Almost all intercontinental traffic runs through cables on the seabed

Comparisons of submarine cables usually pick a winner. This one picks the circumstances, which is more useful.

Fibre on the seabed, not signals in space

A submarine cable is a bundle of optical fibres wrapped in steel and polyethylene, laid directly on the seabed in deep water and buried in shallow water. Light pulses travel through the fibre and are amplified every few tens of kilometres by repeaters powered from shore through a copper conductor in the cable. Powering a device thousands of kilometres offshore through the cable itself is one of the least appreciated engineering achievements in the industry.

Repeaters are designed to work untouched for decades, because visiting one is enormously expensive.

Capacity comes from wavelengths

A single fibre pair carries many separate wavelengths of light simultaneously, each modulated independently. Upgrading the terminal equipment at each end increases capacity on an existing cable without touching the wet plant.

In practice, this is why cables laid years ago carry vastly more traffic now than when they were commissioned. The physical limit is set by the fibre and the amplifiers, and it is being approached on the oldest routes.

Geography concentrates risk

Cables converge on a small number of natural corridors, and a handful of straits and chokepoints carry a disproportionate share of the world traffic. Landing stations are similarly concentrated, so damage near shore can affect many cables at once.

Countries served by only one or two cables experience national outages when one fails, which has happened repeatedly. Diversity of route matters more than total capacity for resilience, and it is expensive.

Damage is routine and mostly accidental

Fishing gear and ship anchors in shallow water cause the majority of faults, followed by seabed movement and abrasion. Deliberate damage is discussed heavily and remains a small share of recorded faults, though attribution is genuinely difficult. Faults are located by measuring reflections along the fibre from shore, which gives a position accurate enough to send a ship to.

Cable protection zones and burial in shallow water reduce the rate substantially and cannot eliminate it.

Repair is slow by nature

A specialised ship must sail to the location, grapple the cable from the seabed, cut it, lift both ends, splice in a new section and lay it back. The global fleet of such ships is small and ageing, and vessels are shared under regional maintenance agreements.

Under load, weather, permits and territorial waters routinely extend repairs from days to weeks. This is why network operators buy capacity on multiple cables rather than relying on any one.

Figures here are typical rather than guaranteed — check the spec sheet for your part.

Who owns and builds them

Ownership has shifted from consortia of telecommunications carriers towards large content companies building routes for their own traffic. That changes which routes get built, because the priorities of a content company differ from those of a national carrier. Regulation of landing rights and of ownership has become a national security question in several jurisdictions.

In the datasheet, latency between financial centres has been enough to justify entire cables on specific routes, which is a reminder that milliseconds have a market price.

Side by side

The takeaway

The internet between continents is a few hundred cables and a small fleet of ships that fix them.

Understanding the failure mode tells you more than the feature list does.