Summary

  • FIBERNET’s current website advertises 50/50 and 150/150 residential plans in Huehuetenango, while recent government purchases and a June 2026 IPTV-app update support the conclusion that the operator is active.
  • AS269989 originated three IPv4 /24s and five overlapping IPv6 announcements in the latest public routing view; every visible IPv4 route reached it through UFINET, while only the IPv6 aggregate exposed Navega as a second logical neighbour.
  • No public record located for this review establishes fibre-route geometry, pole or duct rights, PON split ratios, backup-power runtime, spare-cable stock, crew coverage, restoration targets or physically independent upstream entrances.

A damaged highland span turns a speed promise into a distance problem

Picture the first workable hour after a wet, windy night in Guatemala’s western highlands. A branch has loaded an aerial cable until the span has sagged or parted. The optical alarm identifies a loss somewhere beyond a distribution point, but it does not put a technician under the correct pole. Before any splice can restore customers, someone must locate the break, reach it over a usable road, confirm that the pole is safe, obtain access, bring the right cable and closure, test both directions and decide whether traffic has another path while the work proceeds.

That sequence, rather than the number printed on a residential plan, is the practical meaning of resilience.

This is a risk scenario, not a claim that a documented FIBERNET span failed. No public incident notice found for this article ties a particular storm cut to the company. The terrain and seasonal exposure are nevertheless real operating conditions. Guatemala’s road authority publishes a national network map that shows Huehuetenango connected through a mixture of national, departmental and local roads, with many branches extending through mountainous country; it is a transport map, not a fibre map, and must not be read as one (Dirección General de Caminos road map). The distinction matters because a cable can follow a road, cross it, leave it, share an electricity alignment or take a completely different private route.

The 2026 rainy season supplies the immediate operating context. By 1 July, Guatemala’s disaster authority said it had handled 643 weather-related emergencies nationally, driven principally by floods, fallen trees, strong winds and landslides (CONRED’s 2026 emergency update). A separate 2025 assessment named Huehuetenango among the departments with the most reported road damage during that rainy period (CONRED’s road-impact report). INSIVUMEH also concluded its 2026 agroclimatic technical sessions in Huehuetenango after discussing the May-to-July outlook, evidence that rainfall planning is a live regional concern rather than an abstract national average (INSIVUMEH’s Huehuetenango session).

For a local ISP, those conditions turn geography into labour. A topology can be redundant in a router and still be fragile in the field if both links share the same pole line, bridge, roadside trench, substation or crew. Conversely, a modest network can recover well if it knows every closure, holds compatible spares locally, maintains safe access arrangements and can isolate a damaged segment without taking down the rest. FIBERNET’s public evidence is strong enough to show a functioning service and a routed network.

It is not strong enough to measure the distance from alarm to repair, which is why the article begins at the damaged span rather than at a bandwidth figure.

That distance has at least four components. There is diagnostic distance: how precisely monitoring and optical tests narrow the fault before a vehicle moves. There is travel distance: how long it takes a crew to reach the nearest safe access point under current road conditions. There is authority distance: whether the company can work immediately or must wait for a pole owner, electricity distributor, municipality, landholder or civil-protection team. Finally there is material distance: whether a compatible closure, fibre type, optic, power supply or replacement terminal is already in the west or must arrive from the capital.

A credible recovery claim has to survive all four. Public routing data can illuminate the first minutes of a logical outage, but it cannot show where a cable lies, who may touch it or which spare is on a shelf.

The operating brand predates the company record, and the evidence should stay separated

FIBERNET describes itself as a company created by a young entrepreneur from Huehuetenango that began operations in La Democracia in 2015 and now serves the departmental capital. The same page calls the business “100% Huehueteca” and says it delivers service using what it spells “XGPON” (FIBERNET’s company history). Those are first-party statements about the brand’s origin and technology. They are useful, but they do not establish the date on which the current sociedad anónima acquired every asset, customer agreement or permit associated with the earlier activity.

The legal trail starts later. A commercial-information page derived from Guatecompras reports a constitution date of 25 October 2019 and provisional Mercantile Registry registration on 12 November 2019 for NIT 106261770 (Guatecompras-derived supplier profile). A primary registry notice published in 2023 is firmer on the current business identity: it says José Daniel Alvarado Camposeco, acting as sole administrator and legal representative of FIBERNET, SOCIEDAD ANÓNIMA, requested that the registered commercial address be changed to Torre Visión, Oficina Fibernet, second level, Calle Real, zone 8, Huehuetenango. It also gives commercial registration number 854982 and references file 79932-2019 (Mercantile Registry notice).

