The internet is mostly under the sea
When a video call from Lagos to London stutters, people blame their router, their provider or the weather. The real cause is often thousands of kilometres away, under water. The great majority of international internet traffic travels through submarine fibre optic cables, laid across ocean floors by specialised ships. Satellites matter, particularly for remote places, but they carry a small share of the total.
For much of Africa, that dependence is both a story of progress and a story of fragility.
Why cables matter more than they seem
A country's internet quality is shaped by how it connects to the rest of the world. Content, cloud services and most of the websites people use are hosted elsewhere. Every request travels out and back, and the route determines speed, cost and resilience.
Three things follow from this.
- Capacity sets the ceiling on how much can flow. More cable capacity generally means more room for growth and, with competition, lower wholesale prices.
- Latency is the delay a signal experiences. Shorter and more direct routes help, which is why hosting content closer to users matters as much as laying cables.
- Redundancy decides what happens when something breaks. Cables are cut by ship anchors, fishing, seabed movement and faults. A country with one route has a single point of failure. A country with several can shift traffic.
African coastal countries have seen outages when cable faults have coincided, and the effects reach well beyond browsing: payments, banking apps, hospital systems and businesses that run on cloud software all slow or stop.
What has changed
Older systems on the continent, such as SAT-3 and later WACS and ACE on the west coast, brought the first big improvements. More recently, new systems with far greater capacity have been built or landed, including 2Africa, a large project backed by a group of operators and technology companies, and Equiano, associated with Google, which lands in Nigeria among other countries.
The pattern behind these projects is worth noting. Large technology companies are now investing directly in cables, because they are some of the biggest users of the traffic. That changes who builds the infrastructure and who benefits first. It tends to bring a lot of capacity quickly, and it raises fair questions about how openly that capacity is made available to local operators.
The cable on the seabed is also only half the picture. Capacity has to reach shore and move inland.
The part after landing
A landing station is where a cable comes ashore. From there, traffic needs to cross the country: through terrestrial backbone networks, into internet exchange points, data centres and finally to the networks that serve homes and offices.
This is where many of the gains can be lost or realised.
- Landing station access. If a station is controlled by one party and access is expensive or slow, prices stay high even when capacity is plentiful.
- Terrestrial backbones. Fibre across land is vulnerable to road works and damage, and the cost of laying it is a major reason inland connectivity lags.
- Local exchange points and hosting. When networks in a country exchange traffic locally instead of sending it abroad and back, everything becomes faster and cheaper. Local data centres and caches for popular content strengthen the effect.
- The last mile. Affordable access for ordinary users depends on retail competition, spectrum, devices and electricity as much as on any cable.
A new cable improves the first link in a long chain. The user feels the weakest.
What it means for builders
For anyone building software here, the cable story has practical implications.
Do not assume your users are always online. Design applications that tolerate slow, interrupted or high latency connections: queue actions, retry safely, and avoid chatty designs that make many round trips per screen.
Think about where your systems run. Hosting in a region close to users reduces delay, and a design that continues to work during a regional outage is worth more than one that depends on a single remote provider.
Plan for failure. When a cable fault degrades the connection, payments and critical workflows should degrade gracefully, with clear messages and safe retry behaviour, rather than leaving transactions in doubt.
Use caching and content delivery sensibly too. Serving static assets from a node near your users, and keeping payloads small, makes an application feel faster on the same connection, and it costs far less than waiting for the next cable to land.
A moment, not a conclusion
Calling this a cable moment is fair in one sense: more capacity is arriving than the continent has seen before. But capacity alone does not equal good internet. Competition at the landing station, investment inland, local exchange and affordable access will decide how much of it reaches people.
The sensible question for any company is not "has the cable landed?" but "what happens to my service, and my customers, on the day it breaks?"