After the Cable Breaks: How Repair Capacity Shapes a State's Strategic Choices

On 17 and 18 November 2024, two telecom cables in the Baltic Sea were damaged less than a day apart — one linking Lithuania with the Swedish island of Gotland, the other linking Finland and Germany. The incidents fed European anxiety about critical infrastructure. Yet according to the International Cable Protection Committee (ICPC), neither caused a disruption or interruption to service: repair ships were quickly deployed, both cables were fixed within two weeks, and incidents of this kind are “generally considered unremarkable”.

“No interruption” describes what users experienced. C-Lion1, the Finland–Germany cable, was out of action for more than ten days, until the evening of 28 November; its operator said alternative routes and a rapid repair kept the disruption to Finnish society minimal.

Damage to a cable and a communications crisis do not necessarily arrive together. Between them sit a few things that rarely get attention: how much spare capacity is left, how quickly a repair ship can get there, and which services must be kept running while the link is down.

In normal times these are matters of network operations. In a crisis they shape what a government can do. If communications hold, the authorities can wait for an investigation before deciding how to respond. If banks, ministries and the public are already cut off, they may be pushed into taking a position before the evidence is in.

Finding out who did it, and keeping the lines open: two jobs at once

Europe’s response covered both military surveillance and civilian infrastructure. On 14 January 2025, at a summit in Helsinki, NATO launched “Baltic Sentry”, a military activity using frigates, maritime patrol aircraft and a small fleet of naval drones, and integrating allies’ national surveillance assets.

The EU’s Action Plan on Cable Security, published in February 2025, called maintenance and repair vessels “a major bottleneck” for recovering from an incident. In February 2026 the European Commission amended the Connecting Europe Facility (CEF) digital work programme to allocate €347 million to strategic submarine cable projects. Of that, €60 million across two calls funds repair modules that can be fitted to existing ships; the first €20 million call is a Baltic pilot.

All of these measures face the same problem. After a cable is damaged, a government has to investigate who is responsible — and keep communications running before that investigation is finished. The two jobs need different equipment, legal powers and time: one relies on patrols, evidence and courts, the other on repair ships, backup routes and permits. Both shape how a crisis unfolds.

Finding the ship is not finding who gave the order

Establishing responsibility means clearing at least three separate bars: which ship caused the damage, whether the crew acted deliberately, and whether a state directed or took part. Each needs different evidence, and clearing one does not clear the next. A ship’s flag, its owner or its crew’s nationality cannot by itself supply the evidence for the later ones.

Recent Baltic cases have mostly stopped at the first two. In April 2025 Sweden’s accident investigation authority found that the Chinese-flagged bulk carrier Yi Peng 3 had cut the two cables by dragging its anchor for a day and a half and some 180 nautical miles — but could not conclude that it had done so on purpose. Its investigators were allowed aboard only after more than a month, and never saw the ship’s electronic data. The Vezhen, which broke a Latvia–Sweden cable in January 2025, was confirmed by Swedish prosecutors as the ship responsible; the case was closed as an accident that October.

On Christmas Day 2024 the tanker Eagle S dragged its anchor about 90 km, cutting five cables in the Gulf of Finland, a power cable among them. Finnish prosecutors charged the captain and the first and second officers — not with deliberately damaging the cables, but with knowingly neglecting their duties despite a defective anchor windlass. The Helsinki District Court dismissed the case in October 2025, ruling that Finnish criminal law did not apply; in August 2026 the Court of Appeal overturned that ruling and sent the case back. It is not yet final.

The head of Finland’s security intelligence service, Juha Martelius, put it plainly: a criminal investigation can establish whether there was intent, negligence or carelessness, but whether a state actor was behind it may not be established that way. The EU’s Action Plan on Cable Security likewise concedes that one of the main challenges is establishing intentionality, since such actions are most often prepared with the very objective of obstructing attribution. That is the signature of grey-zone operations: the damage is real, the blame is blurred.

RAND’s October 2026 report notes in the same vein that most of the Baltic incidents “still lack formal attribution even with identified boats”; establishing what happened, who was involved and why can take weeks, months or years, and may never be settled.

