Summary
- APNIC records AS136567 as an active registration held by CyberWorld Data Center Co.,Ltd. in Thailand, but RIPE's route collectors show no current or historical prefixes, no observed neighbours and no visible BGP routes for the ASN.
- The same registrant's CyberWorld-BKK2 network provides an important control: it is currently visible with IPv4 and IPv6 space. The contrast shows that the company can operate a routed network, while supplying no equivalent operating evidence for AMS1.
- Cyber World Tower in Bangkok contains a well-documented CS Loxinfo facility with substantial power, cooling and carrier infrastructure. Public facility records identify CS Loxinfo as its operator, not CyberWorld Data Center Co.,Ltd., and do not list AS136567 among the networks present.
- Until the company identifies the AMS1 site, facility operator, contracted power, cooling design, fibre entrances, carriers, prefixes and tested recovery arrangements, the defensible description is a reserved network identity with unproven data-centre operation.
An active registration is not an active data centre
The strongest fact about CyberWorld-AMS1 is precise but narrow. APNIC's RDAP record for AS136567 names the autonomous system CyberWorld-AMS1, identifies CyberWorld Data Center Co.,Ltd. as the registrant, gives Thailand as the country, and marks the registration status active. It says the ASN was registered on 15 May 2017 and last changed on 22 June 2020. The organisation record points to 12/32 Phaholyothin 48 Road in Bangkok, while the network contact record uses 90 CyberWorld Tower, Ratchadaphisek Road. Those entries establish responsibility for a number resource. They do not establish a data hall, a rack, a server, a utility feed or a customer service.
That distinction is unusually important here because every independent routing view is silent. RIPEstat's AS overview describes AS136567 as unannounced. Its announced-prefix view returns an empty list. The routing-status view reports zero IPv4 and IPv6 prefixes, zero address space and zero observed neighbours. The current BGP-state view contains no routes. Most tellingly, RIPEstat's routing history contains no origin history at its stated visibility threshold, and its neighbour view contains no adjacent networks.
There is no contradiction between APNIC's active status and RIPEstat's unannounced status. The first means the registration remains assigned in the regional registry. The second means route collectors do not see the ASN originating reachability on the public Internet. An organisation may hold an ASN before launch, retain one after a service closes, use one only in a private arrangement, or reserve it for a plan that changes. None of those possibilities can be selected from the registration alone. The absence of routes also cannot prove that no equipment exists. It can prove something commercially significant, however: there is no public BGP evidence that an Internet-facing AMS1 network is operating now.
For a company whose name contains "Data Center", that is the beginning of inquiry, not its end. A data-centre claim has two coupled parts. The physical part requires controlled space, power, cooling, fire protection, security and staff. The network part requires reachable address space, carrier connections and an operating policy. AS136567 presently demonstrates neither part. Its name is evidence of administrative intent; the missing operational signals keep that intent from becoming a service claim.
The company exists, and BKK2 provides a live control
The negative finding for AMS1 should not be stretched into a claim that the Thai company is fictitious or wholly inactive. A Thai corporate-data service drawing on government open data lists CyberWorld Data Center Co.,Ltd. under registration number 0105552000659, says it was incorporated on 6 January 2009, and describes its filed activity as wired and wireless Internet communication services. It reports the legal status as still operating and gives the same 12/32 Phaholyothin 48 address found in APNIC's organisation entry. This is secondary publication of corporate information rather than a facility certificate, but the identity alignment is strong.
APNIC also exposes a useful sign of current administrative attention. In the AS136567 record, the abuse mailbox was validated in March 2026, although the other network contact address was marked invalid. A validated abuse contact is not proof of service, capacity or staffing. It does indicate that somebody associated with the resource answered one registry-maintenance channel recently. That is a modest positive signal, and it should remain modest.
The better control is another ASN. APNIC records AS136566 under the same organisation and calls it CyberWorld-BKK2. Unlike AMS1, RIPEstat says BKK2 is announced. Its prefix view shows 103.20.120.0/24 plus two IPv6 /48s, while its routing status shows the IPv4 prefix visible to all 327 sampled IPv4 peers and the IPv6 routes visible to 319 of 321 sampled IPv6 peers on 10 July 2026. It also reports two observed neighbours.
