The internet is popularly imagined as something in the air, a cloud, a wireless signal, an abstraction with no physical location, and that picture is almost entirely wrong for the traffic that crosses between continents. Nearly all of it travels through a few hundred cables lying on the ocean floor, and that physical network, not any metaphor about clouds, is what actually determines how the regions of the world connect to one another and how resilient those connections are.
A submarine cable is a bundle of optical fibers, armored in steel and insulation and often no thicker than a hose, laid across the seabed by specialized ships that pay it out over thousands of kilometers of ocean. Inside it, the same optical signals that carry a city's traffic carry a continent's. Essentially every message exchanged between, for example, Europe and North America, an email, a video call, a financial transaction, crosses the Atlantic as light within one of these cables rather than by satellite. Satellites handle a small fraction of intercontinental traffic, useful where cables cannot reach, while the cables carry the overwhelming majority, because fiber on the seabed offers lower latency, lower cost, and vastly greater capacity than any satellite link.
What is not widely appreciated is how concentrated and physically vulnerable this infrastructure is. A relatively small number of cables carry the traffic of entire regions, they come ashore at a limited set of landing stations, and they are more exposed than their importance would suggest. Ship anchors, fishing equipment, undersea landslides, and ordinary wear sever them regularly, with faults occurring somewhere in the world on the order of several times a week. The network as a whole absorbs this because it is meshed, traffic reroutes automatically over alternative cables when one is cut, but a country connected by only one or two links has no such redundancy and can lose a large share of its international connectivity when a single cable parts, which makes cable diversity a genuine matter of national resilience rather than a technical footnote.
These cables are not public infrastructure in the manner of roads, they are financed and owned by consortiums of telecommunications companies and, increasingly, by the large technology firms whose services depend on them and which now commission cables of their own. The consequence is a privately held physical backbone that carries almost all international data, along routes shaped by geography, cost, and the politics of where a cable may safely land. The comforting image of a cloud conceals a heavier and more consequential reality, that the global internet is a physical system, predominantly glass and predominantly underwater, resting on a seabed almost no one will ever see, and recognizing that is the first step toward taking its fragility, and the concentration of control over it, as seriously as the traffic it carries deserves.