Summary
- ARIN's fee table looks orderly, but incidence analysis asks how annual tiers, transfer charges, legacy-service choices and participation costs move through smaller operators, buyers, customers and public networks after IPv4 exhaustion.
- The registry bill is rarely the line that first worries a small North American network.
The small invoice in a crowded budget
The registry bill is rarely the line that first worries a small North American network. In a regional hoster's quarterly budget, transit, power, colocation, equipment leases, backbone upgrades, security tooling, support staff, insurance, software licences and customer churn usually speak louder. A rural ISP may be deciding whether to replace access gear before winter. A Caribbean provider may be balancing backhaul, storm resilience and foreign-exchange exposure. A university network may be waiting for a procurement cycle. A municipal broadband project may be trying to keep rates politically acceptable while its costs rise faster than its subscriber base.
Then the ARIN invoice arrives. In absolute dollars it may look modest beside routers, fibre construction or a data-centre contract. Even a tiered annual fee can seem small when compared with the market price of IPv4 address space. For many operators, the easy accounting response is to file it under registry overhead: necessary, irritating, but not decisive.
That is too narrow. The line on the invoice is only the visible charge. Around it sits a bundle of economic conditions: staff time to understand the fee category, payment timing, account standing, future increases, eligibility for transfers, outstanding-fee clearance, agreement status, legacy-resource choices, access to routing-security services, documentation rounds, and the fact that the same ARIN-administered resources cannot simply be moved to a competing North American registry. The invoice is small because it is printed on one page. The relationship behind it is not small.
Fee incidence is the missing test. The legal payer is the organisation named on the ARIN account. The economic bearer may be someone else. A hoster may pass the cost into monthly service prices. A small ISP may absorb it by delaying redundancy. A university may pay through lower network-upgrade headroom. A buyer of a small IPv4 block may pay through escrow timing, transfer charges, counsel and staff hours. A legacy holder may pay through the choice between staying outside an agreement and entering a service perimeter that changes future fees. Downstream customers may never see ARIN's name and still pay through higher prices, slower onboarding, tighter address rationing or carrier-grade NAT complexity.
The question is not whether ARIN needs money. A registry must maintain accurate records, secure account systems, publish Whois and RDAP data, support reverse DNS, operate routing-security services, process transfer requests, answer support tickets, protect continuity and manage corporate obligations. Those functions cost money. A serious incidence analysis begins after that concession. It asks whether the charges are tied to registry work, whether they fall in proportion to ability to bear them, whether they embed scarcity rents, whether they finance broader institutional ambitions, and whether formally equal terms impose unequal burdens on networks of different size, margin and bargaining power.
The North American setting makes the question sharper, not softer. ARIN is not a registry in visible crisis. It is mature, documented and central to a wealthy region that includes the United States, Canada, parts of the Caribbean and North Atlantic, and many of the world's largest cloud, telecom, enterprise, university, security and hosting networks. Precisely because the machinery is orderly, the distributional issue cannot be dismissed as an emergency side effect. A stable registry can still create regressive burdens if its costs, fees and procedures are spread through a captive recognition layer.
After IPv4 exhaustion, ARIN's invoice became more than a membership or service charge. It became a quasi-utility charge attached to recognition in a scarce-address economy. That does not make every fee abusive. It does mean every fee needs a better explanation than the fact that a schedule was published and approved.
Incidence asks who pays after the bill is sent
Fee incidence is the difference between the name on a bill and the place where the cost finally lands. A tax may be collected from a shop but partly paid by customers through higher prices. A payroll charge may be written by an employer but borne by workers through lower wages or by shareholders through lower margins. A port fee may be paid by a shipper but appear later in the cost of goods. The legal payer is a starting point, not the end of the analysis.
For a regional Internet registry, incidence has several channels. The obvious channel is cash: the holder pays an annual fee or a transfer processing charge. The less obvious channels are absorption, pass-through, capitalisation, delay and administrative labour. A large carrier may absorb the fee in general overhead. A hoster may pass it into service prices. A buyer may capitalise it into the all-in cost of acquiring IPv4 space. A seller may discount a block if transfer charges, outstanding fees or documentation uncertainty make the deal harder. A small operator may pay through staff hours that would otherwise be spent on customers, security or network work.
A registry fee also moves through risk. If fee currentness is required before a transfer can be evaluated or completed, a billing issue becomes a settlement condition. If annual fees rise under a capped schedule, a customer with thin margins may experience the same percentage increase more sharply than a large incumbent. If a legacy holder must choose whether to sign an agreement to access RPKI or routing-registry services, the fee question becomes part of a larger bargain over service access and future exposure. If a small buyer must pay a processing charge, prepare recipient documentation and carry financing while waiting for approval, the effective fee includes time.
This is why incidence cannot be reduced to whether the schedule is tiered. A tiered table can be progressive in appearance and regressive in effect. If the largest holders pay more in dollars but much less relative to address value, revenue base, customer count, compliance capacity and market power, the burden may still fall more heavily on smaller networks. A $275 annual charge can matter more to a tiny provider than a six-figure annual charge matters to an address-rich platform. The arithmetic of burden is not the same as the arithmetic of the table.