Internet-number records connect that legal name to the routed operator. LACNIC’s RDAP entry for AS269989 names FIBERNET S.A as registrant, records the autonomous system’s registration on 7 May 2020 and identifies Alvarado Camposeco as the administrative, technical and abuse contact (LACNIC AS269989 record). FIBERNET also appears among Guatemala’s organisations in LACNIC’s 2024 electoral roll (LACNIC 2024 roll). The same named individual attended LACNIC 44/LACNOG 2025 in person under the FIBERNET S.A affiliation, which is a modest but recent sign of operator participation in the regional technical community (LACNIC 44 attendee list).

Other current signals reinforce continuity without proving scale. Verisign’s RDAP record shows fibernetgta.com registered in March 2022, changed in March 2026 and paid through March 2027 (domain RDAP record). Google Play identifies Fibernet, S.A. as developer of the FIBERTV app, gives the same Torre Visión address and shows an update on 3 June 2026 (FIBERTV listing). Together, these records support a careful conclusion: the present legal entity, website, app and AS belong to the same operating identity. They do not prove that every activity claimed since 2015 was conducted by the present company in its current form, so the brand history and the corporate history should not be collapsed into one date.

The verified footprint is a pair of named places, not a departmental route map

FIBERNET’s strongest geographic claim is narrow. Its history page says operations began in the municipality of La Democracia and that the company currently provides service in the departmental capital of Huehuetenango. Its public-facing contact details place the office in zone 8 of Huehuetenango, and the homepage repeats that address and a local support schedule. That is enough to describe a local Huehuetenango operator. It is not enough to draw a line between La Democracia and the capital, claim continuous coverage along that line, or say that the access network spans the entire department.

PeeringDB corroborates the corporate location. Its organisation page lists FIBERNET S.A, the fibernetgta.com website, the Torre Visión address and Huehuetenango, Guatemala, and associates the organisation with AS269989 (PeeringDB organisation record). The record was last updated in April 2024. Location in a peering directory is not a map of cabinets, splitters, fibre rings or wireless relays; it is an operator-supplied identity point. LACNIC’s address record likewise places the registrant in Huehuetenango, but the postal details do not disclose where routers, optical line terminals, aggregation sites or spare stores are housed.

The absence of a coverage map is consequential. A household address inside the city could be served by a short urban drop from a nearby splitter, by a longer aerial feeder entering from another zone, or by a radio-to-fibre arrangement. A customer in the municipality where the brand began could be on the same core, a separate access island, a wholesale partner’s plant or no longer within the marketed footprint. The website does not publish an address checker, neighbourhood list, route diagram, cabinet inventory or map of live service. It also does not say whether the 2015 La Democracia origin remains an active service area.

Public procurement broadens the list of places associated with FIBERNET, but not cleanly enough to redraw the network. The Guatecompras-derived profile groups purchases by the Huehuetenango municipal electricity company, CONRED’s Huehuetenango office, the municipality of Huehuetenango, Banco de Guatemala and the municipality of Concepción Tutuapa in neighbouring San Marcos. Some descriptions are clearly internet service; at least one large electricity-company purchase is described as equipment for its building rather than connectivity.

A supplier can deliver equipment outside its access footprint, and a contract can be fulfilled with a partner or leased circuit. These transactions therefore identify customer and commercial touchpoints, not owned routes.

That leaves a defensible map boundary. The office and current marketing point to Huehuetenango city. The company says the brand began in La Democracia. Government purchases show work for public bodies in and around the western region. No reviewed source plots a FIBERNET fibre path, tower, point of presence, intercity backhaul, alternate road crossing or service polygon. Any map that showed more than the named places would turn inference into false precision.

Even the phrase “service area” needs care. It can mean a municipality in which at least one address is connectable, a neighbourhood covered from a particular optical distribution point, a radio footprint, an area reached through a wholesale partner, or a sales territory where construction is offered after survey. Those meanings carry different installation costs and restoration obligations. A pin at Torre Visión establishes a business location, not the radius of a passive optical network. A historical reference to La Democracia establishes an origin story, not a live circuit.

A public buyer in Concepción Tutuapa establishes a transaction, not an owned interdepartmental cable. The safe geographic model is therefore a set of evidenced points with unknown links between them, rather than a coloured polygon that implies continuous reach.

