Last Updated on August 15, 2026 by Karl Thompson
Undersea Cables: The Infrastructure That Belongs to No Country
On the evening of 20 October 2022, shops in Lerwick stopped taking cards.
Two cables had gone. One, connecting Shetland to the Faroe Islands, had failed six days earlier. The second, running south from Shetland to the Scottish mainland, was damaged just after midnight. Together they took out most telephone and internet service across an archipelago of roughly 23,000 people. Banks closed. Cash machines stopped. Police Scotland declared a major incident and told residents that if they could not reach 999 from a landline or a mobile, they should flag down an emergency vehicle that was not already on a call, or walk into a police station, a hospital or a fire station and report the emergency in person.
The cause was almost certainly a fishing vessel. Possibly two, on separate occasions. Nobody was identified, nobody was charged, and nobody was ever going to be. The cable that failed is operated by a subsidiary of Faroese Telecom, a company in another country. The seabed it crosses is not Shetland’s, not Scotland’s and, for most of the route, not the United Kingdom’s either.
This article is part of a series on the systems that organise everyday life and are difficult to see from inside them. The series hub sets out the argument and lists the other articles.
What Are Undersea Cables, and How Many Are There?
A modern submarine cable is, for most of its journey, about as thick as a garden hose. Inside it are glass fibres roughly the diameter of a human hair, sheathed in plastic and, closer to shore, in steel armouring. Lasers fire down the glass at one end; receptors read the light at the other. Nearer the coast the cables are buried in the seabed, which is why you do not find them lying on beaches. In deep water they simply rest on the ocean floor.
TeleGeography, the industry’s mapping firm, tracks more than 600 active and planned systems as of 2026, amounting to over 1.5 million kilometres of cable in service. These carry between 95 and 99 per cent of the world’s intercontinental telecommunications data. Satellites are not a substitute at national scale: they can handle something in the order of 5 per cent of subsea cable capacity, and evidence given to Parliament in 2025 concluded that they are incapable of delivering data in the necessary volumes without a generational breakthrough.
Cables break constantly. On International Cable Protection Committee figures, there are roughly 200 faults a year worldwide, and two-thirds of them are caused by fishing gear and dragged anchors. Most pass unnoticed, because most large users spread their traffic across several systems and traffic reroutes automatically when one fails. Repairing them is slower than losing them. A repair ship will sail within a day of a fault being reported and travel at around twelve knots; a fault in the Irish Sea might mean two days’ transit and five days’ work, and a mid-Atlantic fault seven days’ transit and ten days’ repair. Bad weather can stretch that to a month.
That is the technical picture, and it is reassuring in the way that averages usually are. The interesting question is not how the cables work. It is who is responsible for them.
How Much Does Britain Depend on Them?
Around 64 cables land in the United Kingdom, of which about 45 are international systems and roughly 50 are thought to be active. Ministers have cited that number as evidence that there is little to worry about: more cables than an island of this size would normally have.
The committee that examined the question came to a more mixed conclusion, and the reason is worth following, because it is a pattern that recurs across every kind of network. Counting connections is not the same as measuring capacity. On evidence given to the Joint Committee on the National Security Strategy, two transatlantic cables carry around 75 per cent of the UK’s transatlantic capacity, and both land at Bude in Cornwall. Seven cables might carry around 93 per cent. Lowestoft houses five cables; Bude houses nine. A vessel travelling from Land’s End towards Aberystwyth would pass over around twenty cables. A ship moving down the Suffolk coast would cross eight major ones in a few hours.
The concentration is increasing rather than easing, because newer cables carry far more data than the ones they replace. As one witness put it, cutting a newer cable would be far more disruptive than cutting an older one. A system built for efficiency accumulates its traffic into fewer, fatter routes — which is another way of saying it accumulates its vulnerability into fewer, more identifiable places.