This sibling comparison removes an easy excuse. If every CyberWorld resource were missing from public routing, stale records or a vanished operator would be the leading explanation. Instead, BKK2 displays the ordinary marks of a small live network while AMS1 does not. The company appears able to originate routes under one identity. It is therefore reasonable to treat the silence around AS136567 as specific to that identity rather than as a blind spot covering the whole company.
The comparison also disciplines what can be inferred. BKK2's routes do not activate AMS1. They do not prove shared equipment, common carriers, common customers or failover between the two ASNs. A routed Bangkok-labelled network cannot be counted as an Amsterdam-labelled node, and a Thailand registration does not disclose where a router is installed. What BKK2 supplies is a standard of proof: prefixes, visibility and neighbours can appear when a CyberWorld network is in public operation. AMS1 does not currently meet that standard.
AMS1 is a label, not a location record
The string AMS1 invites readers to interpret it as a first Amsterdam site. That interpretation may be the registrant's intention, because AMS is familiar shorthand for Amsterdam in infrastructure naming. But the public evidence does not locate a CyberWorld facility, rack or router there. APNIC assigns the resource to a Thai organisation and supplies Thai addresses. The registration contains no Dutch street address, facility identifier, exchange membership, carrier name, latitude, rack reference or commissioning date.
Nor does a neutral interconnection record fill the gap. A direct query to PeeringDB for ASN 136567 returns no network entity. That means there is no PeeringDB page declaring exchange points, facilities, traffic level or peering policy for AMS1. PeeringDB is voluntary, so absence is not proof that a private cross-connect does not exist. It does remove a common way for an interconnection network to substantiate location and availability.
The same restraint applies to Cloudflare Radar's AS136567 page. Radar recognises the registered name and country, but the existence of an overview page is not evidence of material traffic. IPinfo classifies AS136567 as inactive and reports no known IPv4 or IPv6 space. That is a secondary interpretation, but it agrees with the stronger route-collector evidence.
Calling the node Amsterdam without a disclosed site would convert a naming convention into geography. Calling it a data centre without a disclosed operating boundary would convert geography into capacity. The supportable wording is narrower: CyberWorld Data Center Co.,Ltd. holds a Thai-registered ASN named AMS1, and public sources do not presently reveal where or whether that ASN has physical deployment.
Cyber World Tower is evidence of a different operator
The address at 90 CyberWorld Tower creates the greatest risk of mistaken identity. Bangkok's Cyber World Tower does contain substantial data-centre and interconnection infrastructure. It is just not publicly attributed to the company examined here.
PeeringDB's facility record calls the site "CS LoxInfo Data Center - CW Tower", identifies CS Loxinfo Public Company Limited as the organisation, and places it at Cyber World Tower A, 90 Ratchadaphisek Road in Huai Khwang. The record lists two exchanges and 20 networks. Those include BKNIX, CS Loxinfo's own networks, Advanced Wireless Network, DTAC, Jastel, Symphony and UIH. It does not list AS136567 or CyberWorld Data Center Co.,Ltd.
The facility is not a vague directory entry. BKNIX's location page identifies CSL CW at 90 CW Tower A, floors 17 to 20, with CS Loxinfo contact details. BKNIX's explanation of its design says its second Bangkok site was placed in Cyberworld Tower and that operating in two locations provided redundancy and broader member access. BKNIX's current description says its exchange is located in facilities with N+1 support infrastructure and offers a choice of carriers. Together, these sources establish a real interconnection environment at the building.
CS Loxinfo's own material goes further. Its IDC@Exchange brochure locates the Cyberworld data centre on the fourth floor of Tower A. It describes two 2,000 kVA transformers in active-standby configuration, two sets of 400 kVA UPS with N+1 unit redundancy, 2N AC distribution, a 2,500 kVA diesel generator, a 48-hour fuel tank, N+1 precision cooling, six active and one standby cooling unit per room, FM200 suppression, two-hour fire-rated walls and water-leak sensors. A commercial facility directory attributes 1,500 square metres and 385 racks to CSL at the same address, while warning that its location display is approximate.
These details show what a data-centre claim looks like when it is backed by physical specifications. They also show why those specifications cannot be transferred to CyberWorld-AMS1. The brochure names CS Loxinfo throughout. CS Loxinfo's colocation terms define CS Loxinfo Public Company Limited as the service company. A Thai securities filing about CSL's business likewise describes CSL as the provider of Bangkok data-centre services, discusses rack sales and expansion, and identifies electricity as the main cost of that service. None assigns the plant to CyberWorld Data Center Co.,Ltd.
