The first public sign of trouble was a status page. At 08:05 UTC on Saturday, August 8, 2026, Australian provider Aus Internet opened an incident on INDIGO West, the 4,600 km cable from Perth to Singapore by way of Jakarta. Three and a half hours later the update turned blunt: the system was "hard down" with a shunt fault. Before the weekend was over, its sibling INDIGO Central, 4,725 km from Perth around the Great Australian Bight to Sydney, had a shunt fault of its own.
Both faults sit inside the Perth Submarine Cable Protection Zone, where anchoring and trawling have been banned since 2007. Bevan Slattery, founder and co-CEO of SUBCO, which holds substantial capacity on both systems, said the faults happened "straight after each other and in close proximity to each other" and asked the Australian Federal Police to look into it. The AFP told iTnews it "has received a report of crime in relation to this matter."
Nobody has named a vessel and nothing has been proven. But two cables failing in one weekend, in one protected corridor, is the kind of coincidence that gets investigated, and the physical evidence sits 47 km offshore until a repair ship arrives, currently expected August 30.
A shunt fault is a power problem first
A repeatered submarine cable is an electrical system wrapped around an optical one. The fibers sit inside a steel vault wrapped in a copper conductor and polyethylene insulation, and power-feeding equipment at the landing stations drives thousands of volts of direct current through that copper to run optical amplifiers every 60 to 100 km. The return path is the ocean itself, through earth electrodes on the beach.
A shunt fault is a breach in that insulation mid-span: the copper touches seawater and the circuit finds a new ground at the fault. It is a short, not a cut. With enough margin in the power feed, operators can rebalance around the leak and keep every amplifier lit, so cables sometimes run for months with a known shunt fault while a repair ship is booked.
The electrical signature is also what found the damage. Early testing placed the INDIGO West fault about 65 km out; by August 15 the estimate had sharpened to 47.624 km from the Perth landing station, 28.5 km from the nearest repeater. That precision comes from the beach: resistance to the new sea ground tells you how much copper lies between you and the hole.
INDIGO West did not get the graceful version: it was hard down within hours. And because neither INDIGO cable carries the fiber-sensing instrumentation SUBCO has deployed on newer systems, there is no acoustic recording of the moment of damage. "The consortium will not know the root cause until the cable is recovered during repair," Slattery said.
Inside the Perth submarine cable protection zone
Australia protects its cables more explicitly than most countries. Schedule 3A of the Telecommunications Act 1997, added in 2005, lets the Australian Communications and Media Authority declare protection zones over cables of national significance, and damaging a cable inside one carries up to ten years in prison. ACMA declared three zones in 2007: two off Sydney and one off Perth. The Perth zone reaches up to 60 nautical miles seaward from City Beach, into water 2,000 meters deep, a nautical mile either side of each cable route, and anchoring, trawling, and dredging are all prohibited inside it.
The zone exists because the seabed off Perth is crowded. INDIGO West comes ashore at Floreat Beach, and the same corridor carries the Australia Singapore Cable and SUBCO's Oman Australia Cable: nearly everything that connects Australia westward funnels through a few kilometers of coast. "Australia has 16 subsea cables that carry 99 per cent of our international communications," security analyst David Brewster told the ABC.
The law has been tested in this exact spot: between July 31 and August 1, 2021, a ship dragging its anchor damaged cables in the Perth zone, and the AFP later charged its master over damage to the Australia Singapore Cable. The prosecution shows what the zone can do after the fact; a line on a chart stops no vessel that has decided to cross it.
What rerouting looks like from the outside
Aus Internet's first advisory said that "traffic on protected services and Internet Services will route through alternate routes." That sentence is most of the reason Australians did not notice their internet break on August 8.
The arithmetic of the alternatives is set by the speed of light in glass, roughly 200,000 km/s: about 1 ms of round-trip time per 100 km of path. INDIGO West's 4,600 km puts the floor for Perth to Singapore near 46 ms. Fail over to the Australia Singapore Cable and the geometry barely changes. Fail over eastward instead, across the continent and up through trans-Pacific systems, and the path roughly triples: the floor climbs past 120 ms before a single congested queue is added.
