On Christmas Day 2024, a tired tanker called the Eagle S moved through the Gulf of Finland with its anchor down. The anchor — eleven tonnes of steel — dragged along the seabed for about ninety kilometres. It caught the Estlink 2 power cable that ties Finland to Estonia, then four telecommunications lines, before a Finnish patrol vessel boarded the ship and escorted it into custody. The Eagle S was a "shadow fleet" tanker, the kind that moves sanctioned Russian oil under flags of convenience. Finland's lead investigator said plainly that he was looking at "grave sabotage."
Ten months later, a court in Helsinki let the crew walk. Not because anyone decided the cables had snapped by accident, but because the judges concluded they could not apply Finnish law to a ship in international waters, and prosecutors could not prove the anchor had been dropped on purpose. The case was reclassified as an "incident of navigation." The men went home.
That is the strange territory the world now occupies. The seabed carries nearly everything that matters, the damage to it is mounting, and almost no one can say for certain whether the damage is malice or carelessness — or do much about it either way. Most of us never picture the ocean floor at all. It is where the internet actually lives.
The Plumbing Under the Cloud
When people say their photos are "in the cloud," they mean a building somewhere full of hard drives, and a cable on the bottom of the sea connecting it to them. Satellites handle a sliver of long-distance traffic. The rest — by the figure that TeleGeography, the firm that maps the network, and the International Telecommunication Union both put at roughly 99 percent of international data — travels as pulses of light through glass laid across the ocean floor. (For the mechanics of how those pulses find their way to you, see our explainer on submarine cables and internet routing.)
The scale is hard to hold in the mind, and the physical reality is stranger still. More than 1.5 million kilometres of cable are in service today, across roughly 500 systems, with hundreds more planned or under construction. Yet for most of its journey across an ocean, a deep-sea cable is no thicker than a garden hose. The fibres inside that carry the light are each about the width of a human hair. Some of those cables rest deeper beneath the waves than Everest is tall.
There is no backup ocean. When a cable breaks, a ship has to go find the exact spot, which it does by firing a pulse of light down the fibre and timing how long the snapped glass takes to bounce it back. In deep water the ship trails a grapnel — a hook, more or less unchanged in principle since the nineteenth century — drags it along the bottom until it snags the cable, cuts it, and hauls the severed ends to the surface. A jointer then splices in fresh cable by hand, fusing hair-thin fibres in a process where a single speck of dust ruins the joint and forces a restart. One splice can take the better part of a day.
The whole planet depends on roughly 62 of these specialised ships, fewer than twenty of them dedicated to repair, and the fleet is old. A 2025 study by TeleGeography and Infra-Analytics warned that by 2040 about two-thirds of the maintenance vessels will have aged out of service, and that keeping pace will cost around US$3 billion in new ships. Mike Constable, who runs the consultancy Infra-Analytics, compared the situation to buying an expensive car and skipping the insurance: huge sums going into the cables, almost nothing into protecting them.
Anchors, Not Sharks
Ask people what threatens an undersea cable and a surprising number will say sharks. It is a good story. It is also, by the data, nonsense: between 2007 and 2014, fish bites — sharks included — accounted for exactly zero cable faults.
The real culprits are mundane. The network suffers somewhere between 150 and 200 faults a year, roughly three or four a week, and the International Cable Protection Committee attributes 70 to 80 percent of them to ordinary human clumsiness: a trawler's net catching the line, a ship's anchor dragging where it shouldn't. Earthquakes and undersea landslides take most of the rest. Below about a thousand metres, faults are nearly always natural. In the shallows, they are nearly always us.
For decades this was treated as a maintenance problem rather than a security one, and the network was built accordingly. Operators spread their traffic across many cables on the principle of safety in numbers, so that one break reroutes rather than blacks out. When two Baltic cables were cut in November 2024, analysts at Cloudflare and the regional internet registry found barely a ripple in regional connectivity. The system was designed to shrug off the occasional anchor.
What has changed is not the failure rate, which has held remarkably steady even as the network has grown by half. What has changed is the suspicion. The same committee that tracks the faults notes something worth sitting with: deliberate sabotage accounts for less than one percent of cable damage in any given year, and there has not been a verified case of state-sponsored cable sabotage since the Second World War. That is the backdrop against which the past two years have felt like a rupture.
