How undersea cables get repaired when an anchor snaps the internet

Roughly 95 per cent of all intercontinental internet traffic does not move through satellites or radio towers. It travels through hundreds of thousands of kilometres of glass fibre resting on the ocean bed, some of it no thicker than a domestic garden hose.

When one of these lines snaps — and industry fault logs record between 100 and 150 breaks every year worldwide — the surprise for most users is that their screens do not go dark.

What happens when an undersea cable breaks

Most breaks do not stem from sabotage or marine life. Commercial fishing nets and dragging ship anchors cause roughly two-thirds of all cable faults, according to historic incident logs from the International Cable Protection Committee (ICPC). Underwater earthquakes, volcanic activity, and undersea landslides account for most of the remainder.

The moment a fibre snaps, laser light stops bouncing through the glass core. Shore engineers at terminal landing stations immediately run an optical time-domain reflectometer test. By firing a light pulse down the intact segment and measuring the reflection time from the severed end, the system can calculate the break location to within a few metres, even if the damage sits hundreds of kilometres offshore.

Data does not wait for a fix. Modern international networks operate as meshes rather than single lines. Global cloud providers and telecommunications consortia buy capacity across multiple parallel systems laid along divergent geographic routes. Automated routing protocols detect the loss of throughput in milliseconds and shift data streams onto alternative cables.

Widespread disruption usually occurs only in places reliant on a single link, such as remote island territories, or at maritime choke points where dozens of lines crowd through shallow channels. When undersea earthquakes triggered massive seabed slides in the Luzon Strait off Taiwan in 2006, internet speeds across East Asia dropped sharply for weeks because alternative paths were swamped by the diverted load.

How repair ships fix a cable on the ocean floor

Restoring a broken cable is mechanical, slow, and expensive. Only around 60 specialised cable-laying and repair vessels operate worldwide, and getting one to a break site can take days or weeks depending on port availability, sea conditions, and territorial sailing permits.

Once on site, the ship uses sonar transponders or a remote-operated vehicle to find the severed line. In deep water, where cables lie unprotected by steel armour, the crew lowers a grapnel — a heavy pronged anchor — to drag across the coordinates until winch sensors record tension.

Because a cable lying under thousands of metres of water cannot be lifted in a straight pull without snapping under tension, crews must cut the cable on the seabed, raise one end, and attach it to a temporary marker buoy. They then grapple the second end, haul it aboard into a dust-free clean room, and use precision fusion splicers to weld the microscopic glass cores together.

Engineers attach a new section of spare cable to bridge the gap, steam back towards the buoyed end, and make the final joint. Because the added segment introduces slack, the crew carefully lays the repaired line back onto the seabed in a wide loop to prevent snarls. A typical offshore repair takes between one and three weeks on site and routinely costs hundreds of thousands of pounds.

Why route redundancy keeps the internet on

The logistical delay of undersea repairs explains why network operators invest heavily in redundant routing rather than relying on cable armour alone. Over the past decade, cloud providers including Google, Meta, and Microsoft have financed private cables alongside consortium systems, intentionally threading them through separate oceanic valleys.

Yet vulnerabilities remain. Choke points like the Red Sea, the Strait of Malacca, and the English Channel squeeze critical transatlantic and European-Asian paths into congested waters. Until emerging Arctic and overland routes provide genuine alternatives, the reliability of international communications rests on routed software buffers and the small global fleet tasked with hauling broken glass off the seabed.