs, suffered a complete outage on January 27. Traffic and announced IP address space vanished at 17:25 UTC and only recovered at 21:55 UTC that same day—about four and a half hours later.
In a translated Facebook post, Magticom attributed the outage to physical cable damage. The company said its primary connectivity relies on “channels from Europe,” and that heavy snow and avalanche risk in Turkey prevented the partner company’s technicians from reaching the damaged section. On the backup route, Magticom reported “suspicious damage” at three points within Georgia’s Adjara region. The provider’s stated outage window matched the traffic loss observed in Cloudflare data.
France: An Unexplained Morning Incident
On March 11, customers of Bouygues Telecom (AS5410) in France saw a brief disruption. The provider acknowledged the issue in an X post, noting a “technical incident between 5 a.m. and 7 a.m.”
Traffic graphs confirm a dip between 05:00 and 06:45 local time (04:00–05:45 UTC), consistent with the provider’s timeline. Bouygues Telecom offered no further detail on the root cause.
A Mystery Outage in Syria
While Syria’s outages are often linked to cable cuts and are well documented, a multi-hour disruption on February 3 had no disclosed cause. Beginning around 11:00 UTC, national traffic dropped by approximately 80%, with announced IPv4 address space falling by about 60%. Service levels returned to normal by 20:00 UTC.
Syrian Television confirmed the outage on X, but no explanation for the disruption was ever made public.
Outlook
The absence of any government-directed internet shutdowns in the first quarter of 2025 is a positive shift from the previous quarter. However, this trend may prove temporary if countries such as Iraq and Syria resume such tactics during exam seasons. Governments should weigh the broader collateral damage of these actions against their intended goals.