When the Anchor Is on Purpose
A pattern emerged that was hard to wave away. In October 2023, a gas pipeline and two telecom cables between Finland and Estonia were damaged; investigators recovered an anchor from the seabed and traced it to a Chinese-owned container ship. Over a single weekend in November 2024, two more Baltic cables went down on consecutive days — a Lithuania-to-Sweden link, then C-Lion1, Finland's only direct fibre to central Europe. A Chinese bulk carrier, the Yi Peng 3, fresh from a Russian port, became the prime suspect after investigators concluded its anchor had scraped roughly a hundred miles of seabed across both breaks. NATO warships shadowed it for weeks off Denmark. It was eventually allowed to sail on.
Germany's defence minister, Boris Pistorius, said what many were thinking: "No one believes that the cables were accidentally damaged. I also don't want to believe that the ships' anchors caused the damage by accident." By the count of researchers at the University of Washington, about ten cables in the Baltic region have been cut since 2022, seven of them between that November and the following January. NATO responded in January 2025 with Operation Baltic Sentry — frigates, patrol aircraft, a small swarm of naval drones — and a warning from Secretary General Mark Rutte that captains who threatened the infrastructure could expect "boarding, impounding, and arrest."
The same anxiety runs through two other waters. In the Red Sea in late February 2024, three cables were faulted in the narrow corridor between Jeddah and Djibouti — a corridor that carries something like 90 percent of the data moving between Europe and Asia. The likely cause, never officially confirmed, was the anchor of the Rubymar, a cargo ship struck by a Houthi missile, abandoned by its crew, and left to drift for days before it sank. A Hong Kong carrier estimated a quarter of Asia–Europe traffic was disrupted; because the network reroutes, most users never noticed.
Near Taiwan, the story is older and more raw than the headlines suggest. In February 2023, both cables linking the Matsu Islands to the Taiwanese mainland were severed six days apart, by a Chinese fishing vessel and then a Chinese cargo ship, leaving the islands' 14,000 residents on degraded backup radio for more than fifty days. Taipei stopped short of calling it deliberate. The prominent cuts of early 2025, often lumped together with Matsu, in fact struck elsewhere — an international cable off the northeast coast in January, a line to the Penghu Islands in February, the latter prompting Taiwan's coast guard to board a Togo-flagged, Chinese-crewed ship whose name had changed three times across three registries.
That last detail is the heart of the problem. Proving intent at sea is brutally hard, jurisdiction is a thicket, and a vessel that wants to be hard to pin down can simply change its flag and its name. The Eagle S verdict was the lesson stated out loud.
The Hyperscalers Go to Sea
While governments worried about who might cut the cables, four American companies quietly became the people who own them.
For most of the cable era, the lines were built by consortiums of telephone companies and sold capacity to everyone else. Then the content providers — Google, Meta, Amazon, Microsoft — went from being the network's biggest customers to being its biggest builders. The crossover came in 2017, when they overtook the traditional carriers as the largest users of international capacity. By 2024, TeleGeography reckons, those four accounted for nearly three-quarters of all the bandwidth in use worldwide, and more than 80 percent on the busiest transoceanic routes. Much of their traffic now runs on private fibre that the rest of the industry cannot even see.
The flagship of this shift is a cable Meta announced in February 2025 called Project Waterworth. It is planned to run more than 50,000 kilometres — longer than the circumference of the Earth — connecting the United States, India, Brazil, and South Africa, laid as deep as 7,000 metres with twenty-four fibre pairs where most cables carry eight to sixteen. Meta calls it a "multi-billion-dollar" build; the press has reported a figure near US$10 billion and a finish around 2030. It would be Meta's first ocean-spanning cable owned outright, after a decade of co-builds, and the explicit justification is artificial intelligence — the need to shuttle vast volumes of data between data centres on different continents.
Meta is not alone. In November 2025 it finished the core of 2Africa, a 45,000-kilometre system that wraps the entire African continent and touches more than thirty countries. Amazon, the latecomer, announced its first solo cable that same month — Fastnet, a transatlantic line from Maryland to County Cork with a design capacity above 320 terabits per second, due in 2028. Google has assembled a private fleet of more than thirty systems with names like Grace Hopper, Equiano, and Firmina.
This is where the market reports earn their keep, and also where they need a footnote. The research firm MarketsandMarkets values the submarine-cable business at just under US$20 billion in 2025, rising to US$33.75 billion by 2030 — annual growth of about 11 percent. The catch is that this figure bundles two industries that share a seabed and little else: the glass that carries data, and the copper that carries electricity. The first is being driven by AI and the hyperscalers. The second is its own gathering boom.
