Alberto Caraballo
Route diversity is the easiest thing in network design to claim and the hardest to verify. It usually gets reduced to a count. Four cables sounds better than two. Seven sounds better than four.
The count tells you how much capacity reaches a coastline. It tells you almost nothing about what happens when something goes wrong, because two cables that leave from the same place and arrive at the same place are one route carrying two cables.
Four conditions, not one
A genuinely diverse second route changes four things, and changing only one of them produces additional capacity rather than an additional path.
- The landing point. A second cable into the same facility inherits that facility’s power, access and operational risk.
- The trunk. Two systems following the same seabed corridor share the same anchors, the same fishing grounds and the same seismic exposure.
- The terminal metro. If both systems end in the same metropolitan area, a fault, a congestion event or a commercial dispute there affects both. The unit here is the metro, not the state. Two landings a few hundred kilometres apart sit on different grids and different inland fiber.
- The operator. Two paths bought from the same party are one commercial relationship, whatever the physical layer looks like.
Most procurement checks the first condition and assumes the rest. That is the gap between redundancy, which protects against a cable cut, and diversity, which protects against everything else.
Cancún, over twenty-five years
The Mexican Caribbean is a useful place to apply the test, because the record is long and entirely public.
Three submarine systems are confirmed in service at Cancún. MAYA-1 entered service in 2000 with two fiber pairs. ARCOS-1 followed in 2001, a ring system whose end-of-life design capacity is 8.4 Tbps. AMX-1 arrived in 2014. A fourth, GigNet-1, was announced in 2021 with a planned ready-for-service date in late 2022, and no public source confirms it in operation.
Run the four conditions against the three confirmed systems and the terminal metro is where the concentration shows. MAYA-1 lands at Hollywood. ARCOS-1 at North Miami Beach. Both sit inside the same South Florida metropolitan area, sharing a grid, a hurricane corridor and the same inland fiber. AMX-1 is the exception: alongside its Miami landing it also reaches Jacksonville, more than 500 kilometres north, which is a genuinely different termination.
South Florida is where the Americas interconnect, and for twenty-five years routing there has been the correct commercial decision for almost every operator on this coast. Miami’s gravity is real and it was earned.
The consequence is still worth stating plainly: a market can add cables for a quarter of a century and leave most of them terminating in one metro. No additional capacity across the same crossing changes that, because the constraint was never capacity.
What changes it is a different vector. MANTA, now in build, runs a repeatered trunk from San Blas in the Florida Panhandle to Cartagena, with branches to Veracruz and Cancún in Mexico and to North Miami, plus a stubbed branching unit toward the Caymans. Two things there are new rather than incremental. It is the first international system to enter the Gulf of Mexico, which gives Veracruz a landing point on a coast that had none. And San Blas is more than 700 kilometres from Miami, on a different grid, opening a path north toward Atlanta and the Carolinas rather than another arrival into the south of the state.
What to ask the party selling you a second route
The point of the four conditions is that they are verifiable before signature rather than after an incident. Applied to any proposal, in any market:
- Which facility does it land in, and who else is already in that building?
- Does the trunk follow a corridor that any of my existing paths already use?
- Which metro does the traffic terminate in, and does that metro already carry my other routes?
- Am I buying this from a party I already depend on for something else?
Four yes answers mean more capacity, which may be exactly what is needed and should be priced as such. Four no answers mean a second route. Most proposals sit somewhere between, and the useful conversation is about which of the four conditions actually changed.
A count cannot answer any of those four questions. That is the reason it is the wrong unit.
Landing dates and landing points are drawn from TeleGeography’s submarine cable database and Submarine Networks. MANTA details are from SubCom and the consortium announcement.
