How Does the Internet Cross the Ocean? Undersea Cables Explained

Ask most people how a video call reaches another continent and they will point vaguely at the sky. It is an understandable guess, and it is wrong. Satellites carry a small and genuinely useful share of global traffic, particularly to remote places, ships and aircraft. The overwhelming majority of intercontinental data goes the unglamorous way: along a cable, on the floor of the sea, in a bundle of glass fibres about as thick as a garden hose.

This is one of my favourite subjects in economic geography, because it is a case where the modern digital world turns out to run on the same map as nineteenth-century shipping. The cables follow the sea, and the sea has always had preferences.

Why glass beats the sky

The physics is decisive. A modern submarine cable carries several pairs of optical fibres, each capable of transmitting enormous volumes of data simultaneously on many wavelengths of light. Signals weaken over distance, so repeaters spaced along the cable amplify them, powered by electricity fed from shore along a copper conductor wrapped around the fibre core.

Two advantages matter most. Capacity: a single cable can carry a share of an ocean's traffic that no comparable satellite constellation matches. And latency: light in fibre takes a fairly direct route along the seabed, whereas a signal relayed through a distant orbit has to travel much further, which adds delay you can feel in a call or a financial transaction. Low-orbit satellite systems narrow that gap considerably and are transforming coverage in hard-to-reach places, but for bulk trunk traffic between continents, cables remain the backbone.

A Victorian idea that never went away

The first attempt to link continents by cable was electrical telegraphy. After several failures, a cable across the Atlantic carried messages briefly in 1858 before degrading, and a durable connection was established in 1866 using the enormous steamship Great Eastern. The effect on the world was genuinely comparable to what we now attribute to the internet: a message between London and North America went from taking a ship's crossing to taking minutes.

The nineteenth-century network expanded along imperial and commercial routes, which is one reason the modern map inherits old shapes. Telephone cables followed in the twentieth century, and the shift to optical fibre from the 1980s onward multiplied capacity by orders of magnitude. The pattern of ownership changed too: where states and national carriers once dominated, today many of the largest new cables are financed by consortia that include major technology companies moving their own traffic.

How you lay a cable across an ocean

1. 1Survey the route. Ships map the seabed with sonar to find a path that avoids canyons, volcanic zones, coral, wrecks and steep slopes prone to landslides. The straight line is rarely the chosen line.
2. 2Secure permissions. A route may cross the territorial waters and economic zones of many states, each with its own permitting process, plus fishing and environmental considerations.
3. 3Load and lay. A cable ship carries thousands of kilometres of cable coiled in huge tanks and pays it out over the stern at walking pace, laying it to follow the contours of the bottom rather than bridging depressions.
4. 4Bury the shallow ends. Near shore, where anchors and trawls are the real hazard, a plough cuts a trench and the cable is buried. In the deep ocean it simply rests on the sediment.
5. 5Land it. The cable comes ashore to a landing station, an unremarkable building on the coast where the optical signal joins terrestrial networks.

The cable itself is armoured heavily in shallow water and quite thin far offshore, because the deep sea is a calmer place than a busy harbour approach. That difference in construction tells you where engineers expect trouble.

Why the routes cluster

Cables do not spread evenly around the globe. They gather into corridors, and the corridors are chosen by geography rather than fashion.

The result is that a handful of maritime pinch points carry a strikingly large share of the world's data as well as its cargo. The same straits and canals that shaped merchant sailing, which we go through in maritime chokepoints explained, reappear in the cable atlas. Ports are the other constant: the coastal cities that grew rich moving goods are now the places where fibre comes ashore, for the same underlying reason set out in why port cities rule the world. Deep-water access, established trade links and dense population have always travelled together.

Landing stations are therefore a genuine piece of strategic geography, in the plain sense that a small number of coastal sites matter far more than their size suggests. Countries and companies pay close attention to where cables come ashore, how many independent stations serve a region, and whether an island or a landlocked neighbour depends on a single path. It is the same logic as ports, railheads and mountain passes, applied to information.

Faults, ships and redundancy

Cables break with unremarkable regularity, on the order of a hundred or more faults worldwide in a typical year. The great majority of damage is accidental and human: a ship dragging its anchor, a trawl or dredge catching a cable in shallow water. Natural causes account for much of the rest, notably undersea landslides and sediment flows triggered by earthquakes, which can sever several cables in one corridor at once because they all follow the same surveyed path.

Repair is a specialised industry. A small global fleet of cable repair ships sits on standby under regional agreements. When a fault is located, using the electrical and optical characteristics of the line to estimate the distance to the break, the ship grapples the cable from the seabed, raises both ends, splices in a fresh section and lowers it back. In deep water this takes days to weeks, weather permitting.

The reason you usually notice nothing is redundancy. Traffic is designed to reroute across other cables automatically, so a single break shows up as a routing change rather than an outage. Regions served by many independent cables are robust; regions dependent on one or two are not, and that asymmetry is the most useful thing to understand about the whole system.

How to read a cable map

• Count paths, not cables. A place with four cables through one corridor is less resilient than a place with two through separate corridors.
• Look at the landfalls. Clusters of landings in one bay create a shared vulnerability even when the sea routes differ.
• Note the age. Older cables carry less capacity, so a busy-looking map can overstate a region's bandwidth.
• Follow the money. Consortium members reveal whose traffic a route was built for.
• Remember the land legs. A cable is only as useful as the terrestrial network and power behind its landing station.

That is a habit of reading rather than a set of facts to memorise, and it is the same habit that makes any map informative. We go through the general version in how to read a geopolitical map. Disclosure since I am about to mention our own product: MindSnap is our app, and this kind of infrastructure geography is exactly what its two-minute daily lessons and longer narrative collections are built for, with five flagship story collections plus unlimited further topics you can browse at topics.

The closing thought is the one I find most durable. We talk about data as though it were weightless and placeless. It travels through armoured glass, lying in mud, in corridors chosen by the shape of continents, coming ashore in coastal towns that have been strategically important since sail. The cloud has a very specific address, and quite a lot of it is underwater.