A 150/150 plan is a retail product, not a capacity inventory

FIBERNET’s homepage presents two residential offers. The basic plan is labelled 50 down and 50 up; the advanced plan is labelled 150 down and 150 up and includes a Wi-Fi 6 router. The page writes the unit as “MB,” wording that should be quoted rather than silently converted into a technically different measure. It also says installation is free and no contract is required for the basic plan (FIBERNET homepage). These are commercially useful promises: the symmetry suggests an access design intended to support substantial upstream use, and a Wi-Fi 6 gateway may reduce an avoidable in-home bottleneck. Neither statement measures the network behind the gateway.

Retail line rate is only the outermost capacity boundary. To understand usable capacity, a customer would need to know whether the speed is a maximum or committed rate; the PON generation actually deployed on the relevant feeder; the split ratio; optical budget; number of active subscribers per port; uplink size from the optical line terminal; aggregation and transit headroom at busy hour; traffic-management policy; and the performance target during an upstream or equipment failure. None of those figures appears on the public pages reviewed. Prices, business-service terms, installation lead times and service-level remedies are also absent.

PeeringDB adds routing metadata, not engineering capacity. Its API classifies FIBERNET Guatemala as a Cable/DSL/ISP network, records an open peering policy, lists an IRR set and gives self-reported counts of four IPv4 and four IPv6 prefixes (PeeringDB network API). The latest routing observation differs: three IPv4 and five IPv6 announcements were visible. That mismatch can arise from stale profile fields, overlapping prefixes or a changed routing policy. It should not be translated into bandwidth, subscriber count or fault tolerance.

Cloudflare Radar places AS269989 in Guatemala and presents a mid-teens-thousands estimate for the network’s user population alongside routing information (Cloudflare routing view). The estimate is a measurement product, not FIBERNET’s audited subscriber total. “Users” may include multiple people behind one subscription, sampling error, shared access and changing address attribution. Even a perfect subscriber count would not reveal whether the 150/150 tier remains available at peak time, because congestion depends on demand, oversubscription and bottleneck placement rather than on customer count alone.

The evidence therefore supports a product statement and rejects a capacity extrapolation. FIBERNET markets symmetric access at two advertised tiers and supplies a modern customer router with the higher one. There is no public number for installed PON ports, lit feeder fibres, core throughput, purchased transit, traffic at peak, reserved restoration capacity or sold load. In the capacity record, the correct value for each of those fields is “not publicly disclosed,” not an estimate built from address space or marketing speed.

Capacity also changes with the failure being considered. A network may have enough normal-state transit for every sold plan yet fall to half that headroom when one upstream disappears. It may have diverse upstream contracts but only one powered aggregation room. A PON port may be lightly loaded while its feeder is exposed to one aerial cut, or a feeder may be physically protected while the customer-facing Wi-Fi obscures the quality of the fixed line.

Installed capacity is not necessarily lit; lit capacity is not necessarily powered through an outage; powered capacity is not necessarily operational after a control-plane failure; and operational capacity is not necessarily usable at the performance customers bought. None of those stages can be recovered from the “50/50” or “150/150” labels.

A buyer evaluating the advanced plan would therefore need two separate answers. The first is a normal-service answer: typical throughput and latency at busy hour, measured at the provider handoff rather than over an uncontrolled wireless room. The second is a degraded-service answer: what throughput remains after the loss of one feeder, router, power source or transit path, and for how long. An ISP can honestly sell a high line rate without promising either result. The transparency problem arises only when a retail number is treated as evidence for engineering reserves that have never been stated.

AS269989 is active, dual-stack and compact, but address space is not throughput

LACNIC allocated FIBERNET the IPv4 block 138.117.228.0/22 on 30 April 2020. That registration covers 1,024 addresses and names the same organisation and contacts as the autonomous-system record (LACNIC IPv4 allocation). On the same date, LACNIC allocated 2803:bee0::/32 in IPv6 space (LACNIC IPv6 allocation). The allocations prove control of number resources under the registry relationship. They do not identify customers, physical sites or transport equipment.

RIPEstat’s AS overview showed AS269989 announced on 16 July 2026 and identified the holder as FIBERNET S.A (RIPEstat AS overview). Its announced-prefix view found three IPv4 /24s—138.117.228.0/24, 138.117.230.0/24 and 138.117.231.0/24—and five IPv6 announcements: the covering /32, a /36 and three more-specific routes of /40, /41 and /42 (RIPEstat announced prefixes). Because the IPv6 prefixes overlap, adding their nominal address counts would double-count space and produce a meaningless capacity number.

