10,400 km · 11 Landing Points · 8 Countries · Ready for Service: 2016
| Length | 10,400 km |
|---|---|
| Status | In Service |
| Ready for Service | 2016 |
| Landing Points | 11 |
| Countries | 8 |
| Location |
|---|
| Busan, South Korea |
| Changi South, Singapore |
| Cherating, Malaysia |
| Chongming, China |
| Danang, Vietnam |
| Maruyama, Japan |
| Nanhui, China |
| Shima, Japan |
| Songkhla, Thailand |
| Toucheng, Taiwan |
Asia Pacific Gateway (APG) is a 10,400-kilometre submarine cable system that connects eight East and Southeast Asian countries. It entered service in 2016 and today links eleven landing stations across mainland China, Hong Kong, Japan, South Korea, Taiwan, Malaysia, Singapore, Thailand, and Vietnam. APG is one of the largest carrier-consortium cables in the region, with thirteen co-owners — and it is a clean example of the consortium era of submarine cables, just before the hyperscalers changed the economics of the business.
APG was designed by and for regional telecom carriers. Its owner list reads like a roll-call of Asia-Pacific national operators: China Mobile, China Telecom, China Unicom from mainland China; Chunghwa Telecom of Taiwan; KT and LG Uplus of South Korea; NTT of Japan; TIME dotCom of Malaysia; Starhub of Singapore; National Telecom of Thailand; VNPT International and Viettel of Vietnam. Meta (Facebook) bought into the cable as the lone hyperscaler, gaining capacity without owning a consortium share.
Thirteen owners on one cable is a lot. It reflects an era — still ongoing, but fading — when no single carrier could afford a cable of this scale alone, and when long cables were built by committee rather than by a single hyperscaler writing a cheque. Compared with modern systems like Apricot (2025, Google + Meta + regional partners) or Echo (2025, Google + Meta Pacific private), APG represents a coalition-of-the-willing design, where every national operator that terminates the cable also owns a slice of it.
| Specification | Value |
|---|---|
| Length | 10,400 km |
| Ready for service | 2016 |
| Fibre pairs | 6 |
| Design capacity | 30.72 Tbps |
| Landings | 11 stations across 8 countries |
| Owners | 13 carriers |
By 2026 standards 30.72 Tbps is modest. Newer systems carry an order of magnitude more: Medusa, lit earlier this year, moves 480 Tbps across 24 fibre pairs. But APG was at the state of the art for 2016. Its six-pair architecture matched what 100 Gbps coherent transceivers and the first 200 Gbps wavelengths could monetise. A decade on, operators can upgrade transponders to push more bits per pair, but the fibre count itself is fixed for the life of the cable — so the ceiling is already set.
We monitor APG primarily between Maruyama (Japan) and Changi South (Singapore) — the cable's north-south backbone segment of roughly 5,800 km. Over 30 days we have collected 81 clean samples across both directions:
| Direction | Samples | Min RTT | Avg RTT | Baseline | Ratio |
|---|---|---|---|---|---|
| Maruyama → Changi South | 48 | 74.6 ms | 115.3 ms | 98.6 ms | 1.22 |
| Changi South → Maruyama | 33 | 70.2 ms | 109.7 ms | 103.7 ms | 1.06 |
The minimum RTT of 70.2 ms is actually below the theoretical physics floor of 101.8 ms for a great-circle path that would touch every Southeast Asian intermediate landing. This means traffic on APG is not forced through each landing — packets can transit from Japan to Singapore more directly, taking advantage of the cable's ability to drop repeaters and branching units at high speed. The average of 115 ms runs 40 ms higher than the minimum because the cable is shared with other carriers' commercial routing, some of which adds intermediate peering hops at Hong Kong or Taiwan.
Ratio 1.22 on the Maruyama → Changi direction means the cable is currently running 22% above its stable baseline — consistent with our anomaly detector flagging a latency spike at Changi South on the day this profile was written. Typical APG anomalies resolve within hours; a cable at ten years of age is still in operational prime, but every few months it sees congestion, planned maintenance windows, or transient peering issues.
Our probes across Europe and Central Asia reach the APG network at Changi South with these typical latencies:
| Source probe | To Changi South |
|---|---|
| Minsk (Belarus) | 205 ms |
| Jerusalem (Israel) | 219 ms |
| Sevastopol (Ukraine) | 231 ms |
| Tbilisi (Georgia) | 249 ms |
| Almaty (Kazakhstan) | 252 ms |
Almaty is geographically closer to Singapore than Minsk, but its traffic shows higher latency. The cause is routing: packets from Almaty transit through Russian and European peering points before re-entering Asia via Moscow–Frankfurt–Singapore, adding several thousand kilometres of fibre to a geographically shorter trip. This is a common pattern in post-Soviet networks: fibre topology is shaped by historical carrier relationships, not by geography. The same Almaty packet destined for a European city might traverse the same Moscow–Frankfurt path going the other way.
Ten years into its operational life, APG occupies a specific niche in the Asia-Pacific backbone. It is not the newest or fastest cable in the region — that title belongs to Apricot (2025) and Echo (2025), which split the hyperscaler private-cable market. It is also not the longest — 2Africa (45,000 km, spanning a continent) and SEA-ME-WE-6 (21,700 km, Europe-to-Singapore) are much larger systems. What APG is, specifically, is a general-purpose intra-Asian workhorse: a consortium-owned cable that thirteen national carriers keep paying for because each of them needs capacity between East Asian and Southeast Asian markets, and none of them is going to stop.
In that sense APG is a reminder that submarine cable ownership is not fungible. A hyperscaler cable is used when Google or Meta wants to move data between its own data centres; a consortium cable like APG is used when a Malaysian telecom wants to reach a Korean internet exchange, or when a Vietnamese ISP peers with a Japanese content provider. Even as hyperscaler private cables proliferate, the consortium pattern persists — and APG remains one of the clearest examples of how thirteen competing carriers can still cooperate on a decade-long piece of shared infrastructure.
Live latency data on the APG cable page. For comparison, see Apricot (Google + Meta, 2025) and TPU (Google private, 2026) — hyperscaler-era cables along similar routes. Our measurements refresh every two hours.
Because APG crosses eight countries with differing regulatory regimes, its reliability metrics vary by season. In recent windows we have observed transient latency spikes at the Changi South landing in Singapore and at Cherating in Malaysia — both typically resolve within hours but serve as a reminder that a decade-old cable, even a well-maintained one, is a living piece of infrastructure subject to maintenance events, sporadic congestion, and the constant evolution of regional peering arrangements across its eleven landing stations.
| Status | ✓ Normal |
|---|---|
| RTT | 75.42 ms / base 105.97 ms |
| Last checked | 2026-04-19 04:31 |
Monitored using RIPE Atlas probes. Open monitoring →
| Min | Avg | Max | # | |
|---|---|---|---|---|
| 7 days | 74.6 | 107.1 | 254.8 | 49 |
| 30 days | 74.6 | 105.4 | 254.8 | 52 |
| 60 days | 74.6 | 105.4 | 254.8 | 52 |
Find the actual cable routing distance between any two cities
Open Calculator →