Summary

  • LACNIC data-centre address-demand analysis asks how colocation, hosting, edge caches, cross-connects, remote hands and enterprise onboarding turn IPv4 into occupancy-enabling capital.
  • Scarcity changes facility economics because public addresses carry reputation, customer continuity, security evidence and migration optionality that cannot be replaced by rack space alone.
  • A credible regional ledger should make address control portable and legible without turning commercial use into a rationing contest run by the registry.

In Latin America and the Caribbean, the next scarcity problem for colocation is not only land, power or fibre. It is the stable public address inventory that lets an enterprise become a customer without rewriting the way its business is reached.

The customer arrives before the theory

A familiar data-centre sale in the region begins in practical language. A bank wants a disaster-recovery cage outside its main city. A payments processor wants racks with redundant power, upstream diversity and private links to merchants. A software company wants to move an ageing server room into a professional colocation site. A retailer wants hosted firewalls, VPN concentrators, mail gateways and customer-facing applications close enough to local users that latency and support do not become daily complaints. In the Caribbean, a public contractor may want a small but resilient footprint because international circuits are expensive, storms complicate logistics and keeping some services close to citizens is not a slogan but an operating preference.

The conversation is commercial. How many racks? Which power density? What cooling class? Which carriers are on-net? Can the facility arrange cross-connects to an upstream provider, an internet exchange, a payment network, a security partner or a content cache? Is remote hands available at night? Are spare parts stored locally? How quickly can equipment clear customs, reach the site and be turned up? How much of the install can be done without flying an engineer across a continent or between islands?

Then comes the quieter question: what public addresses are available, and under whose control will they remain?

The answer matters more than the sales deck admits. A customer entering a colocation facility is not merely renting metal space. It is attaching business processes to a public identifier layer. It may need stable addresses for VPN endpoints, e-commerce sites, payment processing, DNS, mail reputation, security appliances, monitoring probes, APIs, remote administration, lawful-intercept interfaces, customer portals or legacy software written in an earlier network era. It may need addresses that can be routed through one provider today and another later. It may need address records that satisfy auditors, banks, acquirers, insurers, security vendors and counterparties. It may also need to avoid a forced renumbering exercise that interrupts customers who neither know nor care how address registries describe themselves.

This is why data-centre address demand in the LACNIC region should not be treated as a footnote to cloud computing. The large cloud platforms are relevant, but they are not the whole market and should not define the problem. Much of the region's digital economy runs through ordinary facilities: carrier-neutral colocation buildings, telco-hosted data centres, enterprise disaster-recovery rooms, local hosting firms, payment platforms, content distribution nodes, managed-service providers, public-sector contractors and regional integrators. Their problem is not bargaining with a hyperscale platform. It is continuity of service in a market where public IPv4 addresses have become scarce, tradable, leased, reputationally sticky and operationally embedded.

That distinction separates this subject from adjacent ones. Interconnection dependency is about who sits between networks. Submarine-cable shock is about physical international capacity and route diversity. Cloud-provider address power is about platform gatekeeping. Data-centre address demand is different. It is the economics of turning facility capacity into occupied, revenue-producing, customer-continuous infrastructure under IPv4 scarcity. It asks what happens when a rack, a cross-connect and a cooling unit are available, but the public address continuity behind them is not.

A data-centre address is not a server count

The lazy way to estimate address demand is to count servers. That mistake survives because old allocation narratives often imagined addresses as direct reflections of hosts. Modern colocation demand does not work like that. One server may sit behind network address translation; one small virtualised platform may require many public service endpoints; one security stack may need separate addresses for management, inspection, egress, failover and logging; one reputation-sensitive mail environment may avoid shared address pools even when the underlying compute footprint is small.

The useful unit is not the server. It is the externally reachable function that must remain stable enough for customers, partners and machines to trust it.

Consider a mid-sized enterprise moving from an office server room into a facility in Sao Paulo, Santiago, Bogota, Buenos Aires, Panama City or Kingston. Its physical footprint may be modest: a half-rack, a firewall pair, several virtual hosts, storage and backup. Yet its address needs can be stubborn. Corporate VPNs often depend on whitelisted source and destination addresses. Payment endpoints may be registered with processors, anti-fraud vendors and banks. Mail gateways carry reputation histories that are difficult to replace. Customer APIs may be embedded in partner applications. Monitoring and incident-response tooling may expect stable public addresses. Security teams may segment public services from management, production from staging and customer egress from administrative access. Regulators, auditors and enterprise clients may record network indicators in risk files.

Renumbering is therefore not cosmetic. It can mean updating firewall rules across counterparties, changing certificates and DNS, revalidating payment routes, warming up mail reputation, notifying customers, revising allowlists, changing logs, adjusting SIEM rules, updating disaster-recovery runbooks, retesting VPN tunnels, rebuilding incident-response assumptions and explaining to executives why a facility migration created visible risk. The smaller the customer, the less spare administrative capacity it has for such work. A large bank may assign a programme office to renumbering. A local exporter, clinic network, university, manufacturer, fintech or municipality may have two engineers who also keep the business alive.