The routing-status endpoint counted 768 announced IPv4 addresses, five IPv6 prefixes and two observed neighbouring autonomous systems. It also dated the first visible FIBERNET route to 11 May 2020 and showed near-complete visibility among the RIPE RIS peers in that observation (RIPEstat routing status). This is credible evidence of a globally reachable network, not evidence of the number of access lines. A single public IPv4 address can sit on a router, server, business circuit or carrier-grade NAT boundary; an IPv6 /32 can support an enormous hierarchy without saying how many customers are active.

Route-origin security is a positive operational signal. RIPEstat’s Routinator-backed check marked 138.117.228.0/24 valid for origin AS269989 (IPv4 RPKI validation). It also marked the covering 2803:bee0::/32 valid for the same origin, with a maximum length that permits more-specific customer or engineering announcements down to /48 (IPv6 RPKI validation). Hurricane Electric’s public BGP view independently showed the /32 originated by AS269989 with valid IRR and ROA status (Hurricane Electric prefix view).

RPKI answers one important question: whether the visible origin is authorised by a route-origin attestation. It does not authenticate the entire AS path, prove that two routes use separate fibre, or guarantee that the authorised origin has spare capacity. FIBERNET’s number-resource practice looks more mature than a provider relying only on borrowed addresses. The harder resilience questions begin one layer below the registry and one layer beyond the origin announcement.

The routing observations are also time-bounded. A collector view on 16 July describes what selected observation points could see then; it does not guarantee that every route was continuously present before or after the sample. Prefix counts can change during maintenance, traffic engineering or incident response. More-specific IPv6 announcements may be deliberate policy divisions rather than evidence of five separately resilient networks. Conversely, an unannounced backup arrangement may remain invisible until activated.

The useful finding is not that the public table is a complete network inventory, but that the independently observed routes align with the registered FIBERNET origin and expose a compact, intelligible control-plane footprint.

One quarter of the registered /22 is authorised to originate somewhere else

The most unusual routing fact sits inside the IPv4 allocation. FIBERNET’s LACNIC registration covers 138.117.228.0/22, which divides neatly into four /24s. Three are originated by AS269989. The remaining quarter, 138.117.229.0/24, was visible from AS28461, GRUPO SURNET SA DE CV, rather than from FIBERNET.

RIPEstat’s prefix-consistency view makes the split explicit. It lists FIBERNET’s .228, .230 and .231 /24s in BGP, while .229 appears with AS28461. It also shows older or broader routing objects for the /22 and a .230/23 that were not themselves visible as covering BGP announcements at the observation time (RIPEstat /22 consistency view). A separate prefix overview identifies AS28461 as the current origin of 138.117.229.0/24 (RIPEstat .229 overview).

The visible route should not be described as an unauthorised hijack. RIPEstat’s RPKI check reports a valid ROA for AS28461 to originate exactly 138.117.229.0/24 (RPKI validation for the Surnet origin). That means the origin is cryptographically authorised under the resource-certification system. It does not reveal the commercial arrangement that produced the authorisation.

Several explanations are possible in principle: FIBERNET may have assigned or leased the /24, may use Surnet for a service, may have transferred operational control while registry information remains at the parent block, or may be participating in another documented arrangement. The reviewed public records do not settle which explanation is correct. They also do not show whether traffic in .229 serves Guatemala, Mexico, infrastructure, customers or an entirely separate purpose. Geolocation databases are too indirect to resolve the contractual boundary.

For resilience analysis, the split matters in two ways. First, it prevents the simple statement that FIBERNET actively routes all 1,024 IPv4 addresses allocated under its name; AS269989 visibly originated 768. Second, it demonstrates why registry ownership, route origin and service operation must be treated as separate layers. A valid external origin might be a sensible commercial decision, but it is not automatically a backup route for FIBERNET customers. Without a documented failover design, the Surnet-originated /24 cannot be counted as spare access capacity, a secondary upstream, a recovery network or proof of geographic diversity.

