The Cable Ends at the Repair Ship

Roughly six hundred active submarine cables carry something in the order of ninety-nine per cent of intercontinental data traffic. That figure is repeated so often it has lost weight. It should be read the other way round. Almost everything that is called cloud, sovereignty, cross-border finance, remote work, video, model training, satellite command, and diplomacy passes through a glass thread the width of a garden hose, lying on a seabed that no one guards continuously and few states could realistically defend.

The public debate about that fact has fixated on the wrong layer. It circles the question of who lays the next cable: whether Google and Meta will dominate the trans-Pacific builds, whether HMN Tech (the successor to Huawei Marine) is edging into contracts once monopolised by SubCom, Alcatel Submarine Networks, and NEC, whether the United States will freeze Chinese-linked cables out of certain landing rights. This is not unimportant. But it is the glamorous layer. The harder layer, the one that will decide who keeps the world talking in a bad decade, is the repair fleet.

The world runs on about sixty ships

The global cable-ship fleet is very small. Estimates vary depending on how one counts specialised installation vessels versus dedicated repair ships, but the working number sits somewhere around sixty hulls. Of those, perhaps twenty to twenty-five are true repair vessels held on standby under regional maintenance agreements: the Atlantic Cable Maintenance and Repair Agreement, the Mediterranean pool, the Yokohama Zone, the South-East Asia and Indian Ocean pool, and a handful of others.

These are not sleek objects. Many are twenty-five to forty years old. They were built during the boom of the 1990s and early 2000s and have been in continuous service since, with life-extension refits standing in for the shipbuilding programme that never happened. The specialist labour behind them — jointers, cable engineers, dynamic-positioning bridge crews — is an aging guild. When one of these ships is unavailable, sea-state or crew or dry-dock, an entire ocean region loses a meaningful fraction of its repair capacity.

Every year the world experiences somewhere between one hundred and fifty and two hundred cable faults: fishing gear, anchor drag, subsea landslides, corrosion, and, increasingly, sabotage or plausible-deniable sabotage. In quiet years the fleet absorbs the load with slack. In a bad year, the queue lengthens, priority becomes political, and the sequence of repairs starts to resemble a strategic instrument.

The Red Sea, Taiwan, the Baltic

Recent seasons have made this legible. In February and March 2024, four of the great Asia–Europe cables running through the Bab el-Mandeb corridor — Seacom, TGN Eurasia, AAE-1, EIG — went down within days of each other, in a stretch of water made unsafe by the Houthi campaign against Red Sea shipping. Somewhere between a fifth and a quarter of Asia–Europe capacity vanished. Restoration was not fast. A cable ship had to be sourced, insured, escorted or timed against the war, and worked in seas where the underwriter was as much a decision-maker as the engineer. Some repairs stretched into months.

In early 2023, two cables serving Taiwan’s Matsu islands were cut within days, one by a Chinese fishing vessel, one by a cargo ship. Taiwan holds a modest domestic repair capability, and Matsu’s roughly fourteen thousand residents lived through weeks of degraded or microwave-only connectivity while a ship was pulled into position from elsewhere in the pool. The lesson was not that Taiwan is uniquely fragile. The lesson was that a small islanded population can be softened by anchor drags long before any formal blockade begins, because the healing rate of the network is bounded by a fleet, not by fiber.

In the Baltic, the sequence has been steady. The Balticconnector pipeline and the Estlink 2 power cable, the C-Lion1 fiber between Finland and Germany, the BCS East-West Interlink between Lithuania and Sweden: each incident with the same rhythm of tanker anchors, opaque flags, contested attribution, and repair times measured in weeks. NATO responded with the Baltic Sentry mission, more patrols, more surveillance. That helps with deterrence. It does nothing about the fact that when the cable actually parts, one small ship still has to arrive, hold position, and splice.

The industrial picture: four manufacturers, one serious builder

The chokepoint upstream is even narrower. Almost all serious submarine cable manufacture is done by four groups: SubCom in the United States, Alcatel Submarine Networks (now under French state control after being separated from Nokia), NEC in Japan, and HMN Tech in China. The corresponding cable-ship pool is operated by a similarly short list of firms: Global Marine, ASN Marine, Orange Marine, KDDI Cable Infinity Networks, Elettra, IT International Telecom, NTT World Engineering Marine, and, in China, SBSS.

This is where the trend line matters. The Western pool has not been renewed. Only a handful of new cable ships have been ordered globally in the last five years, and most of the classic installation and repair vessels are living on borrowed hull life. China, meanwhile, has been building. SBSS and adjacent Chinese operators have added new cable ships at a pace nobody else is matching, and HMN Tech’s share of new cable contracts has moved from single digits a few years ago into a range where it now competes credibly for major intercontinental routes, particularly in Africa, South-East Asia, and the Gulf.

A new cable ship costs somewhere in the region of one hundred and fifty to two hundred million dollars and takes three to four years to build. The specialist gear — dynamic positioning, cable tanks, plough, buoy handling, jointing lab — is niche. The crews that operate it cannot be conjured from a shipping-line labour pool. This is not something a strategy paper can accelerate in a crisis. It is a decade-scale industrial commitment or it is nothing.

Europe writes cloud papers; the pool ages

Europe has produced an impressive body of writing about digital sovereignty. Gaia-X, IPCEI-Cloud, the EU Data Act, the Cyber Resilience Act, national sovereign-cloud frameworks in France, Germany, Italy, and the Netherlands, and a steady output of Commission communications on undersea infrastructure. There is now a European Union Toolbox on Cable Security. There is a group of well-connected consultants who can recite it.