IPv6 does not remove this issue on the commercial horizon relevant to most customers. IPv6 may be present, desirable and technically elegant for many internal and new services. But the public internet in the region remains a dual-stack, IPv4-dependent economy. Many consumer connections, enterprise allowlists, fraud tools, legacy applications, security policies and commercial integrations still assume IPv4 reachability. Carrier-grade NAT can help access networks conserve public IPv4, but it does not make an enterprise's public service endpoints disappear. It often increases the value of clean, attributable, stable public addresses for the servers and appliances that remain visible from the outside.

Reputation turns addresses into history. An IPv4 address used for mail, payments, API traffic, hosting or security telemetry accumulates signals. Some signals are good: a clean sending history, a stable customer base, normal traffic patterns and trusted allowlists. Some are bad: spam, malware, abusive hosting, compromised machines, fraud or unresolved complaints. Facility operators and hosting providers therefore care not only about how many addresses they can obtain, but what those addresses have been used for, whether they can be assigned safely, how abuse contacts are handled and whether a customer can take responsibility without losing operational continuity.

Logging and accountability add another dimension. In a shared hosting or managed-service environment, addresses help map public activity to customers, time windows and services. A facility may use private addressing internally, but public attribution still matters when complaints arrive, regulators ask questions or security vendors investigate incidents. In countries with uneven enforcement capacity, limited technical staff or cross-border customer bases, clean records are not a luxury. They reduce the cost of resolving disputes.

For all these reasons, data-centre address demand is lumpy, sticky and function-specific. It grows with customer onboarding, not merely with server installation. A colocation provider may have available floor space, power and carrier access, yet still face a constraint if it cannot offer customers predictable public addressing. Address inventory is therefore part of the commercial product, even when it is not presented that way on the price sheet.

Scarcity changes the contract

IPv4 scarcity changed the economics of colocation more deeply than many facility owners admit. In the allocation era, addresses were treated as administrative inputs. A network planned a service, justified need, received space and expanded. The scarcity price was hidden because the registry process rationed a free or low-cost resource. Once the free pool ceased to behave like a normal supply channel, the price became explicit. Addresses could be bought, leased, transferred, financed, reserved or bundled into hosting contracts. The registry record that once looked clerical became a record above capital.

In a data-centre business this change appears as a contract problem. A facility can sell square metres, rack units, kilowatts, remote hands, cross-connects and managed services under ordinary commercial terms. Address inventory does not fit so neatly. It may be owned or controlled by the facility operator, by an upstream provider, by a customer, by a hosting subsidiary, by an acquired company, by a brokered lease or by a legacy holder that wants revenue without losing title-like continuity. It may be routed from a provider aggregate today and moved later. It may be part of a block whose reputation varies by subnet. It may be subject to contractual limits on reassignment, geolocation, abuse handling, transfer, return, payment default or customer sector.

Scarcity makes each of those terms material. When addresses are abundant, a dispute over a small assignment can be solved by issuing another block. When addresses are scarce, replacement is costly and sometimes unavailable at the necessary speed. A provider that loses control over a customer-facing block may lose the customer. A customer forced to renumber may delay a facility migration. A hosting firm that cannot replenish clean IPv4 inventory may stop accepting certain workloads. A disaster-recovery site without stable public endpoints may be useful for storage but not for rapid failover. A content cache without sufficient addresses may be constrained in how it segments services or peers locally.

The result is a secondary market inside the facility economy. Some operators buy blocks outright when they have the balance sheet. Others lease addresses to match customer demand and avoid capital outlay. Some rely on upstream providers, accepting lock-in in exchange for convenience. Some ask customers to bring their own addresses, though many enterprises in the region have no portable inventory or lack the staff to manage routing and records. Many providers mix all these approaches, creating a portfolio of address sources that must be tracked with more care than commercial teams often realise.

Leasing is especially important for colocation and hosting because customer demand is uncertain. A facility may sign a customer for two racks and discover that the address requirement is much larger than the compute footprint implies. Another customer may need a block for a project cycle, an e-commerce season, a public-sector contract or a migration window. A hosting provider may want to test demand before committing capital to a purchase. Leasing lets address supply match variable occupancy. It also creates risk if the registry layer treats leasing as morally suspect or if records cannot cleanly reflect the party using, controlling or responsible for a block at a given time.

Transfer economics are equally important. If a facility operator acquires a local hosting firm, the real asset may not be servers; it may be customer contracts, domain relationships, mail reputation, address inventory and routing continuity. If the address record cannot follow the economic transaction predictably, the acquisition is discounted. If a registry process introduces unpredictable delay or discretion, the buyer prices the risk. If an address block is trapped by regional or policy categories, the seller receives less value. In a region where infrastructure capital is already sensitive to currency volatility, political risk and power cost, adding registry uncertainty raises the cost of building local capacity.

Scarcity therefore should have narrowed the registry role. A scarce capital input needs clearer records, lower transaction costs, faster recognition of control, better dispute metadata and stronger portability. It does not need moral rationing after the market has already revealed demand. Once a customer is paying for facility space, power, cooling, carriers and addresses, its need is not a theory for a meeting room to approve. It is a capital commitment.

Facilities turn addresses into occupied space

A data centre is a conversion machine. It converts land, electricity, cooling, fibre paths, security, engineering labour and operating discipline into customer continuity. Address inventory is part of that conversion. Without it, many customers cannot turn a rack into a live service.