The Other Cable
A separate set of lines on the ocean floor carries no information at all. They carry power.
Offshore wind farms have to get their electricity to shore, and countries increasingly want to trade electricity across borders the way they trade gas, which means high-voltage cables threaded under the sea. The length of cable installed for offshore wind grew roughly six-fold in a decade, from about 9,000 kilometres in 2015 to 55,500 in 2025, according to the analysts at TGS 4C Offshore. The longest of the new "electricity superhighways," Viking Link, runs 765 kilometres between Britain and Denmark and switched on at the end of 2023.
Demand has outrun supply. The two dominant makers, Italy's Prysmian and France's Nexans, are sitting on record order backlogs — Prysmian's transmission book alone runs to around €17 billion — and the specialised ships that lay these heavy cables are in short supply, the same bottleneck that haunts the data side. Not every grand scheme survives contact with that reality. Britain abandoned a £25 billion plan to import Moroccan solar power through 3,800 kilometres of cable in mid-2025. The most audacious project of all is still standing, and it points straight at Singapore.
Singapore's Wager
For a country with almost no natural resources and very little coastline, Singapore has made itself one of the most connected places on the planet by treating the seabed as strategy. Twenty-eight submarine cables land on the island today, with at least thirteen more in the works, and over 99 percent of the nation's international traffic moves through them. The dependence is not new — telegraph cables tied the port to Australia, Hong Kong, and India back in the 1870s — but the stakes have climbed with the digital economy, which by 2024 made up nearly 18 percent of GDP.
The island also sits at a genuine chokepoint. The Strait of Malacca beside it sees 80,000 to 90,000 vessels a year, every one of them a potential anchor over a cable, and the South China Sea just beyond carries about a third of global shipping. The regulator, the Infocomm Media Development Authority (IMDA), requires new cables buried several metres into the seabed and concentrates them at a handful of designated landing sites. In 2023 the government published a Digital Connectivity Blueprint with a deliberately bold goal: to roughly double the country's cable landings over a decade, an effort expected to draw upwards of S$10 billion, most of it private. The communications minister, Josephine Teo, framed it bluntly. "We are not going for more of the same, just faster," she said. The point was to build for applications — AI among them — that today's bandwidth cannot serve.
Geopolitics keeps complicating the plan. Several new cables bound for Singapore have been rerouted away from the South China Sea and through Indonesian waters to dodge Chinese permitting and the risk that comes with it; the Apricot and Echo systems were both redrawn that way, buying resilience at the cost of longer, deeper, more expensive routes. One regional cable was reportedly held up more than a year over Beijing's concerns about its waters near Hong Kong. Singapore has answered partly by acquiring capability: in March 2025, the local conglomerate Keppel bought a British firm, Global Marine, folding a six-ship cable-repair fleet under Singaporean control at a moment when such ships are scarce. The country's defence minister, Chan Chun Sing, put the philosophy this way at last year's Shangri-La Dialogue: there is no point defending a cable at a single point, because "we need both ends to be secure."
Then there is the Australia-Asia PowerLink, the surviving giant of the power-cable boom. The plan is to build an enormous solar farm in Australia's Northern Territory and run a cable some 4,300 kilometres under the sea to Singapore — five times longer than today's record-holder — to supply low-carbon electricity to a land-starved island that cannot easily build its own. In October 2024 Singapore's energy regulator gave conditional approval for 1.75 gigawatts of imports, enough for roughly 9 percent of the country's power, with supply not expected until sometime after 2035. The project has already collapsed once into administration, in a feud between two billionaire backers, and been rescued. Whether the cable ever reaches the seabed is an open question. That it was approved at all says something about where the world is heading.
A Handful of Ships, a Handful of Firms
Strip away the trillion-dollar customers and the cables themselves are made and laid by a startlingly small club. Four companies build the overwhelming majority of the world's data cables: SubCom in the United States, Alcatel Submarine Networks in France, NEC in Japan, and HMN Technologies in China, the former Huawei Marine. When you read that Google is "building a cable," what it means is that Google hired one of those four.
That concentration has become a front in the contest between Washington and Beijing. The clearest case was SeaMeWe-6, a cable from Singapore to France: China's HMN bid lower, but a sustained American campaign — diplomatic letters, training grants for route countries, and the threat of sanctions — helped hand the contract to SubCom in 2022. Through a review process known as Team Telecom, the United States has blocked or rerouted a string of cables that would have landed directly in Hong Kong; one already-built segment of the Pacific Light Cable Network simply lies abandoned on the Pacific floor. HMN's share of new construction has fallen from around 18 percent to roughly 7. Paul McCann, a consultant who has worked in the business for four decades, told Reuters he had never seen geopolitical influence over subsea cables like it. "It's unprecedented," he said.
