Ready to Sail Within Twenty-Four Hours

There is a clause, and it is the most important sentence in the global communications system that nobody has heard of.

It appears in the terms of the Atlantic Cable Maintenance and Repair Agreement, a non-profit cooperative founded in 1965 and now numbering around sixty members — telecom carriers, power-cable operators, offshore platform owners. When a member’s cable breaks, the maintenance authority notifies the contractor, a vessel is selected, spare cable and splicing gear are loaded, and the ship must be ready to sail within twenty-four hours.

That is the promise on which effectively all intercontinental data depends. It is a private commercial undertaking, not a treaty obligation, and it has been kept remarkably well for sixty years.

Now the arithmetic behind it. ACMA covers the Atlantic Ocean, the North Sea, the Caribbean and the southeastern Pacific. It describes its fleet — one of the largest dedicated maintenance fleets in the business — as three cable ships.

Three.

⁂

The system is regional, and it is structured like insurance. Each maintenance agreement covers a zone. ACMA works out of Brest, in northern France. MECMA covers roughly 71,000 kilometres of cable across the Mediterranean and into the Black and Red Seas, from a base at La Seyne-sur-Mer. Another agreement covers the South Atlantic and Indian Ocean from Cape Town, using a depot belonging to Telkom South Africa. Another covers the Indian Ocean with a remotely operated vehicle aboard a single named ship.

Cable owners pay an annual subscription plus a day rate when the ship is called out. In exchange, a vessel and a crew and a stock of spare cable sit on standby, in a specific port, for years, doing nothing most of the time. Doing nothing is the product. That is what standby means, and it is why the economics are so strange.

Roughly 200 cables are damaged every year worldwide. The overwhelming majority of those faults are accidents — fishing gear and dragged anchors, plus earthquakes, turbidity currents, abrasion. The repair procedure has barely changed in principle since the nineteenth century: a grapnel is lowered on a winch, the cable is hooked and brought up on deck, the damaged section is cut out, a new length is spliced in, and the whole thing is lowered back down.

⁂

Here is what has changed. Cable construction has boomed, driven by hyperscalers financing their own systems, while maintenance capacity has not moved.

Estimates of the global repair fleet cluster in the low tens of vessels. Industry analysts have put the typical maintenance ship at thirty to forty years old; one account found roughly a quarter of repair ships past forty, which for a working vessel is very old indeed. Ships have been booked years ahead. A new cable ship costs upward of $100 million, and the repair market’s margins are thin and getting thinner, because cable owners have spent two decades pushing the price of repairs down. One consultant described it as a crisis of the industry’s own making — the fat squeezed out of the suppliers. Estimates of what it would cost to renew the fleet and hold service at current levels run to about $3 billion.

No one is going to spend $100 million on an asset whose business model is waiting.

And beneath the ships is a constraint that money moves even more slowly: the people. A splice is handmade. Cable jointers, cable engineers, the deck officers who can hold a ship over a fault in weather — these are small trades, learned by apprenticeship, with an ageing workforce and a recruitment problem that industry bodies now name explicitly as a coming bottleneck. You can order a ship. You cannot order thirty years of a jointer’s hands.

⁂

The system has already been tested, and the failures follow one pattern: the repair was technically straightforward, and something else got in the way.

When the Hunga Tonga eruption severed Tonga’s single international cable in 2022, the country was cut off for weeks — not because the splice was hard, but because the nearest ship was far away and the seafloor had been rearranged. When faults hit cables near Taiwan in January 2025, a vessel took weeks to arrive.

The Red Sea is the clearest case. In 2024 a vessel struck by Houthi missiles severed multiple cables as it went down, and repairs stalled for months while operators, insurers and governments worked out whether it was safe and lawful to send an unarmed civilian ship and crew into the area. In September 2025 several cables were cut again near Jeddah; repair timelines were quoted in months.

Nothing in any of those cases was a technical failure. They were failures of access — a ship too far away, a permit not granted, a war-risk premium nobody would underwrite, a legal question about entering a war zone that took longer to answer than the outage lasted.

Meanwhile the Baltic has become a demonstration. Since October 2023, something over ten cables and pipelines in the region have been damaged, in a repeating pattern: a commercial vessel in international waters drags its anchor across the seabed, attribution proves nearly impossible under maritime law, and the repair takes weeks. Finnish special forces boarded a cargo ship on the last day of 2025 on suspicion of exactly that. A Swedish detention, a Latvian incident, a Lithuanian cut, a Finnish–German cut.

⁂

Which brings us to the document.

On 5 February 2026 the European Commission announced its largest subsea infrastructure package to date: about €347 million, paired with a Cable Security Toolbox. Roughly €327 million of it flows through the Connecting Europe Facility to support the development of new cable infrastructure. The Rapid Repair Pilot for the Baltic — pre-positioning modular repair equipment in regional ports — received about €20 million, something like 5.7 percent of the total.

And the pilot’s own limitation is stated plainly in its design: the pre-positioned equipment still requires specialised vessels to deploy it.

Read the ratio, not the press release. Confronted with a crisis whose demonstrated failure mode is repair latency, the largest European response to date allocated roughly one euro in eighteen to repair, and the rest to laying more of the thing that keeps breaking.

This is not stupidity. It is the ordinary political economy of maintenance. A new cable is a capital asset with a landing station, a ribbon, a press release and a growth statistic. A standby ship is a line item that produces nothing in any year it is not needed, and is therefore the first thing a procurement officer questions and the last thing a minister announces. Every institution on earth systematically underfunds the thing that prevents the disaster and overfunds the thing that looks like progress. The subsea industry is simply an unusually clean example, because the numbers are public and the ratio is printed.

⁂

The strategic conclusion follows from the arithmetic, not from any theory of intent.

An adversary interested in disrupting a continent’s connectivity does not need to sever many cables. Networks are resilient; traffic reroutes. What an adversary needs is to exceed the queue — to generate more simultaneous faults than the ships in that zone can service, and then to make the water expensive or legally difficult to enter. Both halves are cheap. Anchor-dragging costs the price of a scrap-value hull and is nearly impossible to attribute. Raising a war-risk premium, or creating a permitting question, costs nothing at all.

And the ship count is public. So is the age of the fleet. So are the zone boundaries, the depot locations and the standby ports. The vulnerability is not classified; it is a business directory.

The current answer to all of this is naval surveillance — NATO’s Baltic Sentry patrols, more monitoring, more attribution effort. Surveillance is worth having. But a frigate cannot splice a cable, and the discussion about who does the splicing in 2040, on what hull, trained by whom, has barely begun.

⁂

Sixty-one years ago a group of telephone companies agreed among themselves that a ship would be ready to sail within a day of a cable breaking, and they have honoured it with an unglamorous consistency the rest of us have simply never had occasion to notice.

The clause is still in force. The ships that have to keep it are, in a number of cases, older than the internet.