Power is the first constraint in much of the region. High-density equipment, GPU clusters, storage arrays and modern enterprise platforms require more predictable electricity than many office buildings can provide. Colocation sells redundant feeds, generators, batteries, switchgear and monitoring. Cooling is the second constraint. Tropical humidity, heat, salt air in coastal markets, urban density and energy prices make thermal management a serious operating discipline. Cross-connects are the third. A rack becomes valuable when it can reach carriers, internet exchanges, cloud on-ramps, payment networks, security partners and content networks without long procurement cycles.

Addresses sit across all three. A customer may buy redundant power and cooling because it wants public services to stay up. It may buy cross-connects because it wants routing options, lower latency or separate upstreams. It may buy managed firewalls because public endpoints create exposure. It may buy remote hands because a public-facing appliance must be replaced quickly. It may buy dual facilities because it needs continuity if one city, power grid or cable path fails. In each case the address is not the main visible product, but it is the identifier through which the service remains reachable.

This makes address inventory an occupancy-enabling form of capital. A facility with good power but no address plan may fill with private workloads, storage and interconnection nodes, but it will struggle to win customers whose public services require stable IPv4. A facility with clean, portable inventory can shorten onboarding, reduce customer friction and support higher-value services. A managed hosting firm with disciplined address records can sell not just virtual machines but business continuity. A carrier-neutral site that helps customers bring, lease or transfer addresses without lock-in can compete against telco sites whose address assignment is tied to upstream service.

The economics resemble parking rights in a city, airport slots or port access, but with one important difference: IP addresses are not physical land. Their value comes from uniqueness, routability, reputation, records and reliance. They can move across facilities and providers if the ledger, routing and contracts allow it. That portability is precisely why the registry must remain narrow. If the registry turns an administrative record into a discretionary licence over commercial use, it converts a portable identifier into a captive facility input. That damages the facility market as much as the address market.

Customer onboarding exposes the issue most clearly. The sales team closes a deal. The implementation team asks for network details. The customer wants public addresses for firewalls, VPNs, applications, out-of-band management, mail relays and sometimes customer tenants. The provider asks whether the customer has its own space. Often the answer is no, or the space is not portable, or the paperwork is old, or the staff who understood the network have left. The provider can allocate from its pool, lease space, route a customer's block or obtain new supply. Each option has operational and legal implications.

If provider-assigned addresses are used, the customer may become dependent on the facility or upstream provider. If the customer later moves, renumbering risk returns. If leased third-party addresses are used, the facility must track contract term, abuse responsibility and return conditions. If customer-held addresses are routed, the facility must support announcements, route filtering, RPKI and operational coordination. If acquired addresses are transferred, the facility must rely on predictable registry recognition. A narrow ledger makes these options visible and manageable. A gatekeeper makes them uncertain.

The region's facility economics magnify the problem. A new data-centre build in Brazil, Mexico, Chile or Colombia can involve large capital expenditure and long planning cycles. Smaller markets may have fewer carrier choices and higher unit costs. Island markets may face imported equipment delays, hurricane risk, limited local spares and dependency on submarine routes. In all cases, empty capacity is expensive. The ability to onboard customers quickly and keep them portable enough to trust the facility is part of the return on invested capital.

The regional shape of demand

Latin America and the Caribbean are not one infrastructure market. The phrase is useful for registry administration, but it hides very different economies. Brazil has scale, local content demand, large enterprises and dense metropolitan markets. Mexico sits between domestic demand, North American supply chains and a strong manufacturing base. Chile has attracted infrastructure interest because of political stability, energy debates and a role as a Pacific-facing hub. Colombia, Argentina, Peru, Panama, Costa Rica and the Dominican Republic each have different combinations of enterprise modernisation, carrier concentration, cable geography, public-sector demand and currency risk. Caribbean islands face a distinct mix of tourism, public administration, disaster resilience, offshore finance, high import costs and limited local technical labour.

These differences matter because address demand follows customer type. In a large metropolitan market, the colocation customer may be a bank, retailer, media group, health provider, logistics platform, outsourcing company, gaming firm, internet exchange participant or regional SaaS provider. It may need multi-homed IPv4 space, separate environments, compliance evidence and reputation continuity. In a small island market, the customer may be a government agency, hotel group, local ISP, disaster-recovery contractor, university, hospital or managed-service provider. It may need fewer addresses in absolute terms, but the addresses can be more important because replacement supply and technical alternatives are thinner.

The region's enterprise modernisation is uneven. Many companies are not moving from clean, cloud-native architectures to sophisticated hybrid infrastructure. They are moving from legacy server rooms, improvised hosting, single-provider broadband, ageing firewalls and undocumented public addresses. Their first professional colocation project often reveals address debt accumulated over years. Old /29s, provider-assigned /28s, forgotten mail relays, untracked NAT rules, legacy VPNs and vendor allowlists become obstacles. The data centre becomes the place where these hidden dependencies must be priced.

Local hosting firms and managed-service providers add another layer. They often serve customers that are too small for direct international platform relationships but too important to run on consumer-grade connectivity. They host websites, mail, ERP software, payment applications, remote desktops, security tools and sector-specific software. Their address needs may be more granular than those of a single enterprise because they must separate tenants, manage reputation and respond to abuse. For them, IPv4 inventory is working capital. A provider that cannot obtain clean addresses cannot accept certain customers, cannot segment risk properly and cannot grow without increasing shared-reputation exposure.

