Submarine communications cable is a fibre-optic cable laid on the sea floor to carry high speed data between countries. [1]
Roughly the width of a garden hose, it contains a small number of hair-thin glass fibres, a power conductor that feeds optical amplifiers spaced along its length, and layers of steel wire and polyethylene that protect it from anchors, trawlers and pressure. There are about 608 of these systems in service, totalling roughly 1.6 million kilometres, and they carry more than 99% of intercontinental data traffic — which makes an infrastructure almost nobody sees the single most important piece of the global internet.[2][3]
Why the cable, not the satellite
The economics are not close. A cable carries far more data at a fraction of the cost of a satellite link, and latency over fibre is lower and more predictable than over a geostationary hop. Statistics from the US Federal Communications Commission put satellites at just 0.37% of US international capacity, and while no precise global figure exists, the share is small everywhere outside remote and maritime use. Satellites are good at reaching places that are not yet wired and at broadcasting from one source to many; they are not a substitute for a fibre pair.[4]
How often they break, and who fixes them
Cable faults are routine: on average about 200 a year, according to the International Cable Protection Committee, with roughly two-thirds caused by fishing vessels and ships dragging anchors, plus earthquakes and other environmental damage. Most go unnoticed because operators spread traffic across multiple cables, so a cut degrades a network rather than stopping it. The shark-bite story is largely myth — fish bites, a category that includes sharks, accounted for zero faults between 2007 and 2014.[5]
Repair is the weak link. The ITU’s International Advisory Body on Submarine Cable Resilience found in 2026 that restoration times are lengthening in some regions, for reasons that have little to do with the sea: complex permitting processes, fragmented regulatory frameworks, and a limited supply of repair vessels and spares. Its recommendations focus on expanding repair assets across more companies and countries, cutting regulatory barriers, and strengthening shared vessel and spare-part arrangements within maintenance zones.[6]
The fleet itself is thin and ageing. Orange Marine is building two new cable ships to join Sophie Germain, delivered in 2023, replacing vessels built in 1983 and 1987, with deliveries expected in 2028 and 2029 — a timetable that says a great deal about how slowly capacity in this industry moves.[7]
A cable network is not designed to never fail. It is designed so that a failure is somebody else’s problem for a few weeks — which works only while there is spare capacity on another route and a ship available to sail.
Chokepoints and politics
The network is far less distributed than a map suggests. Traffic funnels through narrow corridors, and the ITU body identifies this geographic concentration as a systemic vulnerability, because a disruption in one corridor can affect several regions at once. The Red Sea is the clearest case: seventeen cables there carry roughly 18% of global data traffic between Asia, Africa and Europe. In February 2024 a Houthi missile attack led a vessel to drag its anchor across the seabed, cutting three major cables and disrupting about 25% of regional internet traffic; some Red Sea repairs have taken as long as six months. TeleGeography has reported that the Iran war delayed installation of the Gulf portions of major systems including 2Africa and SEA-ME-WE-6, with resumption timing unclear, and Meta has paused part of the 2Africa build in the Persian Gulf.[8][9]
The Baltic has produced a different kind of problem: repeated damage to cables and gas pipelines, followed by the interception of cargo vessels by Finnish and Estonian authorities, has moved subsea infrastructure from a telecoms question to a defence one. The regional response has been layered — prevention, detection and repair — with repair identified as the thinnest layer.[10]
Who owns the pipes
Ownership has shifted decisively. Cables were once built by consortia of telecom carriers sharing cost and capacity; now the largest content companies commission their own. More than half the length of planned cables is part-owned by Google, Meta, Amazon or Microsoft, and Meta’s Project Waterworth, announced in February 2025, is the emblem of the change — a hyperscaler deciding it will own the route rather than buy passage on somebody else’s.[11][12]
India’s exposure
India’s position is one of the most concentrated of any large economy. It had about 309 Tbps of lit subsea capacity at the end of 2025, of which 268 Tbps was activated, spread over roughly 18 systems and 23 landings in Mumbai, Chennai, Kochi, Tuticorin and Thiruvananthapuram. Mumbai and Chennai remain the hubs, and the Mumbai Metropolitan Region alone handles around 95% of the country’s international bandwidth.[13]
The concentration has a precise geography: at least 13 of India’s cables come ashore within about six kilometres of one another near Versova in north-west Mumbai, carrying as much as 95% of the country’s westward bandwidth to Europe, Africa and West Asia. Traffic can be rerouted through Chennai and Kochi, so a cut is not a blackout, but even small increases in latency are costly for banks, exchanges, cloud providers and AI workloads. India accounts for roughly 20% of global data consumption but is connected to only about 3% of the world’s submarine cables, and its 21 cable landing stations are about 1% of the global total.[14][15]
The build-out is substantial and mostly aimed at the same coastline. Four systems were under commissioning as of March 2026 — India Europe Xpress and the Raman cable in Mumbai, SEA-ME-WE-6 in Mumbai and Chennai, and 2Africa in Mumbai — with three more planned: Tata Communications’ Wyasa in Chennai, Reliance Jio’s Indian Ocean Cable at Raigad and Sify’s WABAN in Mumbai. Google announced America-India Connect in February 2026 with a new gateway at Visakhapatnam and three new paths linking India to Singapore, South Africa and Australia, and the 3,600 km I-2SEA system with Microsoft, Singtel, Lightstorm and Tata Communications is due in 2029. Demand is being driven by data centres: India’s capacity grew from about 540 MW in 2019 to roughly 1.4–1.7 GW, with projections of 6–8 GW by 2030 and subsea bandwidth demand growing at about 26% a year.[16][17]
The next decade
TeleGeography’s ten-year forecast expects more than 1,010,000 kilometres of new cable between 2026 and 2035, with the Northeast Atlantic, Northeast and Northwest Pacific and Southwest Pacific each adding over 100,000 km, alongside the retirement of more than 450,000 km of older systems. The direction of travel is towards more routes, more landing points and more owners — partly for capacity, and increasingly because a single concentrated corridor has become an unacceptable risk for the companies whose businesses run through it.[18][19]
Select a year and month above to explore cold archived discussions.
Please select an archived topic above to view the discussion logs.