Summary

  • DE-CIX INTERCONNECTION MEXICO S. DE R.L. DE C.V. operates a Mexican exchange that launched in 2025 across Mexico City and Querétaro; the operator's own page reports nearly 30 connected networks, more than 400% traffic growth in 2026 and a peak near 300 Gbit/s, read on 2026-11-09.
  • The same page yielded two different peak figures across two retrievals — near 300 Gbit/s and approximately 200 Gbps — and publishes no 2025 baseline in Gbit/s, so its growth percentage cannot be converted into an absolute volume.
  • The LACNIC member record behind the entity carries registry country Mexico and a confidence score of 0.82, and carries no ASN, no prefix, no website, no ownership and no leadership.
  • The stake for an ISP, content network or investor deciding whether to connect: a port count is not a substitute for the incumbent path. Until a named participant resolves to an ASN and a live session, the fabric is inventory, and inventory is not leverage.

Everything measured here was read on 9 November 2026. The participant and traffic numbers come from one operator page, retained as a provider-reported excerpt rather than an opened raw document. The registry facts come from a directory record last updated on 20 June 2026. No independent peering record, exchange member list, regulator filing, corporate registry document or incumbent interconnection position was retrieved, and this article asserts none of them.

Two cities, one fabric, one number

The sentence that carries the whole commercial claim is short. "Since launching in 2025, DE-CIX Mexico has rapidly become one of the country's leading neutral interconnection platforms, connecting Mexico City and Querétaro in a single distributed ecosystem," reads the operator's Mexico City location page (de-cix.net). Two metros, one distributed fabric, one brand, and a launch date that puts the platform barely a year old at the moment of reading.

The quantity attached to that fabric is smaller than the adjectives. "Nearly 30 networks connected across Mexico City and Querétaro," the same page reports. Thirty is not a rounding of three hundred. It is a count you could seat in a small conference room, split across two cities separated by roughly 200 kilometres of Mexican plateau, on a platform that describes itself in the same breath as one of the country's leading neutral interconnection platforms.

That is the first arithmetic worth doing, and it is not hostile to DE-CIX. Nearly 30 networks in the first year of a greenfield exchange in a market with entrenched domestic carriage is a real achievement. But it is an achievement measured in tens, and the language around it is measured in superlatives. When a press-facing page says "leading" and then supplies a number in the twenties, the reader learns something about how the operator weighs adjectives against participants.

Two peaks, one page, no reconciliation

Then there is the traffic figure, which is where the page stops being merely promotional and becomes genuinely difficult.

The body text reports "More than 400% traffic growth in 2026, with peak traffic near 300 Gbit/s." The page also carries the headline "DE-CIX Celebrates First Year in Mexico With 200 Gbps Milestone." A second retrieval of the same URL, made fifty-two seconds later, summarised the same page as reaching "approximately 200 Gbps" at the first anniversary.

Both figures are retained here. Neither is reconciled by the source.

There is a benign reading, and it deserves to be stated plainly rather than buried: 200 Gbps at a first anniversary and a near-300 Gbit/s peak later in 2026 could both be true, describing different dates on a rising curve. That reading is plausible. It is also unverifiable from the page, because the page dates neither number. We are given a milestone headline without a milestone date, and a peak figure without a measurement period, an averaging window, or a statement of whether the two metros are counted together or separately.

A reader who wants the absolute traffic cannot get it. The 400% growth claim is a ratio without a base. Multiply anything by five and you get four hundred per cent growth: a jump from 4 Gbit/s to 20 Gbit/s is more than 400%, and so is a jump from 40 to 200. Without the 2025 baseline in Gbit/s, the percentage is a shape, not a size. It tells you the line went up steeply. It tells you nothing about where it ended.

This matters more than it looks. Growth percentages from small bases are the cheapest numbers in infrastructure marketing, because the smaller the base, the more spectacular the arithmetic becomes at no extra cost. That is not a claim about intent. It is a claim about information: a ratio whose denominator is withheld cannot discipline anyone's expectations, including the operator's own.

What the registry can actually say

On the other side of the ledger sits a much smaller set of facts, and they are of a completely different kind.

LACNIC's public associated-organizations listing records DE-CIX INTERCONNECTION MEXICO S. DE R.L. DE C.V. as a member with registry country Mexico, and the directory record BTW holds on the entity — last updated 20 June 2026 — carries that membership as its sole resource facet. Registry country Mexico. Legal type: private company. Registration jurisdiction: MX. External identifier: lacnic:member:de-cix-interconnection-mexico-s-de-r-l-de-c-v:MX, imported by the member-directory process, with a confidence score of 0.82 (milacnic.lacnic.net).

The record then stops. It holds no ASNs. No prefixes. No RDAP or BGP evidence. No website. No leadership contacts. No ownership. No relationship edges to a parent, an affiliate or a facility operator. The record's own stated gap says as much in the record's own words: complete legal registration, official website, ownership and leadership are missing, so the resource record does not yet point to the real legal entity.