Edge caches and content nodes create a different pattern. They are not the whole story, but they are significant in the region because international capacity, latency and user experience still matter. A content cache placed in or near a metro facility reduces transit cost and improves performance for local users. It needs addresses, peering, routing discipline and local support. The value is not simply the number of servers in the cache. It is the ability to make content appear local, reachable and reliably segmented. In smaller markets, even a modest cache can affect user experience and upstream cost. In larger markets, multiple caches and content networks become part of the facility's attractiveness.

Disaster recovery also shapes demand. Latin America has earthquake exposure, flood risk, grid instability in some places, political disruption and currency shocks. The Caribbean adds hurricanes and island logistics. Enterprises that once relied on a single office server room increasingly want secondary sites. Yet a disaster-recovery site that cannot preserve public reachability is incomplete. If failover requires renumbering, reauthorising customers, changing payment endpoints and rebuilding reputation, it is not a rapid recovery plan. Portable address inventory can therefore make a secondary facility more valuable than its physical size suggests.

The cost of renumbering is a regional tax

Renumbering is often described as an engineering task. In the data-centre market it behaves more like a tax on movement. It discourages customers from changing providers, delays migration from poor facilities, increases dependence on upstream carriers and reduces the credibility of disaster recovery. The tax is heavier in markets where technical labour is scarce, documentation is weak and counterparties are slow to change allowlists.

A customer that uses provider-assigned addresses may receive a low initial price and easy setup. The cost appears later. When the customer wants a second facility, a different carrier, a merger, an acquisition or a migration to a better data centre, the addresses do not necessarily move. The provider may not be malicious. It may simply have built its network around aggregate space that cannot be delegated cleanly. But the effect is lock-in. The customer must choose between staying with the provider or absorbing a renumbering project whose business cost exceeds the monthly network bill.

This is especially damaging for colocation competition. A new entrant may offer better power, cooling, security and customer service, but customers tied to non-portable addresses cannot move easily. An incumbent with address control can retain customers even when its facility performance is weaker. That is not efficient competition. It is identifier lock-in masquerading as network service.

Renumbering also undermines resilience. The entire point of a secondary site is to reduce dependency on one facility, one carrier, one city or one grid. If the address plan is not portable, resilience is partial. The customer may replicate data, install equipment and test backups, yet still struggle to make public services reachable under the same trusted identifiers. Some applications can use DNS changes and load balancers. Others cannot, or cannot within the required recovery window. Even where DNS is flexible, upstream allowlists, payment integrations, security policies and human procedures may lag.

Mail reputation is a concrete example. A business that moves mail gateways to new addresses may face warming periods, false positives, blocked messages and support calls. Shared hosting providers know this problem well. A clean block can be worth more than its raw address count because reputation reduces support cost. A tainted block can impose customer churn. If a provider is forced to renumber into unknown space, the facility migration may create business noise unrelated to the quality of the new site.

Security tools create another form of stickiness. Firewalls, WAFs, VPN concentrators, DDoS protection services, SIEM feeds, threat-intelligence allowlists and vendor-managed appliances often bind policy to public addresses. Large multinational customers may have change-management teams. Regional firms may not. Many rely on outside consultants, local integrators or vendor support channels that operate slowly. A forced address change therefore consumes scarce human capacity.

The same is true for public-sector and regulated customers. Government agencies, utilities, health networks and financial institutions may have formal approval processes for network changes. A colocation move that should be a technical improvement can become a procurement and compliance event if addresses change. Smaller jurisdictions may have limited cyber staff, making each renumbering project a queue problem. The hidden cost is not just money; it is delay in modernisation.

The rational response is to value portability. Customers with their own portable space have bargaining power. Facilities that can support customer-routed space, clean leasing arrangements and predictable transfer recognition can reduce migration cost. Address holders that can lease without losing control create useful liquidity. A registry ledger that records control and responsibility without policing the commercial reason for use lowers the renumbering tax. A registry that treats every commercial arrangement as a permission question raises it.

In economics, a switching cost can protect incumbents and reduce welfare. In the LACNIC data-centre market, non-portable IPv4 is one of the most important switching costs. It affects not only price but also resilience, competition and the speed at which enterprises leave weak infrastructure for better facilities.

The ledger is not the landlord

The registry function exists because uniqueness matters. Two networks cannot safely claim the same public number resources. Records must make it possible to know who controls a block, how to contact the responsible party, what security assertions exist, whether a transfer has occurred and whether a dispute is visible. This is a narrow but essential function. It is an address book with economic consequence.

The danger begins when the address book starts behaving like a landlord.

A registry ledger should describe operational, legal and market reality. It should not create that reality by discretionary permission. In a data-centre context, that distinction is practical. A facility may lease addresses to a hosting customer. A customer may bring its own block. A managed-service provider may route addresses through one upstream while maintaining commercial control. A company may acquire a hosting firm and need its address records to follow. A disaster-recovery provider may temporarily announce a customer's space during failover. A content cache may use addresses under a contract that changes as traffic shifts. These arrangements do not threaten uniqueness if records are accurate and conflicts are visible.

They do threaten an institution that wants to treat registration as a licence over commercial structure.