What Europe does not have is a serious cable-ship building programme, a real training pipeline for jointers, or a coordinated industrial strategy to keep ASN, Orange Marine, Elettra, and their partners solvent through the lean years between major intercontinental contracts. It has instead something closer to what continental capitalism has become in general: a large regulatory literature about strategic assets which then depend, in practice, on a handful of privately held companies operating aging platforms with no obvious successor.

This is the pattern one sees repeatedly at the material layer of the European economy. The paperwork about resilience is prolific. The physical infrastructure that would deliver resilience — ships, yards, transformers, chemical plants, cable factories, ammonia terminals — is either aging, sold, or slowly being outbid. Digital sovereignty was always going to end at the wrench, not at the white paper. The wrench, at the moment, is not in European hands.

Priority as a strategic instrument

The deeper point is what happens to the repair queue in a compressed decade.

Regional maintenance agreements were designed for a world of steady, distributed, mostly accidental faults. They allocate ships, share costs, and impose a discipline of neutrality among the customer telecoms and states in each zone. That discipline holds well in peacetime. It bends visibly when the faults concentrate: a war in the Red Sea, a sustained sabotage campaign in the Baltic, a Taiwan Strait crisis, a Gulf incident, a coordinated attack on trans-Atlantic routes.

When two or three theatres fail simultaneously, the pool cannot serve them all. Someone waits. The order in which those waits are resolved — which cable is spliced first, which customer’s traffic is restored first, which ship is diverted from which zone — will be a political decision dressed as an operational one. That is the layer at which digital sovereignty actually operates now, and it is barely mapped in public.

The country that owns the ships, the yards, the jointers, the cable stocks, and the specialist financing is the country whose traffic recovers first in a bad year. Everything else — residency laws, sovereign clouds, encryption regimes — is downstream of that fact.

What to actually build

Three moves would matter more than another sovereignty paper.

First, build ships. Europe, or a serious European-Japanese-Korean coalition, should place standing orders for a new generation of cable-repair and installation vessels, on the same institutional footing as naval programmes: multi-year, industrial-policy financed, with committed refit and replacement cycles. The current market will not build them on commercial terms in time. The current market did not build the last generation on commercial terms either; it inherited them.

Second, secure the crews. Jointing, dynamic-positioning bridge work, and cable engineering need visible career pipelines through maritime academies and technical schools, with public co-financing where the market will not carry it. A ship without a crew is a museum piece.

Third, be honest about the geography of dependence. Europe should stop pretending that trans-continental digital sovereignty is achievable by regulation alone and start negotiating explicit resilience arrangements with the countries that actually own repair capacity in each ocean: Japan, Korea, the Gulf, France, the UK, the US, and, wherever it can be done cleanly, non-aligned Asian operators. Some of these arrangements will be uncomfortable. All of them will be more useful than the current posture of moral commentary about Chinese fleets one has no serious alternative to.

The wrench

There is something almost comic about the gap between the register in which the digital economy is discussed and the physical reality that sustains it. The rhetoric is planetary. The instrument is a jointing van on the aft deck of a forty-year-old ship, a splice done in a rocking box in the middle of the Atlantic, and a captain deciding whether the weather window will hold.

This is the honest shape of the internet. It has never been ethereal. It is a fragile industrial artefact strung across the ocean floor, whose survival under stress depends on a small, aging, geographically concentrated fleet and a shrinking guild of specialists. Whoever renews that fleet writes the operating rules of the next decade’s digital order. Whoever does not will keep publishing sovereignty papers into a room whose lights are held on by someone else’s ship.

Sources

International Cable Protection Committee (ICPC), Submarine Cables and BBNJ and public fleet statistics — used for the count of active submarine cables, share of intercontinental data traffic, and the approximate size and structure of the global cable-ship fleet and maintenance-zone agreements.
https://www.iscpc.org/

TeleGeography, Submarine Cable Map and Cable Faults FAQ — used for the annual range of submarine cable faults, cable count, and the industrial concentration of manufacturers and cable-ship operators.
https://www.submarinecablemap.com/

Reuters and AP wire coverage of the February–March 2024 Red Sea cable cuts (Seacom, TGN Eurasia, AAE-1, EIG) — used for the identification of the affected cables and the estimated share of Asia–Europe capacity affected, together with reporting on the delayed repair timeline in the Houthi corridor.
https://www.reuters.com/world/middle-east/damage-submarine-cables-red-sea-causing-communications-disruption-hgc-says-2024-03-04/

Reuters reporting on the February 2023 Taiwan–Matsu cable cuts and their attribution to Chinese fishing and cargo vessels — used for the account of the Matsu incident and the multi-week restoration window.
https://www.reuters.com/world/asia-pacific/taiwan-suspects-chinese-ships-cutting-islands-internet-cables-2023-03-08/

European Commission, Recommendation on Secure and Resilient Submarine Cable Infrastructures (2024) and the EU Cable Security Toolbox — used for the summary of the European regulatory response and the observation that the paper output is not matched by cable-ship or cable-yard programmes.
https://digital-strategy.ec.europa.eu/en/library/recommendation-secure-and-resilient-submarine-cable-infrastructures

NATO Baltic Sentry mission statements and reporting on the Balticconnector, Estlink 2, C-Lion1, and BCS East-West Interlink incidents — used for the Baltic sabotage sequence and the response posture.
https://www.nato.int/cps/en/natohq/news_232122.htm

Nikkei Asia and SCMP reporting on HMN Tech (formerly Huawei Marine) and the Chinese cable-ship build-out through SBSS and related operators — used for the shift in HMN Tech’s share of new cable contracts and the pace of Chinese cable-ship additions.
https://asia.nikkei.com/Business/Telecommunication/China-s-HMN-Tech-eyes-share-of-global-subsea-cable-market