Asia Direct Cable (ADC)

Asia Direct Cable (ADC) is a 9,988 km intra-Asian submarine cable ready for service in 2024. It connects seven landings across seven countries and territories: Tuas in Singapore, Sriracha in Thailand, Quy Nhon in Vietnam, Shantou in mainland China, Chung Hom Kok in Hong Kong, Batangas in the Philippines, and Maruyama in Japan. The cable is owned by a consortium of major Asian telecoms and one hyperscaler: Meta (Facebook), China Telecom Global, China Unicom Global, NTT, PLDT, SingTel, Tata Communications, and Viettel. It is among the more recent additions to intra-Asian fibre capacity, alongside APRICOT, SJC2, and SJC.

ADC's architectural decisions reflect a pragmatic reading of Asian regulatory geography in the mid-2020s. The cable lands in both mainland China (Shantou) and Hong Kong (Chung Hom Kok) — a configuration some recent cables have declined — but it avoids touching the United States, which keeps the cable out of the scope of US regulatory review. For Asian carriers and for Meta's Asia-Pacific operations, ADC is a regional utility cable that does not inherit trans-Pacific jurisdictional complexity.

69 ms Singapore to Japan

Our monitor measures ADC between Tuas in Singapore and Maruyama on Japan's Pacific coast — the cable's longest end-to-end segment. Over 30 days we collected 45 forward-direction samples, hitting multiple target endpoints as our target selection adjusted for path quality:

The 69.0 ms minimum is notable. ADC's full length is 9,988 km, which gives a theoretical round-trip physics floor of 97.6 ms. Our measurement is below the full-cable floor — about 0.71× — which means the segment we are measuring does not use the full cable length. A packet from Tuas directly to Maruyama does not need to touch Sriracha, Quy Nhon, Shantou, Hong Kong, or Batangas; it takes the trunk route and exits at Maruyama.

The great-circle distance from Tuas to Maruyama is about 5,200 km. At 69.0 ms round-trip, the measured fibre path is approximately 7,050 km — roughly 36% longer than the great-circle, which is typical for Asian cables that curve around the South China Sea rather than taking the deep-water geodesic. At 1.32× the great-circle floor, ADC's Singapore-Japan performance matches the tight end of what we have documented on SJC2 (1.32× on the same corridor) and is noticeably tighter than SJC (2013, slightly longer route).

Seven landings, one regional mesh

Each ADC landing sits in a major Asian internet market. Singapore, Hong Kong, and Tokyo are the three largest peering hubs in Asia; Shantou connects directly to mainland China's internet backbone; Batangas feeds the Philippine data-centre cluster; Sriracha serves Bangkok; Quy Nhon plugs Vietnam into the regional mesh. Each landing is chosen to deliver high-capacity connectivity to the ASN clusters that drive regional traffic.

Three of these landings deserve specific attention:

Quy Nhon, Vietnam, is part of a rapid Vietnamese cable build-out. Alongside SJC2 (2025) and APRICOT (2025), ADC's 2024 Quy Nhon landing gives Vietnam direct connectivity to Japan, Singapore, and China without transiting through Hong Kong or Thailand. For Vietnamese cloud operators targeting regional customers, this is a structural capacity upgrade.

Shantou, in Guangdong province, is one of the most-landed Chinese cable stations. ADC's Shantou landing provides mainland-China capacity in a manner that several US-involved cables (notably PLCN) have had to avoid. For Chinese telecoms, ADC is commercially useful precisely because its ownership structure allows the Shantou landing without political friction.

Batangas, Philippines, is the historical Pacific cable-landing hub, with a dozen cables terminating in the region. ADC's Batangas landing is less strategic than some of the more recent Philippine landings at Baler and Davao, but it plugs into existing infrastructure efficiently.

Intra-Asian hyperscaler economics

ADC is an example of an increasingly common cable-financing pattern: a consortium mostly composed of Asian carriers, with Meta as a single hyperscaler anchor partner. Meta gets dedicated capacity for its Asian data-centre traffic (backend replication, content delivery, interactive sessions). The carriers get capacity on a cable they partly own, with Meta's financing making the overall economics work.

This model sits between the pure consortium cables of the 2000s and the hyperscaler-only cables of the 2020s (Marea, Equiano, Nuvem). The advantage for carriers is that Meta's capacity demand justifies the cable's commercial case, so capacity is available at a better price point than if the carriers had to fund it alone. The advantage for Meta is that the carriers contribute to operating costs and provide the regulatory and peering relationships in each landing country.

Expect this pattern to continue. Cables that span high-regulatory-friction regions (Asia-Pacific, Middle East) increasingly have mixed carrier-and-hyperscaler consortium structures, because neither group alone can deliver the commercial case plus the regulatory access needed to land in each target country.

Multiple fibre pairs in a 2024 design

ADC was designed with a modern multi-fibre-pair architecture suitable for coherent modulation. Public disclosures of exact pair counts and capacity are limited, but the cable is in the same class as its 2024-2025 contemporaries — approximately 8 fibre pairs with per-pair capacity in the 18-24 Tbps range when lit with current-generation transponders, total capacity well over 140 Tbps.

The cable's operational profile is still maturing. At approximately two years of service, ADC is in the early phase where carriers are migrating production traffic onto its capacity, and where our measurements reflect this transition. The variation in target selection and hop counts (11 vs 16 depending on which Japanese endpoint is reached) shows different routes with varying degrees of ADC utilisation. Over the next several years, as traffic engineering matures, expect the minimum-RTT measurement to stabilise at the 69 ms baseline with reduced variance.

What our data proves

• ADC delivers Tuas → Maruyama at 69.0 ms minimum, 1.32× the great-circle physics floor. Among the tightest intra-Asian measurements we have documented, matching SJC2's best performance on the same corridor.
• Below full-cable-length floor indicates efficient segment routing. ADC's direct Singapore-Japan path uses about 7,000 km of fibre (not the full 9,988 km), confirming that production traffic takes the direct trunk rather than detouring through intermediate landings.
• Mixed consortium ownership is commercially functional. ADC's partnership between Asian carriers and Meta as single hyperscaler anchor delivers a cable that serves both groups' needs without trans-Pacific regulatory complications.

ADC is not a glamorous cable. It does not have a celebrity architect, a Google blog post, or a dramatic geopolitical backstory. What it has is seven well-chosen landings, a competent commercial structure, and latency close to physics floor on its primary route. That makes it a workhorse in the new Asian cable mesh — exactly the kind of infrastructure that becomes invisible precisely because it works.

Try it yourself

Live data on the ADC cable page. For contrast see APRICOT (2025 hyperscaler-led), SJC2 (2025 consortium), PLCN (2022 US-landing trans-Pacific), and FLAG-NAL (2001 regional loop).