The narrow ledger asks objective questions. Is the block unique? Who is the holder or controller of record? Is the asserted change authorised by the relevant holder or competent legal process? Are contacts reachable? Are routing-adjacent security assertions consistent with the holder's intent? Is there a dispute that counterparties should know about? Can the state be exported, audited and relied upon if the incumbent registry fails? These questions fit the technical and commercial problem.

The gatekeeper asks different questions. Is the customer's business model approved? Is leasing morally acceptable? Is the customer sufficiently local? Is the address use aligned with a preferred development narrative? Has the holder justified need to a private institution after already committing capital? Should a transfer be refused because a policy room dislikes the movement of value? Should records be impaired as punishment for behaviour better handled by contracts, courts, security vendors or public law?

Those questions do not protect uniqueness. They convert registry control into capital control.

The distinction matters more in Latin America and the Caribbean because data-centre markets often need external capital, patient infrastructure investors and flexible commercial models. A facility build already faces power negotiations, permitting, equipment import, currency risk, customer education and carrier coordination. If address records are also subject to discretionary views about commerce, investors discount the project. They may not describe the discount as registry risk in a board paper, but it appears in slower expansion, higher required returns, narrower customer offerings and greater reliance on incumbent carriers with existing address pools.

A narrow ledger would improve the market. It would make transfers faster and more predictable. It would make leases safer to disclose. It would let customers know who is responsible for abuse and routing. It would preserve title-like continuity without pretending addresses are land. It would allow a holder to change facilities or upstreams without asking a regional administrator to approve the business reason. It would also make disputes easier to isolate: the ledger can mark conflict without disabling unrelated running networks.

This is the institutional economics of the registry layer. A useful registry reduces transaction costs. A sovereign registry increases them. A useful registry makes reality legible. A sovereign registry threatens reality when it dislikes it. A useful registry is valuable because operators trust it. A sovereign registry is powerful because exit is hard. The data-centre market needs the first and should fear the second.

Portability is title-like continuity

Portability is often discussed as an operator right, but in the data-centre market it is also a customer-protection mechanism. A customer that can keep its addresses while changing facility, carrier or managed-service provider has a stronger bargaining position. A facility that can support portable addresses can win customers on service quality rather than identifier captivity. A lender or investor that sees address continuity as reliable will value infrastructure and hosting businesses more fairly.

The word title must be used carefully. An IP address is not a parcel of land, and the legal treatment of number resources varies by contract, jurisdiction and institutional practice. But the market still needs title-like continuity: a stable chain of recognised control, transfer history, reliance protection, dispute visibility and the ability to keep using the resource unless a narrow, objective defect exists. Without that continuity, the address behaves less like capital and more like a revocable permission slip.

Data-centre customers feel this difference even if they never use the vocabulary. A company with portable space can colocate in one facility, buy transit from another provider, peer at an exchange, move to a second site, add a disaster-recovery announcement or change managed-service vendors. The address plan is part of its architecture. A company without portability may discover that its public identity belongs operationally to the provider it wants to leave. It can migrate servers but not reachability. It can copy data but not reputation. It can improve infrastructure but not escape the switching cost.

For facility owners, portability changes sales incentives. If customers fear lock-in, they hesitate to outsource critical services. If they believe the facility supports portability, they can move more workloads into colocation. The provider may lose some captive power, but it gains credibility. In a developing market, credibility can be worth more than lock-in because the larger commercial task is to persuade enterprises to leave improvised infrastructure behind.

Portability also disciplines registries. A service provider with no exit option can be told to accept process delays, policy fashion or institutional errors. A holder with a credible portability path has leverage. The registry must keep records accurate, fees reasonable, processes predictable and services technically competent. Exit is not chaos. It is the ordinary accountability mechanism of voluntary coordination.

The regional implications are important. A small Caribbean provider should not be trapped by the failure, capture or policy drift of any single registry institution. A Brazilian hosting firm should not see acquired address inventory discounted because registry approval is uncertain. A Chilean disaster-recovery provider should not depend on a moral distinction between commercial models when the technical fact is that customers need reachable endpoints. A Colombian managed-service firm should be able to lease, route and document addresses without hiding ordinary business reality. An Argentine enterprise should not lose continuity because currency stress makes one provider unviable and a move becomes necessary.

Portability does not require a registry to approve every business plan. It requires the common layer to preserve uniqueness and chain of control while allowing the holder to choose facilities, counterparties and commercial structures. That is why a narrow ledger and title-like continuity belong together. The ledger records the continuity. It does not own it.

Customer onboarding is where policy becomes price

Institutional theory becomes real during onboarding. A facility can advertise resilience, but the implementation spreadsheet reveals whether resilience is affordable. Each customer adds questions: addresses required at launch, additional addresses reserved for growth, routing design, upstream diversity, abuse handling, geolocation expectations, reverse DNS, RPKI, DDoS plans, mail reputation, logging responsibility, contract term and exit conditions.

If addresses are plentiful and portable, onboarding is a technical project. If addresses are scarce and non-portable, onboarding becomes a negotiation over risk. Who supplies the addresses? Who pays for them? Can they be used for this customer sector? Can they be routed from another facility? What happens if the customer leaves? What happens if the lessor withdraws? What happens if an abuse complaint is mishandled? Can the customer assign addresses to its own tenants? Can a block move after a merger? Are there records proving control? Will a bank, acquirer or auditor accept the arrangement?

