Summary

  • The public network record for AS61832 carries the long name DB3 SERVICOS DE TELECOMUNICACOES S.A. while labelling the organisation "Alloha Fibra - N/NE" and listing "GIGA MAIS - Wirelink - Wire." among its aliases, so the brand named in this directory entry is not the entity that the record names as the network's holder.
  • TeleSíntese reported a board decision taken on 11 August 2025 that replaced diretor-presidente Lorival Nogueira Luz Júnior and installed Felipe Gonçalves Matsunaga as interim chief executive, described EB Capital as controlling the company and as also owning Alloha Fibra, and put the base at roughly 1.64 million subscribers and a 3.1% share.
  • The LACNIC RDAP object for AS61832 could not be retrieved on 2026-10-17, so the registry evidence in this report rests on a self-reported industry record rather than on the authoritative allocation object.
  • Several assertions attached to the directory entry — a different named chief executive, a lower subscriber figure, a city count and a brand list — are not corroborated by the retained sources, and on the chief executive the retained report points elsewhere.
  • The transferable lesson for anyone modelling a consolidated access group is that brand, registrant and equity holder must be observed as three separate objects, because a shock to one of them does not propagate automatically to the other two.

What one network record actually names

The record at the centre of this report is a public network entry for AS61832. Its most consequential field is not the traffic figure but the name attached to the object. The entry lists the long name as DB3 SERVICOS DE TELECOMUNICACOES S.A., presents the organisation as "Alloha Fibra - N/NE", and lists "GIGA MAIS - Wirelink - Wire." under the aliases heading (PeeringDB record for AS61832).

Read carefully, that single object carries three naming layers at once. There is a legal-style long name in block capitals. There is a group-style organisation label that points to a regional division of a larger structure. And there is a set of commercial brands, written with the punctuation of an operator's own shorthand. The directory entry that prompted this report is filed under a fourth name again, a commercial brand presented as a business-to-business arm. Four names, one autonomous system.

The same record reports traffic in the 1-5Tbps range with mostly inbound traffic and lists IX.br exchange points including Fortaleza, São Paulo, Rio de Janeiro and Brasília, with the record last updated in March 2026. Those are the details that make the object interesting to an operator and, for exactly that reason, the details that require the most caution. A traffic band and a list of exchange points are self-reported operational descriptors. They tell a reader where a network says it is present and roughly how much traffic it says it carries.

They do not tell a reader who owns the network, who may sell it, who must answer for its routes, or how much of the group's revenue arrives through it.

The distinction is not pedantic. In Brazilian access markets, as in most consolidated markets, the entity that appears on an interconnection record and the entity that appears on a customer invoice are frequently not the same legal person, and neither is necessarily the entity in which the equity sits. Consolidation accelerates that divergence, because each acquisition brings its own legacy registrations. A group that has bought several regional operators inherits several autonomous systems, several registry objects and several brand names, and rationalising them is slow, optional and rarely announced.

The board decision, as reported

The second retained surface is a trade report on a management change. According to that report, GIGA MAIS FIBRA TELECOMUNICAÇÕES S.A. announced the replacement of its diretor-presidente, Lorival Nogueira Luz Júnior, by a decision approved by the Conselho de Administração at a meeting held on 11 August 2025 (TeleSíntese report on the presidential transition).

The same report names Felipe Gonçalves Matsunaga as interim chief executive, describing him as an investor-relations director and a partner at EB Capital, and describes EB Capital as controlling the company and as also owning Alloha Fibra. It puts the base at approximately 1.64 million internet subscribers, a 3.1% share, and describes the company as the fourth-largest access provider in Brazil as of August 2025.

Three features of that report deserve emphasis, and one deserves scepticism. First, it is a dated record of a governance action, not a comment on strategy: a board met on a specific day and changed the person at the top. Second, the interim designation matters. An interim appointment is a holding pattern, and the report does not establish who holds the position now, whether the interim arrangement was confirmed, or what the terms of the departure were. Third, the description of control is a description by a trade outlet, not a publication of the equity register.

The feature that deserves scepticism is the market-share figure. A 3.1% share and a 1.64 million subscriber base are precise-sounding numbers, but they arrive through a single outlet with no accompanying methodology, no reference date beyond August 2025, and no regulator dataset attached. They are usable as a dated snapshot of what was reported. They are not usable as a measurement. Anyone building a model on them is building on a press citation.