Governments increasingly treat the ability to build cables as a matter of sovereignty. At the end of 2024, the French state bought 80 percent of Alcatel Submarine Networks from Nokia, in a deal valuing the whole company — the only turnkey cable-maker in Europe, with three-quarters of a million kilometres of its cable already on the seabed — at just €350 million. The economy minister called it one of the world's leaders and "the only company of its kind in Europe." A national champion was effectively nationalised for less than the cost of a single major cable system.
What none of this fully fixes is the repair gap. The ships that mend the world's cables are aging faster than they are being replaced, and the people who do the work are a vanishing trade. Aboard the Léon Thévenin, the only repair vessel permanently stationed in Africa — 43 years old, run by France's Orange Marine, responsible for some 60,000 kilometres of cable — a Cape Town jointer named Shuru Arendse is one of a handful of people on the continent who can splice the fibres at sea. The crews can be at it for weeks. "I'm trying to save a country from losing its data or communication," he told the publication Rest of World. He also called himself an absent parent.
The cost of that gap is not abstract. When an undersea volcano severed Tonga's only cable, the country's ATMs stopped working. A line the width of a garden hose had been holding up a nation's banking system, and when it went, so did the cash.
That is the quiet truth under all the capital pouring in. The map of the world that matters most is not the one of borders and capitals. It is the one of thin glass threads on the ocean floor — laid by four companies, guarded by sixty-odd ships, owned increasingly by four more, and cut, on average, three or four times a week. Most of the time, the network reroutes and no one notices. The bet everyone is now making, from Meta's engineers to Singapore's planners to NATO's drone operators, is that most of the time will keep being good enough.
Frequently Asked Questions
What share of the world's data travels through submarine cables?
By the figure used by the International Telecommunication Union and the research firm TeleGeography, roughly 99 percent of international, intercontinental data traffic travels through undersea fibre-optic cables. Satellites — including newer low-orbit constellations — carry only a small remainder. The estimate applies to cross-border traffic specifically, not to all internet traffic, much of which stays within a country over land.
What usually damages an undersea cable?
The network suffers roughly 150 to 200 faults a year. The International Cable Protection Committee attributes 70 to 80 percent of them to ordinary accidents — fishing gear snagging the line and ships dragging anchors — with earthquakes and undersea landslides causing most of the rest. Deliberate sabotage accounts for less than one percent of damage in a typical year, and, contrary to a popular myth, shark and fish bites cause virtually none.
How many submarine cables land in Singapore?
As of 2025, 28 submarine cables land in Singapore with at least 13 more in the works, according to a CSIS case study, making it one of the most connected hubs in the world. More than 99 percent of Singapore's international telecom traffic runs through these cables. The regulator, IMDA, has set a goal under its 2023 Digital Connectivity Blueprint to roughly double the country's cable landings over a decade.
Who owns the world's submarine cables now?
Increasingly, the hyperscalers. Google, Meta, Amazon and Microsoft overtook traditional telecom carriers as the largest users of international capacity in 2017, and by 2024 accounted for nearly three-quarters of all bandwidth demand worldwide, per TeleGeography. The cables themselves are physically built by just four turnkey suppliers: SubCom (US), Alcatel Submarine Networks (France), NEC (Japan), and HMN Technologies (China).
How big is the submarine cable market?
MarketsandMarkets values the submarine cable systems market at about US$19.95 billion in 2025, projected to reach US$33.75 billion by 2030 — annual growth of roughly 11.1 percent. That figure covers two distinct industries that share the seabed: the fibre-optic cables that carry data, and the high-voltage cables that carry electricity for offshore wind and cross-border power trade.
Browse Technology Companies in Singapore
From telecom operators and data-centre providers to the system integrators that connect them, TechDirectory lists verified technology companies across Singapore with company profiles, certifications, and community reviews.
Browse the Directory →Sources include TeleGeography, the International Cable Protection Committee (ICPC), the International Telecommunication Union, MarketsandMarkets, CSIS, TGS 4C Offshore, NATO, and reporting by Reuters, the BBC, the Wall Street Journal, and Rest of World. Market figures are estimates that vary by analyst and by whether power cables are included.