Onshore, the same funnelling happens again. Cables come ashore at landing stations, which are small privately operated buildings on the coast, and this is the point at which an ocean-crossing cable becomes the domestic fibre network, with the terrestrial links onward converging towards data centres. The committee found physical security across landing sites varies, that some could be rendered inoperable by unsophisticated sabotage, and that the terrestrial links between cable, station and data centre can be cut with ordinary tools.
None of this was designed. No authority decided that three-quarters of Britain’s transatlantic capacity should arrive in one Cornish village. Landing points were chosen over more than a century for reasons of geology, shallow approach, existing ducting and the fact that a previous cable had landed there — the same logic that put the Victorian telegraph terminus at Porthcurno a few miles down the coast. Each decision was locally sensible. The aggregate is a map that an adversary can read as easily as an engineer.
Who Has Jurisdiction Over a Cable in International Waters?
Here is the part that has no equivalent on land.
A cable in the deep ocean lies outside anybody’s territory. A state’s authority is organised territorially: it holds sovereignty over its land and its territorial sea, which extends twelve nautical miles from the coast. Beyond that lies the exclusive economic zone, where a coastal state has rights over resources but not general criminal jurisdiction over foreign ships, and beyond that the high seas, where ships are subject to the exclusive jurisdiction of the state whose flag they fly.
Cables run through all three. The legal address of a cable is therefore not the country it serves, and not the country whose data it carries, but — when something happens to it — the country where the offending ship happens to be registered.
Article 113 of the UN Convention on the Law of the Sea obliges every state to make it a punishable offence for a ship flying its flag, or a person under its jurisdiction, to break a submarine cable beneath the high seas wilfully or through culpable negligence. That obligation binds the flag state and essentially nobody else. Britain’s own foundational statute here is the Submarine Telegraph Act 1885, which set a penalty of £100 for damaging a cable through culpable negligence. Fines have been raised somewhat since. The 1885 Act was drafted for a world in which the plausible offender was a careless trawlerman whose own government would fine him, and it has never really been rewritten for any other kind of offender.
The consequences of that architecture stopped being theoretical on Christmas Day 2024.
The tanker Eagle S, registered in the Cook Islands, dragged its anchor across the floor of the Gulf of Finland and severed five cables, including the Estlink 2 power link between Finland and Estonia and four telecommunications cables. Finnish authorities boarded the vessel and escorted it to port. Prosecutors charged the captain and two officers.
On 3 October 2025 the Helsinki District Court ruled that the Finnish Criminal Code could not be applied — because of restrictions arising from the Law of the Sea Convention — even though Finland could be considered the place where the alleged crimes were committed. The damage had occurred inside Finland’s exclusive economic zone but outside its territorial waters. Jurisdiction, the court held, lay with the flag state or with the defendants’ countries of citizenship: the Cook Islands, Georgia and India. The Deputy Prosecutor General appealed six days later, and the case went to the Helsinki Court of Appeal.
Set alongside that, the Yi Peng 3: a Chinese-flagged cargo ship that dragged its anchor for around 300 kilometres in the Baltic in November 2024, cutting two cables. Chinese authorities permitted representatives from Sweden, Germany, Finland and Denmark to board and observe, but did not permit the Swedish public prosecutor aboard or a Swedish criminal investigation. The vessel left the Danish EEZ two days after the boarding. No crew were arrested. The Swedish Accident Investigation Board was unable to conclude whether the damage had been deliberate or accidental, and said so, citing the limitations on what it had been allowed to do.
The International Law Association, reviewing the position, found no examples of states served by a cable that have adopted national legislation explicitly criminalising damage to submarine cables outside their territorial waters by foreign vessels or foreign nationals. Not few examples. None.
This is a genuinely unusual situation, and it is worth being precise about what is unusual. It is not that the cables are unregulated — landing stations are designated critical national infrastructure, operators are licensed, the seabed inside British waters is managed by the Crown Estate. It is that the mechanism which assigns responsibility for an injury to a cable was designed to follow ships, and the thing that came to matter was wires. Nobody chose to place the internet’s physical layer beyond the reach of the states that depend on it. The law of the sea was settled for other reasons, mostly to do with freedom of navigation, and the cables were laid through the arrangement that already existed.
