Southeast Asia-Japan Cable (SJC)

Based on 56 RIPE Atlas measurements from GeoCables monitoring infrastructure, March–April 2026.

The Southeast Asia-Japan Cable, generally referenced as SJC, is an 8,900-kilometre pan-Asia submarine cable system connecting Telisai in Brunei, Chung Hom Kok on Hong Kong Island, Shantou on the Chinese mainland, Chikura on Japan's Pacific coast, Nasugbu in the Philippines, and Tuas in Singapore. The cable came into service on 27 June 2013 and is owned by a ten-member consortium that includes China Mobile, China Telecom, Chunghwa Telecom, Globe Telecom, Google, KDDI, National Telecom Thailand, Singtel, Telkom Indonesia, and Unified National Networks (UNN). The original design capacity was over 15 Tbit/s across six fibre pairs, with current upgraded capacity in the 28 Tbit/s range after several rounds of terminal optic refresh.

SJC sits in an unusual position in the modern submarine cable financing landscape. The consortium model is the traditional one — ten national-incumbent carriers each holding a fixed share of fibre pairs and right of use — but the consortium includes Google as one of its partners, an early example of hyperscaler participation in a multi-carrier system rather than building a single-owner cable outright. The Google membership in SJC predates the company's later shift toward fully owned cables like Honomoana and Curie, and reflects an earlier phase of hyperscaler thinking when partial consortium ownership was the dominant route into the submarine cable economy. Today's mix of consortium and single-owner systems on the same regional routes is a direct legacy of that transition.

0.826× of floor: the chord effect, again

The minimum round-trip we observe between Tuas in Singapore and Chikura in Japan is 71.92 ms, with measurements collected in both directions of this corridor. The physics floor for the cable's nominal 8,900-km length is 87.10 ms; the minimum we observe sits at 0.826× that floor. Below the floor, by about 17%.

That number describes the same below-floor phenomenon that EXA North and South exhibits at 0.716× across the Atlantic, ARCOS-1 at 0.613× in the Caribbean, and LION-2 at 0.235× in the Indian Ocean: the actual route a packet takes between Tuas and Chikura is shorter than the cable's nominal advertised length. SJC's 8,900-km wet plant covers a six-landing branched topology — Brunei to Hong Kong to Shantou to Chikura to the Philippines to Tuas — and the great-circle distance between Tuas and Chikura specifically is well under that full system length. The 71.92-ms minimum corresponds to a transit of roughly 7,300 km, which is the actual fibre length used by the Singapore-Japan corridor within SJC's branched system.

For the Tuas-Chikura corridor specifically, this 0.826× is what a clean transit of just the relevant SJC segment looks like. Singapore-side carriers that route Japan-bound traffic onto SJC's southern branch hand the packet off into the cable, the cable's branching units route it through Brunei and Hong Kong toward Japan, and the packet exits at Chikura with no detour. The 17% below-floor multiplier is the geometric truth that the cable doesn't have a single straight Singapore-to-Japan run — it has a branched system whose Singapore-Japan corridor is shorter than the cable's full advertised length.

Two directions, very different stability

The forward and reverse directions of the Tuas-Chikura corridor produce strikingly different measurement profiles. Across 29 measurements from Tuas to Chikura, the round-trip averages 122.82 ms, with a minimum of 71.92 ms, a maximum of 549.95 ms, and a standard deviation of 84.28 ms. Across 26 measurements in the reverse direction — Chikura to Tuas — the average is just 76.56 ms, with a minimum of 74.33 ms, a maximum of 82.71 ms, and a standard deviation of 1.88 ms. The reverse direction sits squarely at floor; the forward direction is wild.

This is asymmetry by routing decision rather than by cable behaviour. Submarine fibre is symmetric — light travels at the same speed in both directions — and the Tuas-Chikura wet plant should produce identical numbers in both senses if both ends used it identically. The Japanese carrier we measure from on the Chikura end commits its Singapore-bound traffic directly onto SJC: 11-12 hop median traceroute, sub-2-millisecond standard deviation, every measurement falling within 8 ms of the floor. The Singapore-side carrier we measure from does not commit eastbound traffic to SJC the same way: traceroute hops jump between sessions, latency spikes to half a second on some flows, and the routing decision varies measurement-to-measurement. AJC shows the same shape on the Sydney-Maruyama corridor; MENA shows it on the Oman-Italy corridor. SJC adds another data point to the same general pattern: one side committed, the other side load-balancing across alternatives.

The size of the asymmetry on SJC, however, is unusual. The 549-ms maximum from the Tuas side is over six times the floor; on a healthy submarine cable, that maximum would be expected within a few percent of the minimum. The Singapore-side BGP policy is evidently routing some sessions onto paths that do not use SJC at all — likely traversing one of the SEA-ME-WE family cables across the Indian Ocean and onward, an order of magnitude longer than the direct Singapore-Japan run.

Pan-Asia routing through six landings

SJC's six landings span four East Asian and Southeast Asian countries. The cable's value is most clearly anchored not in any single point-to-point corridor but in the meshed connectivity it provides among the seven landing nations: any pair of landings has a direct physical path, and the cable effectively serves as a regional internet backbone for the East Asia and Southeast Asia routes that historically lacked dense interconnection. The Chung Hom Kok landing on Hong Kong Island is geopolitically notable — it places SJC under the regulatory umbrella of Hong Kong's submarine cable framework, which is administered through the city's autonomous Office of the Communications Authority but ultimately subject to Beijing oversight. This has become a more sensitive subject through the 2020s as transpacific cables involving US ownership have increasingly avoided Hong Kong landings; SJC predates that shift and remains operational through Hong Kong.

The Chikura landing on Japan's Pacific coast places SJC into the same coastal cluster that hosts most of Japan's transpacific submarine cables. Japan's 70-plus submarine cable landing stations concentrate on three coastal regions, and Chikura sits in the Chiba/Bōsō Peninsula area alongside Maruyama and the broader Pacific-facing cluster. The Chikura backhaul to Tokyo data centres is the same diversified terrestrial fibre network that carries JUNO's and AJC's traffic — a shared resource for all cables in the cluster.

What we will keep watching

SJC is in its thirteenth year of service and continues to function as one of the workhorse intra-Asia submarine cables. The asymmetry between Tuas-direction and Chikura-direction measurements is the most operationally interesting fact about the corridor right now, and tracking how it evolves — whether the Singapore-side BGP commitment to SJC strengthens or whether the load-balancing across alternatives continues — will tell us something about the relative competitive position of SJC versus newer Pan-Asia systems. The cable's continued operation through Hong Kong is also a structural fact worth tracking, given the regulatory and geopolitical pressures on transpacific cables that involve Chinese landings.

What we measure on SJC — 71.92 ms minimum across what is nominally an 8,900-km cable, with a 0.826× chord behaviour for the Tuas-Chikura corridor and substantial directional asymmetry — is what a thirteen-year-old pan-Asia consortium cable looks like in 2026 with active routing politics on the Singapore side and a tightly committed Japanese-side route policy. The cable continues to carry meaningful Pan-Asia traffic, and its position in the broader regional cable economy remains structurally important to all of its consortium members.