Where the entry and the evidence part company

The directory entry that frames this commission is more expansive than either retained source. It describes the entity as the business-to-business arm of the operating company, places the group under EB Capital control with named principals, gives a subscriber figure near 1.5 million across more than 280 cities, refers to a nine-brand roll-up with named brands, and names a chief executive who took the role in August 2025 after a career at two large mobile operators.

Set that against the retained sources and the results divide into three buckets.

Some of it is corroborated in substance. EB Capital is described as the controller in the trade report. A management change did occur in August 2025. The group context around Alloha Fibra is named in both the trade report and the network record's organisation label.

Some of it is uncorroborated but not contradicted. The subscriber figure differs from the reported 1.64 million, the city count of more than 280 appears nowhere in the retained set, and neither does the brand list. These may well be accurate. This report cannot say so, and does not.

One item is contradicted. The entry names one person as chief executive following the August 2025 change. The retained trade report names a different person, in an interim capacity, on a specific date. That is not a rounding difference or a timing artefact that a reader can wave away. Either the entry is describing a later appointment that post-dates the retained report, in which case the corroborating document was not retrieved, or the entry is wrong. The honest statement is that the retained evidence points to the interim appointment and does not support the named successor.

That single divergence is the reason this article exists in its present form. A directory entry is a working hypothesis about an entity, assembled from whatever surfaces were available when the record was written, and it is useful precisely because it aggregates. But aggregation also imports assertions whose provenance is thinner than the surrounding text implies. The entry itself flags the gap: it asks for verification of the entity's legal country, current services, visible assets, parent or subsidiary links and responsible public leadership contacts.

That is an unusually candid self-assessment, and it should govern how the entry is read.

Why three layers instead of one

For an executive or an investor, the value of separating these layers is not taxonomic tidiness. It is that each layer answers a different question, and each fails in a different way.

The registry layer answers who is accountable for the routing object. The registered holder of an autonomous system is the party that can request a change of delegation, that appears when a route is disputed, and that counterparties contact when something breaks. When the object's long name is a company that does not appear in the brand architecture, that is not necessarily a problem. It is a fact that must be recorded, because a routing dispute will be resolved with reference to that name and not to the brand on the invoice.

The equity layer answers who can direct the business. The controller decides whether to keep buying regional providers, whether to raise price in a contested municipality, whether to fund a network upgrade, and whether to refinance. Where the trade report describes EB Capital as controlling the company and as also owning Alloha Fibra, that description is doing real analytical work: it says the decision rights sit above the operating company, and that a group-level capital decision can be taken without any change at the level of the network registrant.

The commercial layer answers what customers and counterparties see. A brand can be extended across newly acquired bases quickly. Legal consolidation cannot. That asymmetry is the ordinary condition of a roll-up, and it produces the specific hazard this article is written to name: an observer who tracks only the brand will systematically overestimate how integrated the group is, and an observer who tracks only the registry will systematically underestimate how much of the group's commercial behaviour is directed from above.

The economic mechanism runs through three variables that a consolidated access group cannot escape. Subscriber density inside each municipality determines the cost to serve and the pricing power available in that municipality. The price paid per acquired base determines the return on the capital deployed, and it is set in a negotiation that the brand does not control. The cost of refinancing the debt taken on to buy those bases determines whether the group can keep buying.

A change in any one of the three alters capacity to keep acquiring and capacity to defend price in contested areas, and none of the three is visible in a traffic band or an exchange-point list.

That is why the divergence documented above is not a clerical matter. If the registered holder, the brand and the controller sit in three places, then the entity that must answer for a routing change is not the entity that decides whether to make the change, and neither is necessarily the entity that bears the loss if the change is mishandled. Accountability is distributed, and distributed accountability is exactly what a counterparty needs to map before it signs.

The lookup that did not answer

The third retained surface is a failure. The LACNIC RDAP aut-num object for AS61832 could not be retrieved on 2026-10-17; the lookup was reported as not reachable (LACNIC RDAP aut-num object for AS61832).

A retrieval failure is evidence of one thing only: that the authoritative object was not verified on the retrieval date. It is not evidence that the registration is absent, defective, disputed, suspended or hidden. Treating an unreachable registry endpoint as a finding about the registrant would be the single most serious error available in this subject area, and it is a common one, because a blank result looks like a negative result to anyone reading quickly.

The consequence is a hard ceiling on this report. The only registry evidence available is a self-reported industry database entry, not the allocation record maintained by the registry. A self-reported record is decent evidence of how a network presents itself to the interconnection community, which is genuinely useful. It is not authoritative evidence of who holds the resource. That distinction is why the resolution offered below is framed as a verification requirement rather than as a conclusion about the holder.