RAND’s deterrence framework leaves repair out

The study the RAND Corporation published on 20 October organises decisions in undersea competition around three interdependent variables: the level of attribution, the choice to reveal or conceal capabilities or knowledge, and actions ranging from passive defence to escalation. Because undersea activity is hard to observe, actors can hide who they are and what they intend; a government that has the means to respond may be unable to use them without knowing whom to respond to.

That captures the central difficulty of seabed security. But at one point the report’s decision matrix sits awkwardly with practice. It maps a publicly announced passive or active defensive response to a specific incident onto “clear” attribution — evidence confirmed by multiple sources and methods that cannot be repudiated — while defensive steps and planning with no link to an incident need no attribution at all.

NATO did not wait for that threshold. Announcing Baltic Sentry, its Secretary General said recent sabotage had damaged energy and communication cables, without naming any state. The EU chose the Baltic for its repair pilot because of a rise in disruptions “suggesting these critical infrastructures might be subject to hostile acts.” Protective measures are aimed at no one, so they need not accuse anyone. The same goes for publicly arranging repair ships, adding redundancy and reporting on an investigation’s progress.

The bigger gap is repair itself. In mapping capabilities, the report leaves recovery systems out “for simplicity”, and its decision matrix has no column for repair. Yet it grants that a capability to rapidly provision resources and perform repairs “may have a deterrent effect, similar to a ‘fleet in being’” — it limits what an adversary expects to gain from spending resources on an attack. It even lists contracting repair ships on standby among the non-military options. Repair, in other words, is one of the few responses that can be taken openly without attribution — and it sits outside the framework.

The authors call it a rough, initial framework whose thresholds still need mapping to real authorities and testing against real-world scenarios and historical cases. An exploratory study aimed chiefly at the US Navy, it does not show that its thresholds reduce misjudgement or improve deterrence. The Baltic and Taiwanese experience is a place to test it.

Keeping services running buys time for the investigation

From a crisis-management point of view, keeping services running probably leaves a government more time for investigation and diplomacy. If banks, ministries and the public can still communicate, the authorities are under less pressure to respond forcefully before the evidence is in. RAND itself observes that because establishing who and why can take so long, they may not be that important — whereas establishing what happened and where certainly is.

In the language of deterrence theory, repair and redundancy work by denial rather than punishment: they do not raise the price a saboteur pays, they reduce what sabotage achieves. That is the strategic value repair may have — but the available data are not yet enough to show that it actually prevents deliberate attacks.

The everyday experience of the cable industry is also a warning against reading every fault as military confrontation. According to the ICPC, there are roughly 150 to 200 telecom cable faults worldwide each year, and 70–80% are caused by accidental human activity, chiefly fishing and ships’ anchors; events that cut several cables at once now happen seven or eight times a year. A statistical analysis of ICPC records therefore stresses that multiple simultaneous faults are “a known and recurring operational reality”, not something to be attributed automatically to hostile action. Base rates like these frame an investigation; they cannot decide whether any particular case was an accident or deliberate.

Matsu: what backup saved, and what it could not

That some Baltic incidents left services untouched does not mean every place can absorb a cut. Taiwan’s outlying islands show that backup can protect essential communications without preserving normal service.

In February 2023 the two cables linking Taiwan’s main island to Matsu failed one after the other: Taiwan–Matsu No. 2, running from Tamsui to Dongyin, on the evening of the 2nd, and No. 3, from Taoyuan to Nangan, at midday on the 8th, apparently damaged by a cargo ship. Chunghwa Telecom said that when the first went down, traffic switched automatically to the second; once both were cut, a microwave system took over. Voice calls and government, national-security and essential-service circuits carried on unaffected, but mobile and broadband internet and the MOD television service were congested by the limited microwave bandwidth.