Read that alongside the operator page and the asymmetry becomes concrete. One side of the ledger says thirty networks and four hundred per cent. The other side cannot name a single address block, a single routing instance, or a single human being responsible for the company that holds the membership. The BTW directory entry for the entity Only for Strategic Circle is honest about exactly this: it separates a verified RIR membership from assertions about service, footprint and control until something supports them.

One negative result needs its boundary stated, because absence claims are where careful reporting usually falls over. A retrieval of the LACNIC public list on 9 November 2026 did not surface the Mexican entity under its exact legal name in the portion of the page returned. That is one page, one retrieval, one view — not proof of absence from LACNIC's records, and not evidence that the membership is defective. The membership claim rests on the directory record's own account of it, not on a verbatim row read in this run.

The gap is observability, not a verdict

The lazy version of this article writes itself: an exchange that will not name its members must be a shell. That conclusion does not survive contact with the evidence, and it should not be printed.

What the record actually shows is an observability and attribution deficit. It weakens independent verification. It prevents anyone outside the operator from attributing the claimed traffic to named counterparties, or from tying the Mexican legal entity to the ASNs, prefixes, route-server infrastructure, exchange LAN and facility contracts used to run the service. Those are real constraints on what a reader can check.

They are not proof that the exchange is idle, fictitious or ineffective. An exchange can carry genuine traffic and still publish only aggregates. A commercial operator has no obligation to hand competitors a member list. Plenty of working fabrics are opaque from the outside, and opacity is a disclosure choice rather than a technical failure.

The distinction matters because the two readings lead to opposite practical conclusions. If the participant count is real and the fabric is genuinely used, then Mexico has acquired a second interconnection option and the interesting question is how fast it becomes a substitute. If the count is inflated by remote, reseller-mediated, duplicate or provisioned-but-inactive connections, then the port count overstates local interconnection and the leverage story collapses. The page does not define what "connected" means — ordered, provisioned, active, local or remote.

That single undefined word carries the entire difference between the two cases.

Where capacity becomes leverage

Here is the chain that actually decides whether a neutral exchange changes anything, and it is longer than most coverage admits.

Available ports come first. Then named live sessions with economically relevant networks. Then meaningful migration of traffic and of spend away from the incumbent path. Then, and only then, a credible substitute — one a buyer can actually threaten to use in a negotiation. Price pressure is the last link, not the first.

Each link needs separate evidence, and a participant count and a peak figure establish only the first.

So the question about DE-CIX Mexico is not whether it exists or whether it grew. The evidence supports both. The question is whether a Mexican access network, content network or enterprise can now route and buy differently because those two fabrics exist.

Answering that requires two things joined to the same named participant and the same period: a route or traceroute series showing a defined class of Mexican traffic moving off incumbent transit onto exchange peering, and comparable invoices or quotes showing a lower all-in cost for the same capacity, term and service level. Neither is in hand. Neither is even close to in hand.

This is where the aggregate numbers earn their keep as a warning rather than a proof. Thirty networks across two metros is a plausible seed community — enough to make a route server useful, not enough to describe a market. And a peak traffic figure, even taken at face value, has no denominator. Nobody has published what share of relevant Mexican traffic that peak represents. Against total national traffic, a few hundred gigabits per second is either a serious share or a rounding error, and the page gives the reader no way to tell which.

There is a second gap that the operator's own framing makes visible. A link from the Mexican platform to a larger ecosystem in Dallas, mentioned in the retrieval layer's summary rather than in the quoted excerpt, cuts both ways.

Cross-border connectivity gives a young exchange instant reach and a credible value proposition to any network already buying North American transit. It also means that traffic which touches a Mexican port does not necessarily stay in Mexico. An exchange can grow its own numbers while the economics of local carriage barely move, because the traffic it handles is leaving.

What would settle it

The falsifiable test is not complicated, which is why its absence is notable. A named participant list, or an independent peering record, showing Mexican access networks, ISPs or content networks present at the exchange — particularly networks that are not affiliates or wholesale customers of the incumbent carrier — would overturn a concentration reading immediately. So would a single documented case in which a named network moved a defined class of traffic onto the exchange and saw its comparable all-in cost fall.

None of that exists in the public record as read on 9 November 2026.

What exists is a launch year, two metros, a count in the twenties, a growth rate without a base, two peak figures that do not agree, a marketing claim of neutrality that no published access criteria or non-discrimination terms yet support, and a registry row that confirms a legal entity exists and then falls silent.

That is not nothing. A LACNIC membership is a real, narrow, checkable fact, and an exchange that has spent a year connecting networks across two Mexican cities has done something harder than publishing a press release.

But a reader deciding whether to buy, connect, invest or regulate is being asked to accept a momentum story on the strength of numbers the operator controls, produced by a measurement layer only the operator can see.

The honest position sits between the two easy ones. DE-CIX Mexico is a claimed two-metro exchange operation with provider-reported participation and traffic, growing quickly from a base nobody has disclosed, on a fabric whose participants nobody outside the company can name.

Whether that becomes leverage over Mexican paths and prices is still an open question, and the people best placed to answer it have not yet published the part that would answer it.