It is worth stating plainly what would close the gap. A successful retrieval of the aut-num object would establish the registered holder, the country of registration, the administrative and technical contacts and the last-modified date, and it would either confirm or displace the long name carried on the industry record. Until that object is read, any statement about the identity of the resource holder rests on a secondary surface.

Six names, six roles, one date

Because this subject cannot be described with a single actor, the layers are set out here explicitly. GIGA MAIS FIBRA TELECOMUNICAÇÕES S.A. is the company named in the retained trade report as having announced the replacement of its diretor-presidente, and described there as controlled by EB Capital. DB3 SERVICOS DE TELECOMUNICACOES S.A. is the long name carried on the retained public network record for AS61832, and nothing in the retained set states whether it is a subsidiary, a predecessor or an unrelated registrant.

Alloha Fibra appears twice: as the organisation label on that network record, written as "Alloha Fibra - N/NE", and in the trade report as a company that EB Capital is described as also owning. EB Capital is the controller named in the trade report, described as controlling the company and as also owning Alloha Fibra. Lorival Nogueira Luz Júnior is the diretor-presidente whose replacement was announced by the board decision of 11 August 2025. Felipe Gonçalves Matsunaga is the interim chief executive named in the retained report, described there as an investor-relations director and an EB Capital partner.

The commercial brand under which the directory entry is filed — Giga+ Fibra, also given as Giga+ Empresas — is not the entity named as the AS61832 long name in the retained record.

That paragraph is deliberately flat. It uses only names, roles and one date that the retained package supports, because the temptation in a subject like this is to smooth six awkward names into one reassuring corporate story. The smoothing is where errors enter. A reader who collapses DB3 SERVICOS DE TELECOMUNICACOES S.A. into the operating company has made an assumption that no retained source supports, and every subsequent inference about who answers for the routes inherits that assumption.

What would change the conclusion

The conclusions here are falsifiable, and the falsifiers are specific.

First, a retrieved LACNIC aut-num object that names the operating company as holder of AS61832 would substantially collapse the three-layer finding, at least for this resource. The brand and the registrant would then coincide, and the interesting question would reduce to the ordinary one of group structure.

Second, a corporate filing, equity register or regulator record establishing the relationship between the two named companies would resolve the ambiguity that this report is forced to preserve. Such a document would also substitute a primary source for a trade citation on control, which is the weakest load-bearing element in the current picture.

Third, a dated disclosure confirming who holds the executive position after the interim period would settle the one direct contradiction between the directory entry and the retained evidence, in either direction.

Fourth, a published subscriber or market-share dataset from a regulator or from the company itself would displace the single-outlet snapshot and would allow the density and pricing analysis that this report can only frame.

Conversely, no combination of brand announcements, marketing material or directory updates would substitute for any of the four. The article's thesis survives on the evidence currently available; what it lacks is the primary documentation that would either confirm the layered structure as a governance fact or dissolve it as an artefact of registration inertia. The next condition that would materially change the answer is therefore documentary, not narrative.

The boundary of this report

This report is a reported-only treatment. No corporate filing, equity register, acquisition agreement, debt instrument or regulator dataset was obtained. The control relationship, the management change and the subscriber figure rest on a single trade citation, attributed as such throughout. The registry description rests on a self-reported industry record, attributed as such throughout. The one authoritative registry surface that was attempted did not respond, and that failure is dated to 2026-10-17 and confined to the act of retrieval.

Within those limits, several questions that a market participant would naturally ask cannot be answered here and are not guessed at. No purchase price, enterprise value or earn-out is stated. No debt, leverage multiple, covenant or refinancing schedule is stated. No revenue, margin, average revenue per user, churn or capital-spending figure is stated. No regulatory licence, municipal authorisation or regulator decision is stated. Nothing in this report establishes that any brand remains a separate legal person or that any brand has been merged, and nothing establishes how many municipalities the group actually serves.

The positive finding is narrow and durable. On the retained evidence, a reader can verify a chain of names running from a commercial brand through an autonomous system to a described controller, and can verify one dated board decision and one dated subscriber snapshot. A reader cannot verify the directory entry's named chief executive, and on that point the retained evidence points elsewhere. Everything else about the group's economics, its integration and its resilience remains to be documented.

For a counterparty, that is the practical instruction: read the registry object, read the equity description, and never let one stand in for the other.

The entry for this entity in the directory is giga-fibra-ms-pr.