Each uncertainty slows revenue. A rack that waits for address clarity is idle capital. A customer that delays migration because address arrangements are unclear is a lost sales cycle. A managed-service provider that turns away customers because it cannot segment them safely loses margin. A facility that relies on upstream-assigned space may accept lower churn but also lower strategic value because customers see it as a network extension rather than neutral infrastructure.

The price effects are subtle. Address scarcity may appear as a line item in hosting plans, but often it is embedded in higher setup fees, longer contracts, stricter acceptable-use terms, more conservative customer screening, larger deposits, bundled transit, reduced portability or reluctance to serve risky sectors. Customers may not see a separate IPv4 capital cost. They see fewer choices and slower projects.

This is where registry policy can either reduce or increase friction. A registry that records leases, sub-assignments, transfers and contact responsibility in a neutral way makes onboarding safer. It encourages disclosure. It lets the facility, customer, lessor and upstream provider align records with reality. A registry that treats ordinary commercial structures as suspicious encourages opacity. Operators still need addresses, so they route, contract and assign in ways that may not be fully reflected in the public ledger. The database becomes less accurate because being accurate is risky.

Accuracy should be the registry's friend. In a scarce market, the best way to improve record quality is to make truthful recording low-cost and non-punitive. If a customer uses leased space in a colocation facility, the record should make responsibility clear. If a block is temporarily announced for disaster recovery, the record should permit the relevant assertion. If a customer brings a block, the facility should be able to support routing without acquiring economic control. If a transfer occurs after an acquisition, the ledger should update quickly. None of this requires the registry to judge whether the customer's business is sufficiently virtuous.

The alternative is mandate laundering. A private institution uses public-interest language to claim authority over business decisions that properly belong to operators, customers, markets and public law. In the facility market this laundering is expensive. It adds process to customer onboarding, uncertainty to investment and delay to modernisation. It may be defended as fairness, but scarcity controls do not create more IPv4. They allocate discretion.

Small and medium-sized customers pay the highest price. Large firms can buy addresses, hire counsel, influence providers and absorb delay. Smaller firms need clear prices, fast onboarding and transparent obligations. They benefit from liquidity, not paternalism. A data-centre market that wants to serve domestic enterprise modernisation should prefer predictable address commerce to discretionary rationing.

Abuse, reputation and the enforcement temptation

Data-centre operators cannot ignore abuse. Hosting attracts compromised machines, phishing, spam, credential theft, botnet command traffic and fraudulent sites. Facilities and managed-service providers need credible abuse processes because upstreams, peers, payment providers, security vendors and governments demand them. The question is not whether abuse matters. It is which institution should do what.

A registry ledger should support contactability and responsibility. It should record the holder, relevant contacts, delegation, routing-adjacent assertions and validated channels for operational notices. It should make it easier for a complainant to find the responsible party and for counterparties to understand who controls a block. That is a directory-accuracy function.

It should not become a general enforcement authority over hosting conduct. Registries must not become enforcers.

The difference is crucial for data centres. A facility may host thousands of customers through resellers and managed-service providers. Abuse complaints vary in quality. Some are precise and urgent; others are automated, stale, politically motivated, commercially strategic or simply wrong. Public authorities differ in competence and legal standards. Security vendors make errors. Competitors can complain. Customers can be compromised without malice. A registry that turns contact or abuse allegations into a path for impairing address records risks converting operational disputes into identifier sanctions.

This would be especially dangerous in markets where many facilities support politically sensitive customers, media, public-sector services, finance, gaming, adult content, remittances, offshore services or controversial speech. The registry is not equipped to adjudicate the substance of these disputes. Courts, contracts, upstream providers, facility terms, enforcement bodies and security communities all have roles. The ledger's role is narrower: make responsibility visible, not become prosecutor and judge.

Reputation markets already discipline bad conduct. Upstreams can filter. Payment providers can terminate. Security vendors can list. Facilities can evict customers. Insurers can price risk. Courts can order remedies. Governments can enforce law within their jurisdiction. These tools are imperfect, but they carry domain-specific authority and liability. A registry that adds database impairment to the stack may not improve enforcement; it may create a new chokepoint with weak due process and large collateral damage.

Collateral damage is not theoretical. If a block supports shared hosting, mail services, payment endpoints or public-sector services, adverse registry action can affect innocent customers. If the facility loses a block because of one customer's abuse, the harm spreads. If RPKI or reverse DNS continuity is impaired, routing and service reliability may suffer. If a transfer is delayed because abuse allegations are unresolved, an acquisition or migration may fail. The registry's small administrative decision can become a facility-level outage.

The correct model is separation. The ledger records control, contacts, delegation and dispute metadata. Operators enforce contracts. Upstreams manage routing relationships. Courts and regulators handle legal violations. Security vendors publish risk signals. Facilities remove or isolate customers when evidence justifies it. The registry does not use address records as punishment except in narrow cases of fraud, duplicate claims, competent legal order or objective record failure that directly threatens the integrity of the ledger.

