Multiple subsea fiber cuts in the Red Sea impacting global communications
Impact Summary
Starting at 05:45 UTC on 06 September 2025, traffic traversing through the Middle East originating and/or terminating in Asia or Europe regions may experience increased latency due to multiple undersea fiber cuts in the Red Sea. The disruption has required rerouting through alternate paths which may lead to higher-than-normal latencies.
This advisory is intended to raise awareness ahead of increased demand as the regions enter the start of its work week.
Current Status
Multiple international subsea cables were cut in the Red Sea. Our engineering teams are actively managing the interruption via diverse capacity and traffic rerouting, while also discussing alternate capacity options and providers in the region.
Undersea fiber cuts can take time to repair, as such we will continuously monitor, rebalance, and optimize routing to reduce customer impact in the meantime. We’ll continue to provide daily updates, or sooner if conditions change.
This message was last updated at 19:30 UTC on 06 September 2025
cr125rider 24 hours ago [-]
The smart folks who keep these things running, despite failures like this, are incredible. Hats off to you all, good luck in the coming weeks as you have to deal with this.
> "There are 150 to 200 instances of damage to the global network each year. So if we look at that against 1.4 million km, that's not very many, and for the most part, when this damage happens, it can be repaired relatively quickly."
...
> If a fault is found, a repair ship is dispatched. "All these vessels are strategically placed around the world to be 10-12 days from base to port,"
...
> To repair the damage, the ship deploys a grapnel, or grappling hook, to lift and snip the cable, pulling one loose end up to the surface and reeling it in across the bow with large, motorised drums. The damaged section is then winched into an internal room and analysed for a fault, repaired, tested by sending a signal back to land from the boat, sealed and then attached to a buoy while the process is repeated on the other end of the cable.
trelane 19 hours ago [-]
> 1.4 million km
Strange way of writing 1.4gm.
bmicraft 8 hours ago [-]
Should be 1.4Gm
trueismywork 13 hours ago [-]
Do you also write gigagrams?
trelane 9 hours ago [-]
Usually I write "pounds," but what I usually write isn't the point.
pwdisswordfishz 10 hours ago [-]
Gram-metres?
trelane 9 hours ago [-]
Hahaha.
Hosted by my own petard. :)
adolph 20 hours ago [-]
I wonder how much slack is left in the cable over its run in order to facilitate bringing it up for repair.
throwup238 11 hours ago [-]
At first it’s an average of 1% slack across the entire run but up to 5% in rough terrain. They don’t use slack though for repairs, they lay down new cable between the bights that adds up to several miles to the length.
soupfordummies 20 hours ago [-]
Now that’s what i call a service loop!
movedx 22 hours ago [-]
BGP <3
marcosdumay 22 hours ago [-]
Yes, but it's damn time people start upgrading to BGP ==3.
matt-p 20 hours ago [-]
Most likely not. Almost certainly bfd or ldp/mpls
specproc 23 hours ago [-]
[flagged]
crimsonnoodle58 23 hours ago [-]
An act of terrorism is never smart or heroic.
specproc 23 hours ago [-]
[flagged]
sporkxrocket 22 hours ago [-]
[flagged]
20 hours ago [-]
js212 21 hours ago [-]
Free the hostages then we can talk
sporkxrocket 6 hours ago [-]
Yeah all of the Palestinians that Israel has been holding hostage since well before Oct 7th need to be freed.
5 hours ago [-]
typpilol 23 hours ago [-]
Were they cut on purpose or accident?
betaby 23 hours ago [-]
Cables in Red Sea got cut frequently. All the time it's the same story - anchor dropped where it should not. Similar stories happen near Singapore.
mandeepj 22 hours ago [-]
> Cables in Red Sea got cut frequently
On land, we’ve call before you dig. I wish they could do something like call before you drop (anchor) training or just get the word out.
jvilalta 22 hours ago [-]
They are usually marked on nautical charts.
simplesimon890 10 hours ago [-]
Some fishermen won't care as there are less ships where the cables are, they'll fish there. There are a lot of cables and fishing vessels in the Irish Sea.
jmkni 4 hours ago [-]
Couldn't that be a security concern?
anikom15 22 hours ago [-]
(It’s not actually anchors.)
bmelton 21 hours ago [-]
It very often _is_ an anchor, and they are usually dropped and dragged on purpose for the effect of cutting the cables which are clearly marked on charts as places to not drop anchor
gpm 21 hours ago [-]
Why do you say that?