A cable system has no AS number, so a failure never shows in a BGP feed directly; what shows are consequences. Wavelength customers fail over beneath the IP layer, invisibly. But when a carrier loses transit on one coast, AS paths shift: routes that reached Singtel (AS7473) or Telstra (AS1221) through a west-coast landing re-emerge through Sydney, and a traceroute from Perth to Singapore sprouts east-coast hops. A server that answered in 50 ms yesterday and 150 ms today has not moved; its country's geography changed underneath it.
And redundancy in cable systems is not redundancy in geography. The primary backup for a cable landing near Perth is another cable landing near Perth, in the same 60-nautical-mile protected corridor where both faults occurred.
The vessel, the pattern, and a counterexample the same week
The suspicion has a concrete basis. Maritime analytics firm Windward assembled three independent looks at one unnamed vessel: AIS tracking that shows it entering Australia's exclusive economic zone on July 28, setting its ETA to January 1, 2027 (the maritime equivalent of a null value), and repeatedly crossing the cable corridor; an optical satellite image from August 3, about 11 km west of Perth; and a synthetic-aperture radar frame from August 8 at 21:42 UTC, directly on the cable route, inside the window when the faults developed. Windward is careful about what this means: correlation, not attribution. Slattery himself allowed that "this could be a coincidence or an external aggression event."
Skepticism is warranted: the International Cable Protection Committee counts 150 to 200 cable faults worldwide in a typical year and attributes 70 to 80 percent to fishing gear and anchors. Most cable damage is a fishing boat that will never make the news.
But the last two years read like a series. November 2024: C-Lion1 and BCS East-West severed in the Baltic within a day of each other. December 25, 2024: the tanker Eagle S drags its anchor across the Gulf of Finland, breaking the Estlink 2 power cable and four telecom cables. January 2025: the Shunxing 39 linked to damage on the Trans-Pacific Express off Taiwan. February 2024: the crippled Rubymar's dragging anchor cuts Seacom, AAE-1 and EIG in the Red Sea. Each one individually deniable. The accumulation is the story, and Perth now sits on that list pending a better explanation.
The same week supplied the control case. Russia's Kamchatka peninsula lost effectively all mobile data and broadband for roughly 300,000 residents starting August 12, with no mystery vessel anywhere: Rostelecom deliberately shut down the peninsula's only fiber link, the Kamchatka-Sakhalin cable, while divers spliced in about 400 meters of new cable at Cape Levashov, where coastal erosion threatened the shore end. Service returned on August 14. Perth had somewhere else to send the bits. Petropavlovsk-Kamchatsky did not.
Repair day is its own outage
The fix for a shunt fault is surgery. The ship powers the system down, grapples the cable, cuts it, raises both ends, splices in a fresh section on deck, and lays the repaired loop back with slack to spare. While that happens, a cable that had been limping along goes fully dark: Aus Internet has already warned that traffic across the link "may be down for up to 3 days" during the work. The repair completes the outage the fault started.
The bottleneck is the ship. The global fleet numbers around 60 cable vessels, fewer than 20 of them dedicated to repair, and many are more than 30 years old. The INDIGO ETA has already slipped from August 26 to August 30, with repair estimated complete September 4, "subject to clearance, logistics arrangement and weather conditions." Nearly four weeks from fault to fix, in a wealthy country with a friendly coastline. That is normal, and normal is the uncomfortable part.
One more detail from the status history: INDIGO West shunt-fault repairs had originally been booked into an 11-day maintenance window in late July, which slipped. Read plainly, the system appears to have been carrying a known fault into the weekend that took it down, though the updates never spell out the relationship. Cables live with their injuries a long time while the queue for a ship moves.
Australia did get one piece of timing luck. SMAP, SUBCO's 5,000 km system between Perth and Sydney with 16 fiber pairs and more than 400 Tbps on a geographically separate southern route, reached ready-for-service on June 25, six weeks before the faults. Had this weekend arrived in May, the east-west picture would have been a good deal thinner.
Events like this are visible from the IP layer: paths lengthen and latency stops matching the map. When a route change makes a "nearby" server answer from far away, Ipregistry's API gives you the context behind the address: its ASN, carrier, and connection type. You can try it with 20,000 free lookups to get started.
Keep users in the loop Ship release notes that get read. Try Noticeable