Enrique Ortegon
When a submarine cable reaches a new market, the coverage follows a predictable shape. Route, design capacity, landing cities, ready for service date. Underneath it sits an assumption that rarely gets examined: that arrival equals access.
Capacity arrives with the cable. An interconnection point does not. Two coastal cities that got there by opposite routes make the distinction concrete.
Two paths to the same place, neither of them the cable
Marseille is the clearest case of a landing environment doing the work. A carrier-neutral operator opened its first facility there in 2014, and DE-CIX established an exchange inside it in 2015. Within three years the first building had gone from 46 networks to 134. Some cable systems placed points of presence directly inside the facility so they could reach backhaul providers on site rather than backhaul out to find them. By 2020 the campus hosted more than 150 network service providers and the city was being described as the world’s ninth largest internet hub (Interxion, a Digital Realty company, July 2020).
The port authority built to match. Its shared landing infrastructure was designed so that terminal equipment could be installed flexibly at other landing stations or at carrier-neutral facilities, rather than only where the cable physically comes ashore (Port of Marseille Fos).
Fortaleza reached the same destination from the opposite direction. Atlantis-2 entered service there in 2000 and more than a dozen systems followed, landing at stations owned by different parties including Angola Cables, Embratel, GlobeNet and Telxius. What converted that concentration into an interconnection point was not any single operator. It was a national exchange strategy. A 2026 study of peering capacity at the world’s leading public internet exchanges placed four South American exchanges in the global top thirty, Fortaleza among them, and attributed Brazil’s showing to a coordinated national approach to traffic exchange (LACNIC, July 2026).
One city arrived through a single neutral operator. The other arrived through many competing landing parties and a national exchange program. Neither arrived by landing cables. Fortaleza had cables for close to two decades before it appeared in that ranking.
The cable is the precondition. It has never been the variable.
One cable, several owners, one doorway
The reason this matters more now than it did five years ago is structural, and it sits in how modern systems are built.
Older cables were turnkey. A single vendor supplied the wet plant on the seabed and the terminal equipment on shore, and the landing station was where one operator’s cable ended. Newer systems separate the two. Capacity can be assigned as wavelengths, as spectrum, or as whole fiber pairs to different parties on the same physical cable, each running its own equipment on its own upgrade cycle. Diverse fiber pair ownership is now a defining feature of recent systems in the region (TeleGeography, Latin America Bandwidth Market Review, June 2026). Ciena, which supplies terminal equipment for cables of this kind, describes the model as shifting traffic patterns away from landing station to landing station and toward point of presence to point of presence and data center interconnection.
So a single landing point now serves several independent parties who each terminate differently and each answer to different customers. It stops being an endpoint and becomes a shared doorway.
Doorways are defined by who holds the key, and that matters commercially to everyone selling on the cable. A fiber pair terminating where the buyer can reach dozens of networks inside the building is a materially better product than the same pair terminating where they reach one and then pay to get somewhere useful. Open access at the landing point raises what capacity is worth. It does not compete with it.
Which makes access, not capacity, the question worth asking
Three questions separate a coastal facility that will become an interconnection point from one that will remain a landing point, whatever it calls itself:
Who can bring backhaul into the building, and on what commercial terms?
Does the party operating the landing point sell services that compete with the networks landing there?
Once inside, is there anything to interconnect with locally, or does traffic still have to travel inland to find a counterparty?
Regulators have converged on the second question in particular. India has regulated access to cable landing stations since 2007 on the basis that a landing station is an essential facility, with amendments following in 2012 and revised charge regulations in 2018. In 2026 Qatar’s Communications Regulatory Authority issued a framework requiring wholesale access fees to be fair, reasonable and non-discriminatory, and expressly prohibiting landing parties from giving preferential pricing or faster provisioning to their own downstream retail divisions.
It is worth noting the counter-argument, because operators in regulated markets have made it forcefully. Several countries never regulated landing station access, and others withdrew the rules once competition at the landing point had matured. Read that way, regulation is a symptom of concentration rather than a permanent feature. Both readings lead to the same place. The access question gets resolved either by market structure or by a regulator. Structure is considerably faster.
The same problem wearing different infrastructure
There is a precedent for this on land.
At an international fiber crossing, two national networks with no commercial reason to meet still have to meet somewhere. The constraint is not capacity. It is governance: whose building, whose rules, whose competitive interest. Markets resolve it by placing the meeting point with a party that has no stake in the outcome.
A cable landing has the same shape. A subsea system and the terrestrial networks it feeds have to converge, and the convergence point belongs either to a participant or to a neutral party. The physical infrastructure is entirely different. The operating question is identical.
That logic is visible in the model taking shape on Mexico’s coast. The MANTA system, built to open cable specifications, will land at facilities in Cancún and Veracruz that combine cable landing infrastructure with a neutral interconnection environment, with onward paths into Mexico’s main network corridors (MANTA consortium, March 2026). Three carriers holding overlapping commercial interests on the same cable agreed the landing point should be operated by a party that competes with none of them.
What to watch
Several systems will come ashore in Mexico and the wider Caribbean basin over the next two years. Each one will be a commercial success or failure for the parties who own capacity on it, and part of that outcome will be settled on land, by how much of the market their buyers can reach from the point where the cable arrives.
What separates a coastal market that becomes an interconnection point from one that stays a landing point will not be how many terabits come ashore. Marseille took roughly six years. Fortaleza took closer to twenty. The difference was never the cables.