I can't say I'm an expert on the matter, but my understanding from industry reporting on some likely intentional ones in the Baltic where that they were still unambiguously done with anchors. Just by someone intentionally causing the anchor to drop and drag along.
simplesimon890 10 hours ago [-]
Cable cuts are very common worldwide and are very rarely reported widely in the media ( that I've seen ) unless there's an opportunity to tie it to Russia. Then it's reported as suspicious or there's a paragraph in the article speculating about the security implications of purposeful attacks on communications infrastructure. IMO it's a ploy to generate anxiety about how much money is being spent on "security" i.e. the military, and how we ( Europe ) need to increase it.
matt-p 20 hours ago [-]
It's by far the most common.. but.. Multiple cables in this manner is IME /less/ likely to be an anchor, although it has happened.
indigodiddy 16 hours ago [-]
[flagged]
hulitu 2 hours ago [-]
You mean no Russian or Chinese ships deliberating cutting cables on land like in EU ? That can't be /s
saddat 18 hours ago [-]
A captain told me once that it is silly to assume a big ship dropps it’s anchor accidentally unnoted
mmis1000 14 hours ago [-]
You need to run a whole procedure to get the anchor dropped. It can't be an "accident" in any means.
And red sea is quite sunny currently (i just checked windy.com now) . There is absolutely no reason that they need to drop it out of nowhere.
cluckindan 13 hours ago [-]
But it’s a somewhat believable cover story when you’re running military sabotage ops from a civilian vessel
viraptor 23 hours ago [-]
It's only been a few hours. I doubt anyone knows.
MengerSponge 21 hours ago [-]
Every subsea cable service interruption story is an excuse to share one of my favorite pieces of long-form technical history: Neal Stephenson's Mother Earth Mother Board:
Can they splice these? Surely the don't have to run a whole new cable?
staplung 20 hours ago [-]
They can but it's an involved process. Modern cables manage their slack very precisely - a complicated job in and of itself - and usually cannot be hauled to the surface without snapping them. So if the cable is not already cut (an anchor can damage a cable in a variety of ways short of physically snapping it) it will be cut while still on the bottom. The ends are then winched up and a splice is patched in but now there's a rather lengthy section of extra cable to deal with. They deal with that by laying it down precisely on the seabed in a nice sideways loop.
oniony 10 hours ago [-]
So then you have extra latency for each repair.
20after4 8 hours ago [-]
And extra attenuation / reflections.
to11mtm 21 hours ago [-]
I've only done land based fiber design, but I would assume based on my experience in that field that one of two things happens.
First option, there's some slack somewhere nearby to do a pull nearby.
If you ever pay attention to aerial poles, you'll see at various intervals things that are vaguely reminiscent of a snowshoe or tennis racket, in the industry they are in fact informally referred to as a 'snowshoe' and the entire purpose is to have some fiber to 'release' if a cut happens and a resplice is needed. [0]
I'd -assume- there is an analogue for these in undersea cables...
If not, well then you'd have to splice a length in and probably have the right sort of mechanism to ensure the cladding/etc stays contiguous (maybe even weld? IDK) but regardless you're probably talking... Actually wait yeah they've got this all figured out; https://www.youtube.com/watch?v=QiwidMEv8CM
[0] - Real world example, for instance if someone uses a fiber splice can for target practice.
Waterluvian 22 hours ago [-]
I think they raise the cable to the surface, splice it and sink it again.
bigiain 20 hours ago [-]
I've read (or perhaps watched on YouTube) the same thing.
Finding both ends of the cable to grab abd raise it took longer than the raise/splice operation.
I suspect a sufficiently malicious attacker could intentionally cut the cable twice, so when the repair crew drag up one end of the cut they know about, the section between the cuts drags its other end away from it's cut-point making the second repair significantly more difficult...
staplung 20 hours ago [-]
In most cases they can easily figure out approximately where the break is on the line by simply measuring the resistance in the cable's conductors (they have more than one but the big one caries a 10,000V feed to power the repeaters). A double cut would probably show up as resistances that didn't sum to the proper total.
matt-p 20 hours ago [-]
You can only practically do that optically e.g through OTDR. You can't TDR steel conductors exposed to the ocean 500KM away with a ohm meter and a Murray loop, to even have a chance you need to put low voltage on it and it's not even close to as accurate.