What Happens When a Repair Needs Permission?
The same gap appears again, in reverse, at the point of repair.
Repair ships are few, old and privately held. The two main North Atlantic maintenance consortiums have three vessels each and could each field two covering the region. Each ship can work on one break at a time; as one subsea engineering witness told Parliament, once breaks start accumulating you wait until a cable ship becomes free. Britain has no domestically owned repair capability at all. Global Marine, formerly a British company, has been bought by a Singaporean owner. The UK-based CS Sovereign was built in 1991, and with repair ships typically lasting 35 to 40 years, will retire around 2030. Repair ships take about three years to build and their crews about ten years to train.
The committee recommended the government acquire a genuinely sovereign repair ship by 2030. The comparison it drew was with the United States, which pays a private contractor roughly ten million dollars a year through its Cable Security Fleet to keep two US-flagged and crewed cable ships continuously available.
And where a cable lies in the waters of a state, the repair needs that state’s permission — which presupposes that there is an authority in a position to give it. In February 2024 an anchor from a vessel sunk during the Red Sea conflict severed three cables carrying traffic between Europe and Asia. Repairs were held up for months by a dispute over who could authorise work in Yemeni waters, because Yemen’s only telecommunications operator has one branch in Aden under the internationally recognised government and another in Sanaa under Houthi control, and the consortium running one of the cables included the operator. The Yemeni government granted permits for two cables and refused the third. Work on the third did not complete until roughly July.
There is no supervening authority in either direction. No one can compel a repair, and no one can compel permission for one. When a cable is cut, the sequence of events that follows is determined by which private vessels are free, which owners talk to which other owners, and which governments will sign which papers. Priorities, the committee heard, are set by cable owners and repair outfits talking to each other and deciding.
Who Loses When a Single Cable Fails?
For a country with sixty-four landings, a cut is a rerouting event. For a country with one, it is disconnection.
Tonga. On 15 January 2022 the eruption of Hunga Tonga–Hunga Haʻapai and the tsunami that followed broke the Tonga–Fiji cable — the only international cable serving the kingdom — along with its domestic extension. Internet service was restored on 22 February, thirty-eight days later, to a population of around 105,000, in the immediate aftermath of a disaster in which people were trying to establish whether relatives were alive. It was not the first time: a break in January 2019, most likely from a dragged anchor, left the country largely offline for more than ten days. Satellite links carried something like a tenth of the cable’s capacity in the interim.
The cost is only partly measurable in lost output. Remittances from Tongans working abroad are equivalent to something in the region of 40 to 50 per cent of national GDP depending on year and source — among the highest dependencies in the world. Those transfers move over the same connection. A cable fault in Tonga is not an inconvenience to consumers; it is an interruption to the mechanism through which a substantial share of household income arrives.
The mechanism that transferred the cost is worth stating plainly, because it is not neglect and it is not malice. Cables are built where expected traffic justifies the capital. A route serving 105,000 people justifies one cable. Redundancy for a small island economy would have to be paid for by someone who has no commercial reason to pay for it and no legal obligation to. Meanwhile the repair ship that eventually arrived came because a commercial maintenance agreement was activated, not because any authority had a duty to send it. The countries whose traffic reroutes automatically are not avoiding a cost that Tonga bears; they are simply the beneficiaries of a market that found it worth building them a second cable.
The UK’s outlying islands. The same asymmetry exists inside Britain, at smaller scale. The 2022 Shetland incident is one instance. In July 2025 the cable connecting Orkney and Banff was damaged: hundreds of customers were affected, the switchboard at Balfour Hospital went down — 999 calls were not affected — and some businesses reported internet-based systems not working for days. Parliament’s committee identified outlying islands, alongside the financial sector and military communications, as the UK’s particular points of vulnerability.
