APCN-2

Asia Pacific Cable Network 2 (APCN-2) is a 19,000 km submarine cable commissioned in December 2001, forming a ring around East and Southeast Asia. It has ten landing stations across seven countries and territories: Katong in Singapore, Cherating in Malaysia, Batangas in the Philippines, Lantau Island in Hong Kong, Shantou and Chongming in mainland China, Tanshui in Taiwan, Chikura and Kitaibaraki in Japan, and Busan in South Korea. APCN-2 was commissioned in the same 2000-2001 wave as PC-1, FLAG-NAL, and several others — Asia's foundational intra-regional fibre infrastructure that predates nearly all current internet economy.

APCN-2 was built by a consortium of Asian and international telecoms: NTT, KDDI, China Telecom, Chunghwa Telecom, SingTel, Korea Telecom, Telstra, AT&T, Verizon, and others. The original design capacity was 2.56 Tbps. As with nearly all cables of its vintage, subsequent coherent-modulation upgrades have raised practical capacity substantially above the launch specification.

Intra-Asian segment at 70 ms

Our monitor measures APCN-2 between Katong in Singapore and Kitaibaraki in Japan — two landings on opposite ends of the cable's ring route. Over 30 days we collected 44 samples, with the data showing two distinct latency regimes:

The 70.5 ms forward minimum is well below the physics floor for APCN-2's full 19,000 km length (185.8 ms). That is not an error — it simply means the packet is using only a portion of the cable, travelling directly between Singapore and Japan without touching the other eight landings. 70.5 ms round-trip corresponds to about 7,200 km of fibre, consistent with the direct Singapore-to-Japan segment of the ring.

The reverse direction is exceptionally stable: ten samples across four days with a 1 ms spread. That tightness, 25 years after the cable's commissioning, is the same signature we documented on PC-1 and FLAG-NAL — a physical cable body functioning exactly as designed.

The forward-direction variance (70.5 ms minimum, 220.9 ms maximum) is interesting. A 150 ms range on consecutive samples suggests the path is splitting between two distinct routes day to day — one direct (70 ms), one detouring via a longer segment (120+ ms). The packet selection between them depends on BGP convergence at the moment of measurement.

Ten landings on a ring architecture

The ring architecture is the defining technical choice. A packet entering at any landing can travel either direction around the loop to reach any other landing, giving the cable inherent redundancy: a single fault somewhere along the ring does not disconnect any pair of landings, it merely forces traffic to take the longer way around.

Ring topologies were popular in the early 2000s because they solved the redundancy problem without requiring two parallel cables. Later systems (from roughly 2015 onward) typically prefer star-plus-branching architectures that achieve similar redundancy with more efficient capacity utilisation. APCN-2's ring is a technological fossil in that sense — still functional, still useful, but no longer the architecture of choice for new builds.

The consortium behind the cable

APCN-2's consortium reflects the state of Asian telecom in 2000. Large national telecoms dominated capacity planning: NTT and KDDI from Japan, China Telecom and Chunghwa Telecom from China and Taiwan, SingTel from Singapore, Korea Telecom from Korea. International operators with significant Asian interests participated as peripheral members — AT&T, Verizon (then WorldCom), Telstra, Cable & Wireless.

This structure is very different from contemporary cables. A 2026 cable like APRICOT has Meta and NTT as primary sponsors; ADC has Meta plus multiple Asian telecoms; JUPITER is hyperscaler-heavy. Hyperscalers did not exist as cable investors in 2000. APCN-2 was built by the carrier-consortium model that predominated for three decades and is now, in 2026, a receding presence.

A foundational cable of Asian internet

APCN-2 entered service in December 2001 as Asia's first modern high-capacity intra-regional cable. At its commissioning, Asian internet traffic was a fraction of what it is today — Japan's total international bandwidth in 2001 was measured in gigabits; by 2026 it is measured in tens of terabits. APCN-2 was one of the cables that made that growth possible by providing the physical backbone for Asian internet exchange peering to develop.

Many of the commercial peering relationships that define contemporary Asian internet routing trace back to the APCN-2 era. When a Singaporean ISP and a Japanese ISP first established direct peering, the capacity that carried it was often on APCN-2 or one of its close contemporaries. Those relationships have endured through cable generations — the underlying fibre systems have upgraded, but the peering contracts and traffic patterns they underwrite continued.

What our data proves

• APCN-2 delivers Singapore → Japan at 70.5 ms minimum, below the full-cable physics floor. The measurement reflects direct segment routing, not full ring traversal.
• Reverse-direction stability is at the physical limit. 1 ms spread across 10 samples shows the 25-year-old cable performing at spec.
• Ring architecture provides inherent redundancy. Any single fault reroutes traffic around the loop; no pair of landings disconnects absolutely.

APCN-2 is one of Asia's foundational intra-regional cables. It is past the halfway point of its 25-year design life, has been through multiple ownership and consortium restructurings, and continues delivering physics-floor performance on its primary segment. It will eventually be retired — likely within the next 5-10 years — and the capacity will be absorbed by newer systems like SJC2, APRICOT, and ADC. Until then it remains operational.

Coexisting with the next generation

The interesting question for a 25-year-old cable is not whether it works (it does) but what role it plays alongside newer systems. APCN-2 is being increasingly displaced by SJC, SJC2, ADC, and APRICOT for primary production traffic. Modern cables offer 100x or more capacity per fibre pair, lower per-bit cost, and access to new landings (Vietnam's Quy Nhon, additional Philippine sites) that APCN-2 does not reach.

What APCN-2 retains is reliability, redundancy, and existing peering relationships. Carriers that have used APCN-2 for fifteen years have established commercial arrangements they will not disrupt without reason. The cable continues serving as backup capacity, secondary peering paths, and as a known-good fallback when newer systems experience disruption. That is the typical fate of a 25-year-old commercial submarine cable: not retired, but demoted to roles that suit its age and capacity profile.

Try it yourself

Live data on the APCN-2 cable page. Compare with other foundational Asian cables: FLAG-NAL (2001 ring), SJC (2013 successor), and the 2024-2025 intra-Asian mesh of ADC, APRICOT, and SJC2.