Almost all traffic between continents travels through fibre optic cables lying on the seabed rather than through satellites. The physical arrangement of that network shapes where the internet is fast and where it is not.
Light carries the data
Each cable contains a small number of glass fibre pairs, and information travels as pulses of light reflected along the fibre core.
Repeaters spaced along the route amplify the signal, drawing power supplied along a copper conductor running the length of the cable from landing stations at each end.
Capacity has grown mainly by improving the equipment at the ends rather than by laying more glass, since the same fibre can carry more data as encoding improves.
Routes follow geography and history
Cables are laid along paths chosen for seabed conditions, avoiding steep terrain, known fishing grounds and heavy anchoring areas where possible.
Landing points require suitable coastline, existing terrestrial backhaul and permission from the coastal state, so the same locations are used repeatedly.
The result is a network concentrated into a small number of corridors, with several major routes squeezed through the same narrow straits.
Ownership shifted toward the platforms
Cables were traditionally built by consortia of telecommunications carriers sharing the cost and the resulting capacity.
Large content and cloud companies now finance systems directly, because they move enormous volumes between their own data centres and prefer to control the path.
This changes who decides where capacity goes, since a company builds toward its own users rather than toward general connectivity needs.
Breaks are common and repairs are slow
Most faults are caused by fishing gear and ship anchors in shallow water rather than by anything happening in the deep ocean.
Repair requires a specialised vessel to locate the fault, raise the cable, splice in a replacement section and return it to the seabed.
Because the global fleet of these ships is small and weather-dependent, a repair can take weeks even where the fault is quickly identified.
Redundancy determines who notices
Regions served by many cables absorb a break by rerouting traffic, and users see nothing beyond a small change in latency.
Places served by one or two systems lose a substantial share of their capacity at once, and the effect on daily life is immediate.
This is why cable diversity, rather than total capacity, is the measure that best predicts how a country experiences a fault.