This separation is not soft on abuse. It is disciplined about institutional competence. A data-centre economy needs clean reputation and reliable enforcement, but it also needs confidence that public addresses will not be converted into hostages during ordinary disputes. Stability comes from narrow roles, not from giving every institution a bigger stick.

The Caribbean and the edge make the point sharper

The Caribbean illustrates why small absolute address demand can still be economically significant. Many islands have limited domestic markets, high logistics costs, exposure to storms, dependence on submarine links and concentrated telecom sectors. A local data-centre footprint may be small compared with Sao Paulo or Mexico City, but the marginal value of stable public addressing can be high because alternatives are fewer.

A hotel group, offshore service provider, government agency, hospital network or regional managed-service firm may need only modest colocation. Yet it may need public endpoints that survive outages, serve local users, support remote administration and keep international counterparties connected. If the island relies heavily on external hosting, latency, cable disruption and jurisdictional concerns become part of the service calculation. A small local facility with reliable power, cooling, cross-connects and address continuity can therefore have importance beyond its size.

Edge infrastructure has a similar logic. The point of an edge cache, security node, DNS platform or application point of presence is to bring a function closer to users and networks. It may not require a huge address block, but it needs addresses that are clean, routeable and operationally stable. In markets where international transit remains expensive or cable routes are vulnerable, such nodes can reduce cost and improve user experience. Address friction can slow their deployment.

The same is true for disaster recovery. An island government or bank may want a secondary site in another island, a mainland facility or a nearby regional hub. If address portability is weak, failover becomes more complex. If address records cannot reflect temporary operational control, disaster arrangements become less transparent. If a registry or provider treats customer geography as a moral condition, regional resilience suffers. In a storm, earthquake or political crisis, the relevant question is not whether an address sits inside an elegant administrative narrative. It is whether the service remains reachable.

Small markets are also more vulnerable to incumbent lock-in. A limited number of carriers may control local routes, facilities and address pools. Customers may accept provider-assigned addresses because there is no easy alternative. A neutral colocation entrant may struggle to win customers if address portability is weak. A managed-service provider may become dependent on one upstream inventory source. This can reduce competition even when physical infrastructure improves.

A narrow registry ledger helps because it lowers the cost of alternative arrangements. It supports customer-held space, leased inventory, transparent sub-assignments, neutral facility routing and cross-border disaster recovery. It does not need to decide whether a small island's demand is morally deserving. The customer, facility, lender and counterparty price that demand. The ledger preserves uniqueness and continuity.

What LACNIC should not become

The temptation for any regional registry is to treat scarcity as an invitation to govern. The language is familiar: stewardship, community, fairness, conservation, proper use, regional need, protection of scarce resources. Some of these words once had administrative meaning when free-pool allocation was the main task. In a market of scarce, leased, transferred and operationally embedded IPv4, the same words can become instruments of capital control.

For the LACNIC data-centre market, the risk is not that the registry will run facilities or choose customers directly. The risk is subtler. Policy can make ordinary address commerce slower, less portable and less transparent. It can privilege incumbents with old inventory. It can make leasing hard to disclose. It can make transfers uncertain. It can blur the line between contact accuracy and conduct enforcement. It can treat regional administration as if it created regional ownership. It can force renumbering by making portability conditional. It can make address records less truthful because operators fear the consequences of telling the truth.

That would damage precisely the customers the rhetoric usually claims to protect. Smaller networks, domestic hosting firms, local enterprises and island providers need low transaction costs. They need supply to move toward use. They need address holders to lease or sell without fear that the registry will punish commercial reality. They need clean records that lenders, auditors, customers and counterparties can understand. They need the ability to leave bad providers and choose better facilities.

The registry should therefore avoid becoming an admissions committee for data-centre business models. It should not decide whether enterprise hosting is more worthy than content caching, whether a managed-service provider has too many addresses for its rack count, whether leasing is less virtuous than direct holding, whether a disaster-recovery arrangement is local enough or whether a facility's customer mix satisfies a regional development narrative. These are not uniqueness questions.

Nor should the registry turn non-adoption of preferred practices into invalidity. The internet works because operators adopt what they can run and counterparties accept what is useful. If a facility supports a customer's portable block, if an upstream carries it, if security assertions are valid and if the ledger records control accurately, the commercial arrangement should not require a separate moral blessing. Running-code primacy means that operational reality has weight. A policy document should not override a working, non-conflicting network merely because an institution wants broader authority.

This is not an argument for lawlessness. Fraudulent claims, duplicate records, forged authority, hijacking, broken contact records and inconsistent security assertions are real registry concerns. They threaten the ledger itself. The registry can and should deal with them. But it must distinguish threats to the ledger from discomfort with commerce. A data-centre customer leasing addresses for enterprise hosting is not the same kind of problem as a forged transfer. A facility routing a customer's portable block is not the same kind of problem as a duplicate claim. A content cache using local addresses is not a constitutional issue.

The institutional test should be severe: what breaks in the running internet if this rule is not centralised at the registry? If the answer is not uniqueness, accurate control records, security integrity, fraud prevention, dispute visibility or operational continuity, the rule belongs elsewhere.

Number Resource Society as the positive model

Critique is incomplete without a positive model. The direction that fits data-centre address economics is not a stronger regional gatekeeper, a government takeover or a new priesthood with better slogans. It is a holder-centred model of narrow coordination, portability, transparency and collective protection. That is the significance of the Number Resource Society idea.