IPv4 converges on UFINET; IPv6 reveals a second logical neighbour

The latest public path evidence is sharper than the physical evidence. RIPEstat’s neighbour view found two autonomous systems immediately to the left of AS269989 in observed paths: AS52468 and AS26617 (RIPEstat neighbour view). The protocol split is the important part. UFINET’s AS52468 was visible beside FIBERNET for IPv4 and IPv6. Navega’s AS26617 appeared for IPv6, not for the observed IPv4 routes.

The full BGP-state response shows how that distinction materialised. Each of FIBERNET’s three IPv4 /24s ended with AS52468 immediately before AS269989. The IPv6 /36 and the three IPv6 more-specifics did the same. The covering IPv6 /32, however, ended with AS26617 before AS269989 in the sampled paths (RIPEstat BGP state). RIPEstat identifies AS52468 as UFINET PANAMA S.A. (UFINET AS overview) and AS26617 as Navega.com S.A. (Navega AS overview).

An independent IPv4 summary from CIDR Report reached the same topological result: one upstream adjacency, AS52468, for AS269989’s three originated IPv4 prefixes (CIDR Report for AS269989). That does not prove FIBERNET buys only one commercial service. BGP collectors see advertised paths, not dark fibre, inactive backups, default-only sessions, private interconnections or contractual options that are not exporting these routes. It does establish that the public IPv4 control plane did not display an independently visible second adjacent AS at the time examined.

PeeringDB supplies a complementary absence. Its network-exchange API returned no public exchange connection for network 29872 (PeeringDB exchange-connection query), and its facility API returned no listed facility connection (PeeringDB facility query). Empty PeeringDB results do not prove that FIBERNET is absent from every exchange or carrier site; participation can be private, indirect or simply unrecorded. They mean a buyer cannot use that public profile to verify a second interconnection location.

The resulting resilience statement must be protocol-specific and modest. FIBERNET has two visible logical neighbours across its routing set, which is better than universal dependence on one AS. But all three public IPv4 routes converged on UFINET, and most visible IPv6 routes did too. Even the Navega-served /32 does not prove physical independence: both carriers could enter the same building, use the same intercity corridor, share a wholesale fibre owner, depend on the same local power or reconverge before leaving Huehuetenango. Logical diversity is a necessary clue, not the requested proof of route diversity.

The asymmetry between IPv4 and IPv6 is commercially relevant because many customer sessions, public websites and legacy applications still depend on IPv4 or traverse translation systems. A second visible IPv6 neighbour does not automatically rescue an IPv4-only service, and a customer router capable of both protocols does not guarantee that an application will fail over cleanly. Nor does the existence of two neighbouring AS numbers prove simultaneous usable capacity. One may be primary and the other selective, one may carry only the covering IPv6 route, or policy may direct different prefixes over different contracts.

The public view supports “two logical neighbours observed,” not “dual-carrier service for every customer.”

Physical validation would require a different evidence package: carrier letters or diagrams naming handoff sites; photographs or engineering records showing separate building entrances; route descriptions sufficient to identify shared bridges, poles and corridors; power-domain information for each termination; and a controlled failover test recording convergence time and degraded throughput. Commercial confidentiality can limit the precision published openly, but it need not prevent a customer under agreement from receiving proof.

In the absence of that package, the single visible IPv4 adjacency remains the most decision-relevant fact for an organisation whose continuity depends on IPv4 reachability.

Route diversity cannot be inferred from XGPON, a road map or two carrier names

FIBERNET’s use of “XGPON” is a statement about access technology as the company describes it. A passive optical network can eliminate active electronics between an optical line terminal and customer premises, but it does not eliminate common points of failure. Feeder fibres can share one cable. Multiple splitters can sit behind one cabinet. Separate customer drops can join the same pole line. Two uplinks can terminate on one aggregation router. A ring can exist in a diagram yet run through the same bridge or roadside cut in both directions.

The physical proof needed here is concrete. For the urban network, it would include feeder and distribution route classes, whether fibres are aerial or buried, the locations and power domains of optical line terminals, the number and placement of aggregation sites, and the first point at which nominally diverse paths share a structure. For any continuing La Democracia footprint, it would include the intercity handoff, mountain or road crossings, relay sites if radio is used, and the restoration authority on plant that FIBERNET does not own. None of those details is published.

Power is equally bounded. Guatemala’s electricity regulator lists the Empresa Eléctrica Municipal de Huehuetenango as the local distribution company and publishes its Huehuetenango contact point (CNEE distributor record). CNEE also approved an expansion of the Huehuetenango substation through a new 13.8 kV line exit in 2024 (CNEE-48-2024). Those records describe the surrounding electricity system. They do not show which FIBERNET sites use that distributor, which feeder supplies them, whether two network sites share the same outage domain, or whether the network has batteries and generators.