What Shetland lost first, in 2022, was the ability to pay for things: the payment networks that shops and cash machines depend on run over the same connection, and a card terminal that cannot reach an authorising bank is an ornament. Evidence to Parliament noted that some card systems have offline backup functionality but that it is unclear how well this would hold up in a protracted outage, and pointed to the Bank of Finland’s work on offline card payments as a model worth studying.
Neither Shetland nor Tonga was failed by an institution that could be named. That is the difficulty. There is no regulator with a remit to ensure that a place has two cables rather than one, because a cable is an international commercial undertaking crossing waters that belong to nobody, and the state whose citizens depend on it has jurisdiction over roughly the first twelve miles.
The Sea Where the Settlement Was Never Made
Over the twentieth century most infrastructure arrived at a settlement of some kind. Railways, water, electricity and telephony all ended up with an owner, a regulator, a duty and someone who could be summoned to explain a failure. The arrangements differ by country and they are frequently unsatisfactory, but they exist, and they exist because the infrastructure sits on territory and territory is where state authority operates.
The cables never reached that settlement, and the reason is not that anyone resisted it. It is that the framework governing the deep ocean was built to answer a different question — who may sail where, and whose law applies to a ship — and the answer it gave was: the flag. When telegraph cables began crossing oceans in the nineteenth century they were slotted into that framework, on the assumption that the only realistic offender was a fisherman and the only realistic remedy a fine levied by his own government. The 1884 convention and the 1885 Act encode that assumption. Article 113 carries it forward into 1982. And then almost the entire communications capacity of the world was laid along the seabed, and nothing about the assumption was revisited.
So a British archipelago’s connection to the mainland can be severed by an unidentified vessel with no consequence to anyone; a NATO member can board a tanker that has dragged an anchor across five of its cables and find that its own criminal law does not reach the seabed where the damage was done; and the repair of a cable serving three continents can wait five months on a permit from a ministry whose control of its own coastline is disputed. These are not failures of enforcement. They are the system working exactly as its rules provide, applied to a use nobody had in mind when the rules were written.
That is the thing about a mechanism designed for one purpose and inherited for another. It goes on allocating — responsibility, priority, repair, redress — long after the question it was built to answer has stopped being the question that matters.
More in this series
Previously: How the UK Electricity Grid Works — the one resource every other system needs, and the queue that decided who got it.
Next: Semiconductors — how a global division of labour gave the most leverage to the stage that earns the least. (Placeholder — drafted, not yet published. Point at the hub until it goes live.)
All articles in this series: How Modern Society Works.
References
References — Undersea Cables
Paste-ready. Formatted as a reader-facing list rather than a working table. Read the note at the end before publishing: three entries need URLs I do not have.
References
Parliamentary and official
Joint Committee on the National Security Strategy, Undersea cables: the ultimate global chokepoint?, HC 723 / HL Paper 179, 19 September 2025. The principal source for this article: UK landing points and their concentration, fault causes and repair times, the absence of British repair capacity, the legal position under the Submarine Telegraph Act 1885 and Article 113 of the UN Convention on the Law of the Sea, the Yi Peng 3 incident, and the 2022 Shetland and 2025 Orkney outages.
Joint Committee on the National Security Strategy, Government response, 1 December 2025.
Office of the Prosecutor General, Finland, Prosecutor appeals judgement in the Eagle S case, 9 October 2025. The Helsinki District Court’s ruling of 3 October 2025 that Finnish criminal law could not be applied to damage caused inside Finland’s exclusive economic zone, and the Deputy Prosecutor General’s appeal against it.
Industry
TeleGeography, Submarine Cable Frequently Asked Questions. Cable counts and total length, annual fault numbers, and the shift toward content-provider ownership. TeleGeography maps and sells data on this industry; its figures are used here as the industry’s own account.
International law
United Nations Convention on the Law of the Sea, 1982, Article 113. Submarine Telegraph Act 1885. Both discussed in the Joint Committee report above.
International Law Association, Committee on Submarine Cables and Pipelines under International Law, findings on the absence of national legislation criminalising damage to cables outside territorial waters — quoted in the Joint Committee report above.