Several explanations remain possible. CyberWorld Data Center Co.,Ltd. may once have been or may still be a customer, tenant, reseller or network user at the building. Its network contacts may simply have worked there. The organisation may have moved while older role records retained the operational address. It might use services without advertising AS136567. Public evidence does not choose among these cases. What it does establish is the ownership boundary: the known Cyber World Tower data centre belongs in CS Loxinfo's evidence column, not in CyberWorld-AMS1's, unless a contract, operator statement or live interconnection record links them.
That boundary matters because name similarity can manufacture scale. "CyberWorld Data Center" and "Cyberworld Datacenter" look almost interchangeable in a search result. Legally and operationally they are not interchangeable. One is the registered name of the AS136567 holder. The other is the product and location name used by CS Loxinfo for a Bangkok facility. Sound infrastructure research keeps the rack owner, facility operator, network operator and building landlord separate until evidence joins them.
Power evidence begins at the utility handoff
A data centre is a power-delivery system with computing attached. For AMS1, the missing power facts are therefore not supplementary details; they are the missing asset.
Thailand's own investment conditions provide a useful benchmark. The Board of Investment's digital-infrastructure conditions require a qualifying data-centre project to have a continuous-rated generator capable of carrying the full load, a backup generator for a unit failure, UPS and cooling backup that operate immediately when the main system fails, and independent electricity-distribution paths. The same conditions require concurrent maintainability, efficient backed-up air conditioning, fire prevention and 24-hour security. These rules do not certify CyberWorld-AMS1; there is no public evidence that it sought or received those incentives. They demonstrate the level at which Thailand itself distinguishes a project from a resilient operating facility.
For AMS1, no public source states the site utility, contracted megavolt-amperes, IT load, transformer arrangement, UPS topology, generator rating, fuel runtime or refuelling agreement. There is no single-line electrical diagram, no commissioning statement and no load-test result. Without those items, even a photographed rack would not reveal whether it can survive a feeder fault or maintenance event.
Dual utility feeds are especially easy to overstate. Two cables entering one room may return to the same substation, duct bank or upstream breaker. Two transformers may still depend on one utility service. A and B rack feeds may converge at a common switchboard. Genuine diversity must be followed from the utility source through transformers, switchgear, UPS modules, distribution boards and rack power units. It must also be tested while carrying a representative load. No such chain is public for AMS1.
Bangkok's utility is not assumed to be failure-free. The Metropolitan Electricity Authority publishes planned outage notices because maintenance can require temporary supply interruption, and its outage-management description explains how distribution failures are detected and restored. The point is not that a particular outage affected the subject; none has been identified. The point is that a credible data-centre service must bridge ordinary utility events without customer interruption and must show how it does so.
Installed capacity is not usable capacity
Infrastructure numbers become misleading when their denominators are hidden. A site may advertise gross floor area while only part is fitted for racks. It may quote transformer nameplate capacity while utility contracts, cooling plant, generator capacity or electrical losses constrain actual IT load. It may count rack positions that lack saleable power density. It may reserve capacity for internal systems or have capacity physically installed but not commissioned.
No capacity number at all is public for AMS1. That means there is no basis for estimating its scale from the company's name, from AS136567, or from the CS Loxinfo plant at Cyber World Tower. Even the 385-rack figure reported for the separate CSL facility should be treated as a facility-directory claim, not as CyberWorld capacity and not as proof of occupied racks.
The distinction is economically material. In a data hall, the scarce unit is often not square metres or rack shells but deliverable kilowatts under the promised resilience class. If a room can physically fit 100 racks but its electrical and cooling systems support only 300 kW, then average density constrains what can be sold. If the generator can carry only critical plant and not the full IT load, nominal utility capacity overstates outage capacity. If one UPS module is held as the N+1 reserve, usable capacity must preserve that reserve instead of selling it.
Thailand's policy direction makes this constraint more visible. In May 2025, the Board of Investment linked top incentives to power-use effectiveness and resource management. In July 2026, the Board said power readiness, water, environmental effects and clean-energy transition would be considered together when examining data-centre projects. Those measures concern large promoted investments, not proof of this small operator's status. They reinforce the central point: a label and an allocated ASN reveal nothing about energised, supportable IT load.