Customer premises create another boundary. The advanced plan’s Wi-Fi 6 router still depends on household power. A provider can keep an optical line terminal and aggregation path alive while customers remain offline because their optical network terminals and routers have no backup. Conversely, a customer UPS cannot help when the serving cabinet, upstream handoff or aerial feeder has failed. Meaningful outage communication therefore needs to distinguish network-side loss, local-power loss, damaged drop, degraded capacity and customer-equipment failure.

Nothing in the public evidence shows pole ownership or attachment rights. The company may own some structures, lease space from an electricity or telephone operator, use municipal rights of way, bury portions of the network, or combine these methods. Each arrangement changes repair authority. A crew that can splice cable may still have to wait for a pole to be declared safe, a road to reopen, an electricity team to isolate conductors or another owner to grant access.

Until those dependencies are disclosed or tested, the recoverable footprint is smaller than the marketed footprint: it is the area that can actually be reached, powered, isolated and repaired under adverse conditions.

Local support is valuable only when the crew, spares and escalation path are real

FIBERNET publishes a local office, telephone number, email address and opening hours, with weekday service and a shorter Saturday window. That is more useful than an anonymous national sales form because customers can identify an accountable local operator. The recent technical-community participation associated with the company also suggests that network stewardship is not wholly detached from the region. Yet a contact point is not the same thing as a restoration organisation.

The decisive labour questions are absent. The public pages do not state how many technicians are on call, whether support continues outside office hours, where crews are based, how far the farthest active customer lies from a depot, or whether the company maintains separate teams for access fibre, core routing, power and customer installations. There is no published mean time to repair, severity matrix, incident channel, maintenance calendar, after-hours escalation number or service-credit policy.

Spare depth is just as important as headcount. A resilient small provider needs compatible drop cable, feeder cable, closures, splitters, connectors, optics, customer terminals, routers, power supplies, batteries and test instruments close enough to the service area. Some items are interchangeable; others must match the installed platform. A reel in Guatemala City is not the same recovery asset as a reel in Huehuetenango if roads are disrupted. One optical time-domain reflectometer shared across distant crews can turn fault localisation into a queue. None of these operational facts can be derived from the retail plan or ASN.

The economics are unforgiving. Local ISPs compete partly by keeping prices accessible, installations fast and support personal. Redundant transit, diverse fibre entries, stocked spares, trained standby crews and battery replacement programmes all impose costs before a failure occurs. If those costs are not visible in tariff design, enterprise options or service commitments, customers cannot tell whether resilience is funded, cross-subsidised or simply assumed. The absence of a posted price makes even that comparison impossible from the website alone.

A credible support promise would bind labour to geography. It would define normal and emergency response windows by zone; explain whether crews can repair aerial and buried plant; identify dependencies on pole owners and civil works; state how incidents outside office hours are handled; and separate a home visit from a network restoration. FIBERNET may already perform some or all of this work. The point is not that the capability is absent, but that the public evidence does not let a household, business or public office verify it before service fails.

Scale does not determine whether that model works. A small local team may reach a familiar fault faster than a national operator dispatching from another department, especially when technicians know the roads, pole lines and landowners. The same concentration can become a weakness when two incidents compete for one splicer, one test set or one vehicle, or when the person authorised to change routing is also coordinating the field repair.

The relevant measure is not employee count in isolation but simultaneous incident capacity: how many faults can be diagnosed, accessed, made safe, repaired and communicated at once without abandoning routine support.

Recurring public purchases prove service, while also raising the cost of ambiguity

Current procurement records offer some of the clearest evidence that FIBERNET remains commercially active. Banco de Guatemala’s February 2026 low-value purchase page records Q230 for internet service from FIBERNET, SOCIEDAD ANÓNIMA (February 2026 purchase). The March record contains further Q230 internet-service entries (March 2026 purchases), and the April record shows another Q230 internet-service purchase on 29 April (April 2026 purchase). Repeated payments over consecutive months are stronger operating evidence than an undated marketing page.

The broader procurement aggregation reports dozens of transactions and identifies recurring buyers including the municipal electricity company, CONRED, the municipality of Huehuetenango and Banco de Guatemala. The descriptions include internet service to CONRED’s departmental or regional Huehuetenango office, alongside equipment and other purchases. Because the aggregation is derived from government data and can change as records are updated, it should be used for pattern recognition, not as an audited revenue statement. It nevertheless establishes that FIBERNET’s customer surface is not purely residential.