The microwave link saved the most critical layer, not the whole. Even after an emergency expansion to 3.8 Gbps in March, it could back up only about 42% of traffic. An international repair ship originally due no earlier than 20 April was persuaded to come sooner, and No. 3 was restored at the end of March, ending some 50 days of congestion; No. 2 took four months and 23 days, a delay that Taiwan’s deputy digital minister later attributed partly to “interference from Chinese coast guard ships”.

That difference is why cable security cannot be assessed by counting breaks. For a network, which services are affected, how much capacity remains and how long the outage lasts say far more about the real impact than the number of damaged cables. Even without an immediate outage, backup gets used up and repair bills rise — and a second incident before the first cable is fixed would hit much harder.

For Taiwan this is not a rare risk. Statistics compiled from ICPC records show that in the East Asian waters around China, Taiwan and Malaysia some areas see roughly one fault a month, and the Asia-Pacific now accounts for more than 40% of repairs worldwide; the Baltic sees one to four faults a year. Matsu’s microwave backup was expanded to carry all of its traffic by the end of 2023, and a fourth Taiwan–Matsu cable entered service in September 2026.

Most repair time is lost before the ship sails

Repair speed turns on a factor easily lost in discussions of military deployments: when the repair ship can leave port.

A study published in September 2026 in a NATO Science for Peace and Security volume by Andy Palmer-Felgate, a cable engineer writing for the ICPC, drew on ICPC repair records from 2007 to 2025. It found that it takes on average about 22.5 days from a fault to a repair ship sailing, plus about 7.5 days to reach the repair ground; the longest recorded delay was 947 days. A few extreme cases pull the average up: an earlier version of the same database, cited by the ITU, puts the median at just 8 days.

The trend is more worrying. The International Advisory Body on Submarine Cable Resilience, set up by the ITU and the ICPC, reports that the global average time from fault to start of repair rose from under 20 days in 2012 to over 50 days in 2024. Repair volumes and the number of available ships held roughly steady over the period; the main reason given is permitting and other regulation. In parts of Asia and the Middle East, a repair permit can take months, and some states demand a bond for the full value of the ship, require it to be imported and re-flagged locally, or put it through lengthy security and sanctions clearances.

Conversely, the Netherlands, Belgium, France, the UK and South Africa pre-permit repair ships, which can often sail within a week and finish the whole job within a fortnight. The advisory body’s report likewise recommends handling repair permits separately from permits for new cables, and giving operators a single government interface.

The difference has direct budget implications. Where the bottleneck is permits and inter-agency coordination, settling the procedures in peacetime does more than sending extra ships in a crisis — and costs far less. Only where the bottleneck is ship schedules and equipment does more repair capacity need to be bought. The two call for different budgets, held by different ministries.

Permits are hardest to obtain in conflict zones. In February 2024 three cables in the Red Sea were severed by the dragging anchor of a cargo ship that Houthi militants had attacked and that later sank. Repairs took months of negotiation between the cable operators and the two factions that control Yemen — the internationally recognised government and the Houthi-backed authorities — and the first cable came back online only in mid-July. On the Red Sea itself, see The Houthis Push Toward Bab el-Mandeb as the Middle East’s Two Shipping Gateways Come Under Pressure at Once.

Who pays for a repair ship to wait?

Ships are scarce too. The advisory body estimates that around 70 vessels worldwide could repair cables in 2025, but many were mainly laying new cable or no longer kept in active service; only some 20 to 25 were dedicated to repair under maintenance agreements or long-term charter. A significant share of the world’s repair fleet was built in the 1980s and 1990s.

Cable protection also contains a commercial paradox. Palmer-Felgate notes that Northwest Europe cut its annual faults from 40–45 to around 10–15 by burying cables deeper and retiring older systems. But fewer faults weaken the business case for keeping repair ships on standby there, and operators may move them to Asia, where faults are rising. A region with only one or two ships on standby, he warns, is significantly more vulnerable than one with four. Success in protecting cables does not automatically bring more emergency repair capacity.

The mechanism still needs testing against data from individual markets; fewer incidents cannot simply be equated with ships leaving. But it makes a case for public spending: what a government pays for is capacity available the moment an incident happens, whose value never fully shows up in everyday repair demand. The industry’s view is that rather than governments building their own fleets, maintenance agreements, capacity retainers or strategic subsidies can keep commercially operated ships in strategically important waters.