A credible AMS1 capacity statement would therefore separate gross building power, contracted utility capacity, installed critical power, commissioned IT power, occupied IT power and saleable reserve. It would give the rack-density range and the redundancy basis under which capacity is calculated. Until those figures exist, "capacity" is not merely unknown; it is not a defensible attribute of the subject.
Cooling turns electrical resilience into service resilience
Power is useful only if heat can be removed. Servers can remain electrically alive while inlet temperatures climb, components throttle and equipment eventually shuts down. Cooling failure is therefore not a secondary comfort issue. It is a direct route from plant fault to lost service.
The separate CS Loxinfo brochure provides a concrete comparison. It describes underfloor cold-air delivery, N+1 precision cooling, automatic rotation and a standby unit for each room. Whether those historical specifications remain unchanged would need current confirmation, but the document at least identifies a topology and an operator. AMS1 has no equivalent disclosure. There is no stated cooling medium, chiller or direct-expansion arrangement, component redundancy, containment design, temperature envelope, alarm threshold or backup-power coverage.
The coupling points deserve particular attention. A generator may support IT racks but omit part of the cooling plant. An N+1 computer-room unit arrangement may still share pumps, controls or heat rejection. Both utility feeds may supply a common cooling switchboard. A facility may meet its design on a mild day but lose reserve during Bangkok's hottest, most humid conditions. If AMS1 is outside Thailand, an entirely different climate and plant design may apply; the absence of a site means even the environmental loading case is unknown.
Operating evidence would include recent integrated systems testing: remove a power source, fail a cooling unit, transfer to generator, and observe rack inlet temperatures and service continuity. Maintenance records would show whether filters, refrigerant circuits, pumps, controls and sensors are serviced without reducing resilience below the promised level. Customer-facing service terms would state what temperature excursion or cooling interruption counts as a breach. None of this is public for AMS1.
That missing evidence changes the language a buyer should accept. "N+1 cooling" is not enough without the system boundary. N+1 at the room unit can coexist with a single upstream dependency. The question is whether heat removal remains available through every planned maintenance state and credible single failure, at the contracted load. For CyberWorld-AMS1, there is not yet enough information to ask that question against a real diagram.
The carrier test is stricter than an address
Carrier diversity is another area where appearances mislead. A building with many networks can be highly connected while one tenant buys only one cross-connect. Two carriers can enter through the same street duct. Two logical services can ride the same metro fibre. An Internet exchange presence can improve local reach but does not replace full transit, and a transit contract does not create physical route diversity.
Cyber World Tower clearly has a rich connectivity environment. Its PeeringDB record lists 20 networks, and BKNIX's PeeringDB page shows the exchange operating at CSL CW as one of several Bangkok locations. The APNIC account of BKNIX explains why two sites were used for resilience. Those are strong facts about the building and the exchange. They do not show that AS136567 has a port, cross-connect or session there.
The route data points the other way. With no announced prefix, AS136567 cannot currently be observed as an origin. With no observed neighbour, there is no public AS path from which to infer an upstream. With no PeeringDB entity, it declares no exchange port or private interconnection location. There is also no public routing policy naming imported or exported routes. A buyer cannot identify a first carrier, let alone verify that a second carrier follows a physically independent path.
The contrast with BKK2 is again instructive. AS136566's routes show two observed neighbours and global visibility. That does not reveal every commercial agreement or conduit, but it proves that packets can encounter the ASN on the public Internet. AMS1 supplies no comparable packet-level trace. Assigning BKK2's connectivity to AMS1 would erase the reason separate ASNs exist: each autonomous system can apply its own routing policy and originate its own space.
A defensible carrier claim for AMS1 would identify the facility and meet-me room, the cross-connect provider, at least two upstreams, the exchange ports if any, the prefixes originated, the maximum-prefix policy, RPKI route-origin authorisations, and the physical entrance routes. Looking glasses or route-collector views should corroborate the live sessions. Evidence should also show that a carrier maintenance event or fibre cut actually moves traffic to the alternate path without an unacceptable outage.
This is why an address in an APNIC contact object is not carrier evidence. It may be where an administrator once sat. Even a rack address is not enough. The relevant asset is an end-to-end path from server interface to diverse external networks, with common ducts and intermediate providers exposed. CyberWorld-AMS1 has not made that path visible.
Routing silence changes the failure analysis
An operating network leaves traces. Prefixes appear at route collectors. AS paths expose neighbours. Exchange directories may show ports. Route-origin authorisations may bind prefixes to the ASN. Latency observations can suggest geography, although they rarely prove a room or building. Customers may publish traceroutes or service addresses. None of the strong traces exists for AS136567.