That distinction matters during failure. A residential outage interrupts work, education, communication, entertainment and household administration. A public-office outage can affect coordination and service delivery, even when the link is not formally designated critical. The evidence does not show whether any listed institution buys a dedicated circuit, a residential-style plan, a backup line or a managed service. It would be irresponsible to claim that FIBERNET is the sole connection for emergency response, banking operations or municipal electricity. The records prove supply, not exclusivity or criticality.

FIBERTV adds another dependency surface. The app invites users to obtain credentials from their provider and supports mobile viewing and Chromecast. Its June 2026 update shows an actively maintained service adjunct, but it does not state where video is hosted, how much traffic is local, whether multicast is used, or what happens when transit is constrained. An IPTV product can increase evening demand precisely when residential internet usage is already high. Without peak-traffic and cache information, the app is evidence of a service offering, not proof that the network has spare headroom.

Public and residential customers therefore face the same transparency gap at different stakes. They can verify that the company exists, sells internet, runs an AS and continues to receive payments. They cannot verify whether their circuit shares a feeder with the next district, whether a second contract exits through the same UFINET path, whether backup power covers the full outage window, or which customers receive restoration priority. Those are procurement questions, not merely engineering curiosities.

For an institutional buyer, the practical remedy is to specify evidence instead of accepting labels. A tender can require the legal service provider and the physical last-mile operator to be named separately; request the first shared structure between primary and backup links; define whether diversity applies to conduit, pole line, building entrance, aggregation router and upstream carrier; and require a witnessed failure test. It can ask for battery autonomy at each provider-controlled active site, the location class of critical spares, after-hours escalation roles and an incident-report timetable.

None of those requests presumes that FIBERNET lacks capability. They convert an otherwise ambiguous “internet service” purchase into a recoverability requirement that can be checked before the rainy season.

The useful test is whether the marketed footprint can be restored, not merely reached

FIBERNET’s public evidence has improved beyond the “thin footprint” suggested by a name in a registry. The operator has a coherent legal and technical identity, current retail offers, recent public-sector payments, an updated customer app, LACNIC resources, valid route-origin authorisations and globally visible dual-stack routing. It should be described as an operating regional ISP, not as an unverified directory entry or a generic technology company.

The resilience evidence is weaker. A buyer can see one adjacent AS on all three IPv4 routes and a second logical neighbour on the covering IPv6 announcement. The buyer cannot see where either carrier enters the local network, whether their paths are geographically separated, where they reconverge, or whether a failure shifts traffic automatically. The authorised Surnet origin for one quarter of the registered IPv4 block is a real and explainable routing arrangement, but the public record does not explain it and it cannot be counted as recovery capacity.

The next proof should be operational rather than promotional. FIBERNET could publish a service-area map at neighbourhood or municipality resolution without exposing sensitive cable detail; distinguish owned plant from leased or partner infrastructure; describe access media by zone; state whether core sites have battery and generator coverage; disclose the number of physically separate upstream entrances; and provide a restoration target for access cuts, power failures and upstream loss. Enterprise customers could receive route-diversity letters showing the first shared structure and carrier handoff, subject to confidentiality.

Capacity disclosure need not reveal commercial secrets. The company could state whether 50/50 and 150/150 are maximum or committed rates, publish typical busy-hour performance, identify any fair-use or traffic-management policy, and say whether backup operation preserves full or reduced throughput. A simple status page with incident start time, affected area, failure class and next update would make repair labour visible without naming staff or exposing network security.

Most important, resilience should be tested against the actual highland sequence. Can monitoring distinguish an upstream loss from an access cut? Is there a crew available after hours? Are compatible closures and cable stored locally? Can the crew reach both ends of a damaged span if a road is blocked? Does a second path avoid the same poles, bridge, substation and aggregation room? How long will customer equipment remain powered? Which public-service or business links are prioritised, and on what contractual basis?

Until those questions have evidence-backed answers, the correct conclusion is balanced. FIBERNET can reach the global internet, serve customers and operate modern address space. It has not publicly demonstrated that the network advertised across Huehuetenango can survive and recover from the same physical events that make local service valuable. The difference between those two propositions is the difference between a marketed footprint and a recoverable one.