The Internet Is Under the Ocean

Send a photograph from Singapore to San Francisco and it feels as though the image leaves your phone, enters the air and somehow reappears on another continent. The language we use encourages that illusion: wireless networks, cloud storage, satellite internet, cyberspace. But the long-distance internet is stubbornly physical. Your photograph probably travelled through glass fibres resting in darkness on the ocean floor.

The modern world depends on a network most people never see. Submarine telecommunications cables carry more than 99 per cent of international data traffic, supporting cloud services, financial systems, government communications, education, healthcare and the ordinary messages exchanged every second.¹ The “cloud” is not floating above us. Its most important international roads run beneath the sea.

Your message enters the sea

A mobile phone is wireless only for the first small part of its journey. The signal travels from your phone to a nearby cell tower or Wi-Fi access point. From there it usually enters terrestrial fibre, passes through network facilities and, if the destination is abroad, reaches a coastal cable landing station.

The landing station is where the apparently weightless internet becomes unmistakably industrial. It houses transmission equipment, power-feeding systems and connections into domestic fibre networks. Beyond the building, the cable crosses the beach, is buried in shallow water for protection, and then descends toward the deep ocean.

A garden hose carrying a civilisation

For most of a transoceanic journey, a modern submarine cable is roughly as wide as a garden hose. The glass fibres carrying the light are much thinner, around the diameter of a human hair. What surrounds them depends on where the cable is laid.²

In deep water, where anchors and fishing gear are less likely to reach, the cable can be comparatively slender. Near the coast it receives heavier steel armour because the shallow seabed is the dangerous part of the route. The cable is often buried close to shore, then laid directly on the seabed in the deep ocean.

How light crosses an ocean

Data is translated into patterns of light. Lasers at one end flash through the optical fibres at extraordinary speed, using multiple wavelengths so that many streams can travel through the same fibre pair at once. At the other end, receivers turn the light back into electronic information.

Light weakens as it travels. Long transoceanic systems therefore include optical repeaters, highly reliable amplifying equipment integrated into the cable at intervals. Power-feeding equipment at the landing stations sends electricity through the cable’s conductor to operate them. A modern submarine system is not simply a long piece of glass; it is a chain of optical and electrical engineering designed to function for decades in a place where maintenance is exceptionally difficult.³

Why not send everything through space?

Satellites are indispensable where fibre cannot economically reach: remote communities, ships, aircraft, disaster zones and mobile users across wide areas. Low-Earth-orbit constellations have also reduced the latency traditionally associated with satellite connections.

But satellites and submarine cables solve different problems. Fibre offers vastly more aggregate capacity at a lower cost per unit of data. It is the industrial freight network of international connectivity; satellites are flexible aerial routes that extend reach and provide valuable resilience. The two are complementary, not interchangeable.²

Who owns the ocean internet?

The early cable world was dominated by telecommunications carriers that formed international consortia, shared construction costs and divided capacity. That model still exists. So do privately owned systems operated by specialist infrastructure companies.

The major change is the arrival of hyperscale technology companies. Google, Meta, Microsoft and Amazon have become major investors because their global cloud platforms, video services and data centres consume extraordinary amounts of bandwidth. Owning or co-owning routes gives them capacity, control over network design and less dependence on third-party carriers.²

ITU data indicates that annual investment in submarine infrastructure rose from about US$0.8 billion in 2015 to US$9.7 billion in 2025, with hyperscale technology companies playing a leading financing role.⁴ The seabed is therefore not just engineering. It is also capital allocation, corporate strategy, permitting and geopolitics.

The map is a map of power

TeleGeography tracks more than 600 active and planned submarine cable systems and over 1.5 million kilometres of cable in service as of 2026.² But the lines are not evenly distributed. They converge at coastal hubs, narrow seas and landing corridors where geography, regulation, existing networks and commercial demand make connection practical.

A landing station can turn a coastline into a digital gateway. Multiple independent routes can make a country resilient and attractive to data centres. Dependence on one or two systems can make an island or underserved region vulnerable. This is why the ITU’s 2026 resilience recommendations focus not only on protecting individual cables but also on route diversity, repair permissions and the geographic concentration of infrastructure.¹

What actually breaks a submarine cable?

Cable faults happen regularly, roughly 150 to 200 times worldwide each year. The surprising part is how ordinary the causes are. Industry data indicates that around 70 to 80 per cent are caused by accidental human activity, particularly fishing gear and ships’ anchors. Natural hazards such as earthquakes and underwater landslides account for part of the remainder; equipment failure also occurs. Deliberate sabotage attracts headlines, but it is not the everyday explanation for most faults.⁵

Risk is concentrated near coastlines, where the cable shares space with ports, vessels, fishing grounds and construction. In the deep sea, the network is generally quieter and safer, until geology moves.

How do you repair glass on the ocean floor?

The network survives because a specialised maintenance system exists around it. Operators detect a fault from shore and estimate its location using electrical and optical measurements. A cable repair vessel is dispatched from a regional maintenance base, carrying spare cable, grapnels, remotely operated equipment and technicians trained to splice fibre at sea.

At the site, the ship retrieves the cable from the seabed. The damaged section is cut away, replacement cable is spliced in, the connection is tested and the repaired section is lowered back into the water. The engineering may be precise; the delay is often administrative and geographic. A vessel may be far away, weather may be poor, and permits, customs rules or restrictions on foreign ships can add weeks or months.¹

Why one break is usually invisible

Most users never notice the average cable fault because networks are built around redundancy. Large operators buy capacity on several systems and route traffic through different corridors. When one path fails, software and network engineers shift traffic onto surviving routes while the damaged cable is repaired.

Redundancy is not infinite. Losing several cables in the same corridor can reduce capacity, increase latency and create congestion. Countries connected by only one or two international systems face a much harsher risk. For them, a cable break can become a national communications crisis rather than a quiet maintenance event.

The security story is more complicated than sabotage

Submarine cables have become a geopolitical concern because they are critical, difficult to guard continuously and concentrated in certain routes. Governments worry about deliberate damage, surveillance, ownership and dependence on infrastructure controlled by foreign companies.

Those concerns are legitimate, but the public story often becomes distorted. A suspicious break is not automatically an attack. Determining cause requires physical inspection, vessel tracking, technical evidence and time. The less dramatic vulnerabilities, slow permitting, insufficient repair capacity, weak route diversity and accidental damage, can be just as consequential.

Resilience is therefore not achieved by militarising every kilometre of seabed. It comes from mapping risks, protecting routes near shore, coordinating with fishing and shipping industries, maintaining repair vessels, simplifying emergency permits and ensuring that countries have more than one way out.

The bottom line

The internet feels immaterial because the physical work has been made invisible. A message crosses borders in milliseconds, but only because surveyors chose a route, factories manufactured thousands of kilometres of cable, ships laid it across the seabed, landing stations power it and maintenance crews remain ready to lift it back from the ocean when it breaks.

Satellites, mobile networks and cloud platforms matter. But beneath them sits a quieter structure: glass, copper, steel, coastal buildings, marine law, repair agreements and the geography of the seabed.

The cloud has always had a seabed.