SAT-3/WASC

Based on 58 RIPE Atlas measurements from GeoCables monitoring infrastructure, March–April 2026.

SAT-3/WASC — the South Atlantic Terminal / West Africa Submarine Cable — is, on paper, a fossil. Its 14,350 kilometres of fibre were laid down in 2002 on behalf of a consortium of 32 telecom operators, ranging from AT&T and BT to Maroc Telecom and Angola Telecom. It has twelve landings across ten countries, running from Sesimbra on the Portuguese coast down through West African ports — Dakar, Abidjan, Accra, Lagos, Douala, Libreville, Cacuaco — to Melkbosstrand near Cape Town. Its commercial structure belongs to an era that the submarine-cable industry has largely moved past.

And yet, twenty-four years after it was first lit, our measurements place its performance at 1.017× the theoretical physics floor. Among published cables in our monitoring queue, that is one of the tightest floor ratios we see anywhere — better than many modern 2020s hyperscaler-funded cables. SAT-3 is still delivering essentially the shortest possible round-trip that its geometry allows, for the specific Portugal-to-South-Africa path our probes measure.

What the measurements show

Our probes exercise SAT-3 in both directions between Sesimbra, Portugal and Melkbosstrand, South Africa — the two endpoint landings that bookend its 14,350 km arc.

The theoretical floor for a 14,350 km optical path is 140.44 ms. The measured floor is 142.82 ms. That is a 1.7% overhead — essentially the minimum you can achieve on a real network once you account for the transit-operator equipment on both ends. For comparison, many modern cables we've published show floor ratios of 1.1–1.3× or even below floor when spurs are used instead of the full cable length. SAT-3 is sitting at the idealised end of the distribution.

The 32 ms gap between forward-direction min (142.82) and reverse-direction min (174.02) isn't an asymmetry story — it's a 3-hop difference in median path length (13 hops forward, 16 reverse), reflecting the normal BGP-driven variation in how transit providers hand traffic off on each end. Above the floor, both directions show similar 22 ms standard deviation — well-behaved routing with occasional minor detours, none of which dominate the measurement.

The consortium model, preserved

Reading the SAT-3 owner list is like reading the industry directory of 2000. AT&T, BT, Deutsche Telekom, KPN, Orange, Vodafone, Verizon, Tata Communications, Telxius, Altice Portugal, Cogent, Singtel, Chunghwa Telecom, KT, PCCW, Telstra, BICS, Sparkle, China Telecom, Maroc Telecom, Orange Côte d'Ivoire, Telecom Namibia, Camtel, Angola Telecom, Mauritius Telecom, Ghana Telecommunications, NATCOM Nigeria, SBIN Benin, Liquid Intelligent Technologies, Cyta — 32 separate entities, each holding a capacity share proportional to their investment, each with their own PoP and their own commercial agreements downstream.

This is not how submarine cables are built anymore. The 2020s model, visible in cables like Equiano (Google, 2022), Nuvem (Google, 2025), Malbec (Meta, 2021), and Bifrost (Meta, 2024), is single-owner or small-consortium construction where a hyperscaler funds the cable entirely and sells capacity on commercial terms. The older pattern — capex shared across dozens of national telcos, each retaining access rights in perpetuity — has almost entirely disappeared from new construction.

SAT-3 is one of the last major cables still operating under that older pattern, and the economic logic that sustains it is intact: every one of the 32 owners still has customers whose traffic benefits from the cable, every one still paid their share of the capex 24 years ago, and none have any incentive to exit as long as the physical infrastructure continues to work. The consortium doesn't need new capital; it just needs the cable to keep running.

The Africa connectivity monopoly that SAT-3 created

When SAT-3 came into service in 2002, it was the only high-capacity submarine cable serving most of West and Central Africa. For countries like Cameroon, Gabon, Benin, and Côte d'Ivoire, the entire international internet transit of the nation flowed through a single SAT-3 landing operated by the local consortium shareholder. This was, predictably, extraordinary commercial leverage.

For roughly a decade — 2002 through 2012 — SAT-3 landings functioned as bottleneck assets, and the consortium owners sold international capacity to their domestic competitors at premiums that reflected the absence of alternatives. Anyone running a data centre, ISP, or corporate link in Douala, Libreville, or Cotonou paid whatever the local SAT-3 owner charged. This pricing structure was widely criticised at the time as a brake on African internet adoption, and it was only broken by the construction of later cables that bypassed the consortium model entirely.

WACS (2012) and ACE (2012) added competing West African routes with different ownership structures. A decade later, Equiano (2022), PEACE (2022) and 2Africa (in progress) added further capacity, most of it either single-owner (Google) or hyperscaler-led. The SAT-3 pricing premium has long since collapsed; prices for Africa-Europe transit are now set by the newer cables and by the hyperscaler cloud providers' own economics.

What SAT-3 continues to provide, even in this competitive environment, is reliable baseline capacity that its consortium owners can use without paying anyone else. For a Camtel or an Angola Telecom, owning their existing SAT-3 share is operationally cheaper than buying equivalent capacity on a newer cable — even if the newer cable offers better technology and higher capacity ceilings.

Why 1.017× floor at 24 years

The performance number is the surprise. Submarine cables of this vintage have often been quietly retired once their repeaters — the in-line optical amplifiers every 60–100 km along the cable — begin to degrade. Others have been upgraded with modern DWDM equipment at the landing stations, which multiplies capacity several times without touching the cable itself. SAT-3 has received multiple such upgrades over its 24-year life, most recently pushing its design capacity from the original sub-Tbps level into the low tens of Tbps.

But upgrades at landing stations don't change the physics floor. The physical length of the cable is fixed; the speed of light in silica fibre is fixed; the theoretical minimum round-trip for 14,350 km of path is 140.44 ms and nothing will improve that number short of laying a shorter cable. What SAT-3 demonstrates at 1.017× floor is that the optical path itself is still clean. Repeaters along the route still work. Signal-to-noise ratios still allow the ~140 ms latency to make it through without retransmissions or path diversions. The consortium's ongoing maintenance — however distributed across 32 operators — is still keeping the physical asset healthy.

What we're watching

The core metric is whether that 1.7% overhead stays flat over time. A gradual rise in the floor would be the earliest signal of something degrading — not catastrophically, but enough to matter for a 24-year-old asset. Any of the 32 owners might then propose retirement, or an additional upgrade, or eventual replacement. Until that happens, SAT-3 continues to demonstrate something unusual in a market that has largely moved on: a consortium-era cable, still held collectively by a cross-section of the global and African telecom industry, still running at essentially the best performance its geometry allows, twenty-four years into service.