Silence creates two very different failure interpretations. The first is that AMS1 is not in public operation. In that case, zero routes are the steady state and there is no live customer traffic to fail. The second is that the network exists privately or as cold reserve. In that case, route collectors cannot assess it, and private operating records become essential. A third possibility is that a once-planned node never launched. RIPEstat's empty history makes that more plausible than a visible network that recently withdrew, although low-visibility or private use remains possible.
This matters for outage monitoring. A fall from normal traffic to zero can be an incident signal. A network that is always invisible provides no public baseline. Cloudflare's recognition of the ASN name does not solve that problem, and a registry contact validation does not create telemetry. Without an announced service endpoint, an outside observer cannot distinguish planned maintenance, equipment failure, commercial dormancy and non-deployment.
It also matters for recovery claims. A company might say AMS1 is a disaster-recovery node for BKK2. The naming pattern could tempt that theory, but no public source says so. If it were a cold site, proof would include replicated data, reserved power and racks, carrier circuits, activation procedures, recovery-point and recovery-time objectives, and a recent exercise. If it were a hot site, routes or other live-service evidence would normally be expected. Neither case is established.
The prudent conclusion is not that failure has occurred. It is that operating status cannot be verified from public network evidence. That is a lower-confidence position than a live route withdrawal, but it is more useful than filling the silence with a story.
Five physical failure paths remain unanswered
The most important infrastructure question is not whether a component can fail. It is where a failure can cross a redundancy boundary and reach customers. For AMS1, five paths deserve explicit evidence.
Utility and generator failure. A utility outage should transfer the IT and cooling load through UPS to generators without interruption. The critical details are transfer time, battery autonomy, generator start reliability, continuous rating, fuel on site, refuelling access and performance after one generator is unavailable. A prolonged city outage can coincide with traffic congestion, fuel demand and reduced staffing. No AMS1 document states the generator topology or runtime. The CS Loxinfo brochure's 48-hour tank and 2,500 kVA generator are useful examples but belong to CSL's Cyberworld facility, not to this company.
Cooling failure. Loss of a precision-cooling unit should be absorbed by reserve capacity, but a common control, pump, condenser, power board or water supply can defeat unit-level redundancy. The sequence from alarm to rising inlet temperature to server shutdown can be short at high rack density. No AMS1 source identifies the cooling architecture, environmental limits or response time. It is therefore impossible to estimate thermal ride-through or to know whether standby generation carries the full cooling load.
Carrier-meet interruption. A backhoe, building fire, meet-me-room fault, failed optical platform or carrier maintenance action can remove connectivity while servers remain healthy. Recovery requires physically diverse entrances and providers, not two invoices riding one cable. BKNIX and numerous carriers are present at CS Loxinfo's tower, but no record puts AMS1 on those systems. The absence of prefixes and neighbours prevents even logical path diversity from being assessed.
Fire or water exposure. Fire suppression must protect both people and equipment while containing the affected compartment. Water can enter from flood, roof, pipe, cooling or firefighting systems. The World Bank's Thailand Economic Monitor describes Bangkok as exposed to frequent and severe flooding and recalls the scale of the 2011 disaster. That does not establish flood exposure at an unknown AMS1 site. It does show why a Bangkok-linked operator should disclose elevation, barriers, drainage, leak detection, fuel placement and access arrangements rather than rely on a city-level assurance. CS Loxinfo's brochure says its separate fourth-floor rooms use sealed junctions and leak sensors; again, that mitigation cannot be borrowed by the subject.
Construction or commissioning delay. If AMS1 is a planned node, the principal failure may occur before service: utility energisation slips, cross-connects are late, generators fail testing, permits remain open or customer migration starts before resilience is complete. An ASN can be allocated years before any of those dependencies are ready. AS136567 was registered in 2017 and still has no visible route history. The elapsed time does not prove cancellation, but it makes a current milestone, commissioning certificate and service date essential before treating the label as deployable capacity.
Each path also has a common-mode version. Power and fibre can share a street corridor. Both carriers can depend on one meet-me room. Generator controls and cooling controls can share a network. Primary and recovery staff can be unable to reach the same building during flood. Two sites can use the same upstream or utility substation. Redundancy counts components; resilience tests whether the components fail together. Public evidence for AMS1 does not reach either level.