Spread-spectrum time-domain reflectometry (SS-TDR) is often used to locate and characterize a fault on active cables. Traditional time-domain reflectometry, which sends signals down a conductor to determine where any impedance discontinuities are by timing any reflections that come back from any impedance discontinuities along the way, can’t be used on in-use undersea cables thanks to the high voltages involved. SS-TDR, which was originally developed to detect faults in the wiring of airplanes using 400-Hz AC power, uses modulated pseudo-noise (PN) signals rather than a plain square wave pulse. The signals still bounce off any impedance changes introduced by damage, but an algorithm is used to correlate the returned PN codes with what was sent and when, making it easier to make measurements in a noisy environment. Optical TDR can also be used to locate fiber breaks
> To repair the damage, the ship deploys a grapnel, or grappling hook, to lift and snip the cable, pulling one loose end up to the surface and reeling it in across the bow with large, motorised drums. The damaged section is then winched into an internal room and analysed for a fault, repaired, tested by sending a signal back to land from the boat, sealed and then attached to a buoy while the process is repeated on the other end of the cable.
Impact Summary
Starting at 05:45 UTC on 06 September 2025, traffic traversing through the Middle East originating and/or terminating in Asia or Europe regions may experience increased latency due to multiple undersea fiber cuts in the Red Sea. The disruption has required rerouting through alternate paths which may lead to higher-than-normal latencies.
This advisory is intended to raise awareness ahead of increased demand as the regions enter the start of its work week.
Current Status
Multiple international subsea cables were cut in the Red Sea. Our engineering teams are actively managing the interruption via diverse capacity and traffic rerouting, while also discussing alternate capacity options and providers in the region.
Undersea fiber cuts can take time to repair, as such we will continuously monitor, rebalance, and optimize routing to reduce customer impact in the meantime. We’ll continue to provide daily updates, or sooner if conditions change.
This message was last updated at 19:30 UTC on 06 September 2025
An article about people who fix this
Also, the teams are quite busy:
> "There are 150 to 200 instances of damage to the global network each year. So if we look at that against 1.4 million km, that's not very many, and for the most part, when this damage happens, it can be repaired relatively quickly."
...
> If a fault is found, a repair ship is dispatched. "All these vessels are strategically placed around the world to be 10-12 days from base to port,"
...
> To repair the damage, the ship deploys a grapnel, or grappling hook, to lift and snip the cable, pulling one loose end up to the surface and reeling it in across the bow with large, motorised drums. The damaged section is then winched into an internal room and analysed for a fault, repaired, tested by sending a signal back to land from the boat, sealed and then attached to a buoy while the process is repeated on the other end of the cable.
Strange way of writing 1.4gm.
Hosted by my own petard. :)
On land, we’ve call before you dig. I wish they could do something like call before you drop (anchor) training or just get the word out.
I can't say I'm an expert on the matter, but my understanding from industry reporting on some likely intentional ones in the Baltic where that they were still unambiguously done with anchors. Just by someone intentionally causing the anchor to drop and drag along.
And red sea is quite sunny currently (i just checked windy.com now) . There is absolutely no reason that they need to drop it out of nowhere.
https://www.wired.com/1996/12/ffglass/
https://news.ycombinator.com/item?id=20497098
https://web.archive.org/web/20210107003422/https://www.wired...
The Verge also has a great article which is more recent and more specifically about a repair ship:
https://web.archive.org/web/20240422050805/https://www.theve...
First option, there's some slack somewhere nearby to do a pull nearby.
If you ever pay attention to aerial poles, you'll see at various intervals things that are vaguely reminiscent of a snowshoe or tennis racket, in the industry they are in fact informally referred to as a 'snowshoe' and the entire purpose is to have some fiber to 'release' if a cut happens and a resplice is needed. [0]
I'd -assume- there is an analogue for these in undersea cables...
If not, well then you'd have to splice a length in and probably have the right sort of mechanism to ensure the cladding/etc stays contiguous (maybe even weld? IDK) but regardless you're probably talking... Actually wait yeah they've got this all figured out; https://www.youtube.com/watch?v=QiwidMEv8CM
[0] - Real world example, for instance if someone uses a fiber splice can for target practice.
Finding both ends of the cable to grab abd raise it took longer than the raise/splice operation.
I suspect a sufficiently malicious attacker could intentionally cut the cable twice, so when the repair crew drag up one end of the cut they know about, the section between the cuts drags its other end away from it's cut-point making the second repair significantly more difficult...
https://en.m.wikipedia.org/wiki/Murray_loop_bridge
https://hackaday.com/2024/07/30/undersea-cable-repair/A little knowledge goes a long way
An article about such repairs (150-200 yearly)
> To repair the damage, the ship deploys a grapnel, or grappling hook, to lift and snip the cable, pulling one loose end up to the surface and reeling it in across the bow with large, motorised drums. The damaged section is then winched into an internal room and analysed for a fault, repaired, tested by sending a signal back to land from the boat, sealed and then attached to a buoy while the process is repeated on the other end of the cable.