The useful part of the model is not branding. It is architecture. Resource holders need a common way to insist that number governance remain a coordination layer rather than a discretionary control layer. They need representation that begins with actual holders and operators, not with a room claiming to speak for absent users or an institution invoking a region as if it were an owner. They need mechanisms for exit, portability, redundancy, ledger accuracy, dispute visibility and failover. They need a way to make registry-side risk visible before each holder faces it alone.

For data-centre markets, this matters because address demand is distributed across many customers. A single facility, hosting firm or enterprise may not have enough leverage to resist creeping registry authority. Together, holders can defend the principles that make the facility market work: addresses must remain unique; records must remain accurate; holder rights must be respected; commercial use must not be policed by the ledger; leasing and transfers must be recorded rather than driven underground; portability must be real; and running networks must not become hostages.

The Number Resource Society model also avoids a common error in reform debates. It does not need to claim that all registries are evil or that all institutions should vanish overnight. The problem is structural. A voluntary coordination layer becomes unsafe when exit is constrained and discretion is centralised over scarce capital. The answer is not outrage. It is to shorten failure domains, create alternatives, make records portable and give holders a collective means to resist mandate laundering.

In the long run, the common layer should be thin enough that the identity of the incumbent registry matters less. The ledger should preserve uniqueness, proof of control, transfer state, delegation state, security assertions, dispute metadata and auditability. The operator layer should decide facilities, customers, leasing, routing practice, financing and business model. Later changes should become real through adoption by operators and counterparties, not through institutional declaration alone. That is not anti-registry. It is pro-ledger.

For LACNIC-region data centres, such a model would be practical. It would make address inventory more financeable. It would help facilities support customer portability. It would reduce the discount attached to registry uncertainty. It would allow leased inventory to be documented cleanly. It would make disaster-recovery arrangements easier to trust. It would let small markets participate without accepting identifier lock-in as the price of modernisation.

The positive future is therefore not a registry that grows more ambitious as IPv4 becomes more valuable. It is a coordination society in which holders, operators and counterparties can rely on narrow, portable, auditable records while leaving commercial choices outside the common layer. The registry may remain useful as a service provider. It should not remain powerful because exit is impossible.

The next build-out needs a narrower book

The LACNIC region's data-centre build-out will not be decided only by IPv4. Power prices, renewable supply, cooling technology, fibre routes, cloud regions, cable landings, tax policy, equipment imports, urban planning, security, skills and capital markets will all matter. But addresses sit underneath many of these decisions because public reachability is the point at which infrastructure becomes customer service.

If addresses are treated as mere administrative entries, facility investors will miss a real constraint. If they are treated as political property of a region or community, the market will inherit a different problem: capital will be trapped, transactions will slow, customers will be locked in and records will become less truthful. The correct view is more sober. IPv4 addresses are scarce, valuable, operationally embedded identifiers. Their economic value comes from the services they enable and the continuity they preserve. The registry ledger should reflect that reality without trying to rule it.

A good data-centre market needs address supply that can meet varied demand: enterprise hosting, colocation, managed services, edge caches, mail, VPNs, payment endpoints, security appliances, disaster recovery and local public-sector services. It needs leasing and transfer channels because not every customer can buy inventory and not every holder uses all it controls. It needs customer onboarding that does not turn every address question into a constitutional negotiation. It needs portable records so that facilities compete on power, cooling, cross-connects, service and trust rather than on captivity.

The economics are straightforward. Scarcity makes addresses capital. Capital needs clear rights, low transaction costs, reliable records, predictable transfer, dispute isolation and exit. Where those elements are present, resources move toward higher-valued use and facilities can turn infrastructure into services. Where they are absent, addresses are discounted, customers delay, incumbents gain lock-in and the region pays a hidden tax.

The institutional conclusion follows. LACNIC's useful role, judged from the data-centre floor rather than the meeting room, is to keep the book narrow and trustworthy: uniqueness, holder and control records, contacts, transfer history, security assertions, dispute metadata, auditability and portability. It should not police whether a rack customer, hosting provider, edge cache, enterprise VPN platform or disaster-recovery site has the right moral claim to scarce IPv4. It should not convert regional service boundaries into title. It should not turn customer geography into a condition of record validity. It should not make the facility economy ask permission to be ordinary.

The customer who arrives at a colocation site wants racks, power, cooling, cross-connects, remote hands, upstreams and stable public addresses. It wants confidence that if the facility fails, the provider changes, the business is acquired or the country suffers a shock, its public identifiers can continue. That expectation is not an ideological demand. It is how businesses price continuity.

The region's next data-centre cycle will reward facilities that understand this. Address inventory is not a side cupboard of numbers. It is part of the capital stack that converts physical infrastructure into reachable services. The registry record above it should be as narrow, portable and boring as possible. In a scarce IPv4 economy, boring is not weakness. It is the condition for investment.

The ledger should not become the landlord. The book should not become the gate. The region needs data centres that can onboard customers, preserve continuity and compete on infrastructure quality. For that, it needs a registry layer that records reality, protects uniqueness and then gets out of the way.

Sources and further reading

These references provide the article's public doctrine and background context. They are used for institutional-economic framing, not for adopting any registry or official-sector narrative.