Who is affected when the system fails
The affected population cannot be counted because no AMS1 customer list, hosted prefix or service catalogue is public. That uncertainty should block invented scale, not block analysis of the mechanism.
If AMS1 provides colocation, a power or cooling incident would affect tenants whose servers occupy its racks. Their end users might experience application, website, payment, communications or storage failures depending on workload design. If it provides transit or hosted routing, a carrier failure could isolate customer prefixes even while equipment remains powered. If it is a recovery site, its failure may remain invisible during normal operation but become critical precisely when the primary site is unavailable. If it is only a reserved ASN, there may be no operational customer impact today.
Responsibility also depends on the operating boundary. A tenant may own servers; a facility operator may own power and cooling; a network operator may own routers; carriers own external paths; the building owner controls some structural and access systems. The Cyber World Tower evidence demonstrates why those roles must not be collapsed. CS Loxinfo publicly claims the known facility systems. CyberWorld Data Center Co.,Ltd. holds AS136567. Without a contract or current service statement, liability and restoration ownership between them cannot be assigned.
Customers therefore need a dependency map tied to their service: rack and power circuit, cooling zone, meet-me room, carrier, prefix, upstream, remote-hands contact, escalation path and recovery site. A generic promise of "data centre" availability does not identify which organisation must act when the lights, chillers or routes fail.
Maintenance and recovery are where redundancy becomes real
Design diagrams describe what should happen. Maintenance and recovery records show whether it does.
The Thai investment conditions use the demanding phrase "concurrently maintainable": clients should continue to receive service while equipment is maintained or replaced. That requires more than spare components. Operators must be able to isolate switchgear, UPS modules, cooling units, generators and network paths without placing the whole service on an unprotected chain. Procedures, staffing and spare parts matter as much as topology.
For AMS1, no maintenance window policy is public. There is no notice period, list of intrusive work, escalation contact, change-freeze rule or record of maintenance completed without outage. There is no stated remote-hands coverage or response time. The APNIC role contact and abuse mailbox are not substitutes for an operations desk with authority over physical systems.
Recovery evidence should include transfer tests under load, generator runs long enough to expose fuel and thermal problems, UPS battery tests, cooling failover, dual-carrier traffic shifts, fire-system inspection, water-sensor tests and restoration exercises. The result should say what failed, how long transfer took, whether customers noticed and what was corrected. A tabletop discussion can test communication, but it cannot prove that a breaker opens, a generator starts or BGP converges.
Customer failover is equally important. Two sites do not provide recovery if data is not replicated, DNS changes are too slow, address space cannot move, application dependencies remain at the primary site or staff lack credentials. If AMS1 is intended to back up BKK2, the company should be able to show a recent exercise in which representative workloads moved and remained usable. The public route history offers no sign of such activation.
The right recovery question is therefore empirical: when the primary dependency was deliberately removed, what continued to work? For power, answer with load and transfer records. For cooling, answer with temperatures. For fibre, answer with route and packet telemetry. For people, answer with response times and escalation logs. CyberWorld-AMS1 has no public answers at that level.
Signals that matter, and signals that do not
Sparse companies often attract a cloud of indirect clues. Some are useful for forming questions, but they must not be promoted into proof.
The corporate listing suggests the legal company continues to operate and files under Internet services. The validated abuse mailbox suggests one contact channel was maintained in 2026. BKK2's live routes show that the registrant operates at least one publicly visible network identity. The Cyber World Tower contact address suggests some historical or operational connection to that building. These are relevant signals.
They cannot prove an AMS1 facility. A company can be legally active while a particular product is dormant. An abuse contact can be answered without a network being routed. A sibling ASN can run independently. A contact address can be an office, a former workplace, a customer location or a service provider's premises. None discloses power or cooling.
Third-party hosting offers that mention "CS Loxinfo / CyberWorld" are weaker still. They support the continuing market recognition of CS Loxinfo's Bangkok location, but the wording often abbreviates the building name and may confuse it with the Thai company's legal name. They do not identify AS136567, and they cannot transfer CSL specifications to the subject. Facility aggregators are useful discovery aids, not ownership instruments.
The absence of a PeeringDB record is also bounded evidence. Many networks do not maintain one, and private connectivity may not appear there. Yet an operator seeking customers for an interconnection location normally benefits from declaring facilities, exchanges and policy. Absence raises the cost of verification even if it does not prove absence of equipment.