The EU’s repair modules take another route. Rather than buying ships, they keep cable-repair equipment at ports or shipyards, ready to be fitted to existing vessels in an emergency. The first €20 million call, a Baltic pilot, is open only to public bodies with an emergency-response mandate — civil protection agencies, coastguards, navies — and later phases are to extend to the Mediterranean and the Atlantic. A multi-purpose “EU Cable Vessels Reserve Fleet” appears in the action plan as a medium-term idea.

International cooperation also decides whether this capacity is there when needed. Operators have long run regional maintenance agreements that pool standby ships and spare-cable depots; the advisory body recommends extending them, especially in Africa, small island developing states and parts of Latin America, where repair resources are thin. For governments, the negotiations involve who pays, how ships are allocated in an emergency, and whether a repair ship can get its permits in time. An ally’s security commitment can shape how fast a country recovers in a crisis through precisely these practical arrangements.

Taiwan shows how this calculation plays out. It has no dedicated cable repair ship: Chunghwa Telecom belongs to two international repair zones, which between them have six cable ships that can come to Taiwan, and domestic cables queue behind international ones for repairs. The company’s chairman has said that building its own repair fleet would probably not make economic sense. The digital ministry has chosen to pay for speed instead: since the second half of 2025, operators that repair a cable faster than the historical average receive a subsidy, and the faster the repair, the larger it is. Domestic cables have taken an average of 108 days to repair; the first three accelerated projects aim to cut that to between 64 and 86 days.

Another cable is not always another way out

New cables involve a similar trade-off. The advisory body notes that international cables are heavily concentrated in a few chokepoints — the Luzon Strait, the Strait of Malacca, the Red Sea — where a single natural hazard or geopolitical disruption can take out several systems at once, and recommends greater geographic diversity of routes. If a new cable shares the same corridor, landing station or onshore connection, it adds capacity, but one event can still cut several links together: a textbook single point of failure.

As investments, extra capacity and lower common-failure risk have to be weighed separately. More diverse routes can cost more to build and need permits and cooperation from more countries. Where cables go is therefore shaped at once by commercial demand, geography and diplomacy. How much satellites can fill in is covered in the explainer If the Cable Breaks, Can Satellites Take Over?

What repair cannot do

None of this makes patrols, investigation and enforcement less necessary. An actor whose aim is to keep raising repair costs or draining backup can achieve part of it without ever causing a full outage — and, as RAND notes, the supply chain that makes repair possible is itself a target. Repair capacity reduces the damage; on its own it cannot settle responsibility or stop repeated sabotage.

The courts can reach ships and crews; they rarely reach states. The Chinese captain of the Hong Tai 58 was sentenced by the Tainan District Court to three years for damaging a Taiwan–Penghu cable with what the court called “indirect intent” — Taiwan’s first successful prosecution for cable damage. But prosecutors did not bring the aggravated charge of endangering national security, and found no communications pointing to security or political motives. When NATO’s Secretary General launched Baltic Sentry, the consequences he warned of — possible boarding, impounding and arrest — were likewise addressed to ship captains.

The Eagle S case exposes a less obvious tension too. In bringing charges, prosecutors noted that although the damage posed a serious risk to Finland’s energy supply and telecommunications, services were secured through alternative connections. The District Court then held that the incident had not caused the kind of consequences for energy supply or telecommunications that the offence of criminal mischief requires. That judgment has since been overturned, but it holds a lesson for policymakers: where an offence is defined by disruption to supply, the better the resilience, the harder prosecution may become.

Three questions for measuring results

Europe’s patrols and repair investment, and Matsu’s backup, offer different ways of measuring the same thing. The effectiveness of cable security can be judged by three questions: how long critical services held, how soon they were restored, and whether the government kept room to respond before responsibility was established.

The answers will decide whether damage to undersea infrastructure is just a repair bill — or something that further narrows a country’s strategic choices.