How undersea internet cables are repaired, and why it takes weeks
Undersea cables are repaired by ships that grapple both ends, splice in new cable and relay it. See each step, ICPC fault figures and why delays happen.
A broken undersea cable is repaired by a specialised cable ship that locates the fault, hauls both cut ends to the surface, splices in a new section of cable on deck and lowers the result back to the seabed. The sailing and the splicing take days. The weeks usually come from the steps around them: waiting for a free ship, getting permission to work in a country's waters and waiting for weather.
This guide walks through each stage, explains what drives the total time, and covers how many faults happen in a year. Figures are as of October 2026 and come mainly from the International Cable Protection Committee (ICPC), the industry body for cable owners and operators.
How common are cable faults?
The ICPC says roughly 150 to 200 submarine telecom cable faults occur globally each year, and describes that count as stable. It puts the global network at about 500 cable systems; its figures for total cable length vary between pages (around 1.7 to 1.8 million km). A fault does not mean an outage: most cables sit in a mesh of alternative routes, so traffic is rerouted and users often notice nothing.
The ICPC attributes 70 to 80 percent of faults to accidental human activity, mainly commercial fishing and ship anchors. The rest come from natural hazards and equipment failure. Deliberate sabotage is, according to the ICPC, under one percent of damage worldwide each year, and the committee says it knows of no verified state-sponsored sabotage since World War II. That is the ICPC's own assessment, and it holds no data on the question. Individual incidents, such as those in the Baltic, are reported and investigated case by case in our networks coverage.
Anchors alone are significant. An ICPC viewpoint on dragged anchors puts them at about 30 percent of incidents each year, roughly 60 faults, and notes that several simultaneous faults can create a backlog for repair vessels.
Step 1: Finding the fault
Operators monitor each cable continuously from its landing stations. When a break or a degradation appears, they measure the signal from the shore end to estimate how far along the cable the damage lies. Electrical and optical reflection measurements can narrow the location to a stretch of seabed, which tells the ship where to start.
The estimate is not exact. The cable's laid length does not match the straight-line distance, and the route may have been adjusted since installation, so crews expect to search around the predicted point.
Step 2: Mobilising a ship and getting permission
Repair ships are stationed at strategic ports and work under cooperative maintenance agreements between cable owners. The ICPC says vessels load cable and jointing equipment and sail to the repair ground. The ICPC's cableships list showed 63 entries as of October 2026, and the committee says the list is illustrative and not complete. Not every vessel on it is dedicated to repair work.
Permits are a separate hurdle. The ICPC calls for timely and efficient permitting for maintenance and repair work, and lists fewer regulatory delays in obtaining permits as a way to protect cables. It has compared repair start times across countries using anonymised maintenance-agreement data, to support talks with regulators. Its comparison report is available on request, so we do not quote its findings here.
Step 3: Recovering the cable
At the repair ground the ship drops a marker buoy near the predicted break. It then drags a grapnel along the seabed to snag the cable. The grapnel type depends on the bottom: rigid prongs that plough below the surface suit sand, while a flexible grapnel with hooks along its length suits rock.
In deep water the cable may be too weak to lift as one long piece. In that case a cutting grapnel brings up one end at a time. In shallow water, divers or a remotely operated vehicle can be used instead. In both cases the aim is to get each cut end on board.
Step 4: Testing, jointing and relaying
On deck, technicians test each end to confirm where the fault is and that the cable on either side is sound. They then join in a new length of spare cable, and a repeater if the section needs one. Splicing the optical fibres and sealing the joint against water pressure is careful work, and it is the part that cannot be rushed.
The repaired cable ends up longer than the original, because the spare section is added. The ship lays the extra length in a loose U shape on the seabed, so the cable is not under tension. Operators then test the full link from shore to shore before returning it to service.
How long does a repair take?
There is no standard figure. The ICPC says it analyses repair times with maintenance providers, but route and fault data belong to cable owners and are confidential. What can be said is that the duration depends on:
- Ship position. A vessel may be a few days' sailing from the fault, and may be finishing another repair.
- Permits. Waiting for approval to work in a country's waters can delay the start of a repair.
- Depth and seabed. Deep water makes recovery slower than shallow water.
- Weather and sea state. Grapnel work and jointing stop in heavy seas.
- Security. Conflict zones can delay or prevent access.
One documented example: in the November 2024 Baltic incidents, two cables were damaged and the ICPC says both were repaired within two weeks, with no service interruption. Other cases, such as those in contested waters or where approvals are slow, can take longer.
Why resilience matters more than speed
Because repairs are slow and ships are few, the network is built so that a cut cable does not stop service. Traffic can be rerouted over other cables, and branching designs limit the effect of one fault. The ICPC says its priorities include protecting cables and reducing delays to repairs, and the infrastructure teams that run cloud and content networks plan capacity with several routes in mind. For readers who look at the risk side, security reporting covers deliberate attacks on infrastructure.
Frequently asked questions
How are undersea cables repaired?
A repair ship finds the fault, hauls both cut ends aboard with grapnels, joins in a new section of cable and lowers the cable back to the seabed. The repaired link is tested end to end before it returns to service.
How long does it take to repair an undersea cable?
It varies with ship availability, permits, depth and weather, and no official average is published. The ICPC reports that two cables damaged in the Baltic in November 2024 were both repaired within two weeks.
How many undersea cable faults are there each year?
The ICPC says about 150 to 200 faults occur worldwide each year. It attributes 70 to 80 percent to accidental human activity such as fishing and anchors.
Are undersea cable cuts usually sabotage?
No. The ICPC says deliberate sabotage accounts for less than one percent of cable damage worldwide each year, though it also says it holds no data on state-sponsored sabotage.
Why do some repairs need permits?
Ships generally need permission to work in a country's waters. The ICPC calls for timely, efficient permitting and lists fewer regulatory delays as a way to protect cables.