Finally, the AMS1 string itself is the weakest physical signal. It may encode a plan, an old node, a reseller region, an internal naming scheme or an intended first Amsterdam presence. Only a site address, contract, live route or operator statement can settle which. The discipline is to let these clues shape the next question while leaving the factual conclusion unchanged.
What would prove current operation
CyberWorld Data Center Co.,Ltd. could convert the current low-confidence picture into a strong one with a compact set of current disclosures.
First, identify the asset. Give the legal facility operator, street location, building and data-hall boundary for AMS1. If the equipment is in a third-party colocation site, say which systems belong to the host and which belong to CyberWorld. If AMS1 does not refer to Amsterdam, explain the label. If the node has closed or never launched, state that directly so customers and registries do not infer capacity that is not available.
Second, identify the energised service. State contracted utility capacity, commissioned IT load, occupied load and saleable reserve. Describe utility sources, transformers, switchgear, UPS, rack distribution, generators, fuel autonomy and refuelling. Supply the redundancy class and its system boundary. A recent integrated test should show that IT and cooling survive a utility loss and a component failure.
Third, identify heat removal and physical protection. Disclose cooling topology, design conditions, reserve margin, backup-power coverage and environmental monitoring. State fire detection and suppression, compartmentation, leak detection, flood elevation and access arrangements. Provide current certification scope and expiry dates rather than an unscoped logo.
Fourth, identify connectivity. Name the facility meet-me room, cross-connects, carriers, exchanges, diverse entrances and upstream physical paths. Publish the prefixes intended for AS136567, corresponding route-origin authorisations, routing policy and a looking glass. Once routes are live, independent collectors should see them. If the ASN is deliberately private, provide customers with equivalent private evidence and stop presenting public reachability as implied.
Fifth, identify recovery. State maintenance notice periods, remote-hands coverage, incident contacts, recovery objectives and the last tested failover. Demonstrate traffic movement across independent carriers and workload recovery to an independent site. Explain any dependency shared with BKK2 or a Bangkok facility.
These disclosures would not need to reveal customer names, security-sensitive floor plans or exact cable routes. They would reveal enough to distinguish a functioning service from a reserved label. That distinction is routine in serious infrastructure procurement because customers are buying continuity, not nomenclature.
The investability question
Thailand's data-centre market is attracting large, power-intensive projects, and the country's authorities are tightening attention to electricity, water and environmental readiness. That context can make any company with "Data Center" in its name appear strategically positioned. It can also make small or dormant records look more valuable than their evidence supports.
CyberWorld's live BKK2 footprint is a real asset, albeit a small one in address-space terms. The company also has a long legal history and a current resource registration. Those facts could support a niche hosting or network operation. They do not establish a second geography, a powered data hall or a recovery platform under AMS1.
For a customer, lender or partner, the value of AMS1 is therefore option-like until deployment is shown. The ASN may ease a future launch, but number-resource ownership is not the expensive part of a data centre. Utility capacity, resilient electrical and cooling plant, secure space, carrier paths, staff and tested operations carry the capital and execution burden. No public figure can be assigned to those elements here.
The decision should be staged around evidence. Corporate and APNIC identity is sufficient to begin a conversation. BKK2's routes justify technical diligence. A purchase, migration or resilience dependency should wait for the site, power, carrier and recovery proofs described above. That is not a penalty for being small or private. It is the minimum needed to place production risk on an infrastructure service.
Verdict
CyberWorld-AMS1 is a validly registered autonomous-system identity held by an apparently operating Thai company. It is not, on the public evidence available on 10 July 2026, a verified operating data-centre node.
The decisive evidence is not one missing marketing page. It is the combination of no announced prefixes, no historical route origin, no observed neighbours, no current BGP state, no PeeringDB network entry, no disclosed facility and no subject-specific power or cooling specification. The live BKK2 sibling proves that the registrant can produce visible routing evidence when one of its networks operates. The rich infrastructure at Bangkok's Cyber World Tower proves that a similarly named location exists, but public records attribute that facility to CS Loxinfo and do not place AS136567 there.
The fair conclusion is neither accusation nor assumption. AS136567 may be reserved, private, dormant or planned. Any of those states can be legitimate. What it cannot yet carry is a claim of current, resilient data-centre capacity. To earn that claim, CyberWorld-AMS1 must connect its label to an identified operating boundary and then make that boundary stand up to power, cooling, carrier and recovery evidence.