The same applies to transfer fees. A $500 source request charge may look low beside a large IPv4 transaction. It may matter materially in a small-block sale where legal fees, broker spread, escrow fees, ARIN documentation, staff time and financing cost are already large relative to the block. Recipient processing charges scaled by transfer size may be efficient to administer, but the economic question is whether the scale tracks work performed or the value of the scarce address resource being moved. A charge that follows operational complexity is easier to defend than a charge that quietly behaves like a levy on scarcity.
Incidence also includes who pays for attention. Reading fee consultations, modelling tier changes, understanding outstanding-fee rules, monitoring annual increases, following changes to legacy treatment, predicting transfer costs and raising concerns in time all require specialised attention. Large networks can spread that attention across legal, finance, policy and registry teams. Small networks often assign it to one overloaded engineer, founder or operations manager. A policy that is open to all can still be used mainly by those with capacity to read it.
The incidence question is therefore concrete: when ARIN raises, collects or conditions a fee, where does the cost go? Does it stay with an address-rich holder, pass to customers, reduce investment, delay a transfer, capitalise into market price, induce agreement signing, consume scarce staff time or make small networks more dependent on intermediaries? Without that map, a fee table is not enough evidence of fairness.
ARIN is closer to a settlement utility than an ordinary vendor
The economic weight of ARIN fees comes from the nature of the registry relationship. A dissatisfied buyer can switch many vendors. Transit can be renegotiated. A data-centre contract can be moved, painfully but plausibly. Software can be replaced. Consultants can be fired. Even trade associations and industry forums usually depend on voluntary attachment; leaving may be costly, but it does not normally make an operator's address records less recognisable to the Internet.
ARIN is different. For resources administered in its region, ARIN is the shared recognition layer. A holder cannot take the same resources to another North American registry and ask for a competing authoritative record. A buyer in a transfer cannot obtain full registry finality merely because a private contract is signed. A legacy holder cannot get ARIN-region reverse-DNS authority, public registration records, ARIN-linked routing-security services or ARIN policy treatment from a substitute supplier. Exit is limited because uniqueness itself depends on a common reference.
That gives ARIN utility-like features even though its legal form is not that of a public utility. It operates a critical settlement layer for number-resource recognition. It records which organisation is associated with which resources, processes changes, maintains public data, supports related services and applies policy conditions. Its daily work is technical and administrative. Its market position is closer to a monopoly-like ledger.
The utility analogy should be used carefully. ARIN does not sell electricity, own the networks that route packets or guarantee reachability. Operators still make routing decisions, sign customer contracts, buy transit and manage their own infrastructure. Yet ARIN's record sits underneath a great deal of reliance. A buyer wants the record updated. A lender wants the address capacity in a diligence file to be credible. A hoster wants reverse DNS and abuse contacts to remain coherent. A security team wants routing-origin data to match operational reality. A legacy holder wants historical recognition not to become a discretionary favour. A small ISP wants its account standing not to become a sudden barrier to transfer, support or service access.
Because exit is limited, a fee has a different moral status from an ordinary price. A normal vendor can say: if the service is too expensive, leave. A settlement utility must say something harder: here is the narrow work the charge funds, here is why the charge is proportionate, here is how the burden falls across unequal users, and here is how affected parties can contest the design before it becomes unavoidable. The more captive the relationship, the stronger the explanation must be.
IPv4 exhaustion tightened the captive relationship. When ARIN still had a meaningful free pool, the fee relationship sat beside allocation policy. After depletion, the stock of existing registrations became the durable base. Address holders, transfer participants and service users fund a registry that now administers records around scarce, transferable and operationally embedded resources. Scarcity did not make ARIN less important. It made its recognition function more valuable and its fees more distributional.
The distinction between a vendor and settlement utility also changes the test for cross-subsidy. A voluntary association may fund conferences, outreach, public programmes and broad community work from member dues because members can decide whether that association is worth supporting. A monopoly-like registry should separate essential ledger costs from broader institutional programmes more clearly. If fees are compulsory in practice, they should first fund the narrow functions that captive users cannot obtain elsewhere: record accuracy, security, publication, support, transfer processing and continuity. The farther a charge moves from that core, the stronger the incidence concern becomes.
ARIN's public orderliness is valuable. Its materials disclose fee schedules, transfer categories, legacy distinctions, membership mechanics and many operational facts. But disclosure does not end the utility question. The relevant test is whether the payer can understand the cost standard behind the invoice and whether the burden matches the service, not merely whether the invoice was visible.
A tiered schedule can still be regressive
ARIN's 2026 Registration Services Plan is tiered. That is the first fact to keep in view. The schedule moves from 3X-Small at $275, 2X-Small at $550, X-Small at $1,100, Small at $2,205, Medium at $4,410, Large at $8,820, X-Large at $17,640, 2X-Large at $35,280, 3X-Large at $70,560, 4X-Large at $141,120 and 5X-Large at $282,240. The scale is not a flat poll tax. Larger resource holders pay more in absolute dollars.
That structure has a fairness intuition behind it. A larger holder uses more of the registry's resource-recording system, has more at stake in public recognition and may have a larger customer or business base over which to spread registry costs. A small holder should not pay the same amount as a large platform with a much larger address portfolio. Tiering recognises that obvious difference.
The problem is that regressivity is not measured only by whether the largest line is larger than the smallest line. It is measured by burden relative to ability to absorb the charge, the value received, the bargaining power available and the alternatives open to the payer. On those measures, a tiered registry schedule can still be regressive.
A 5X-Large holder paying $282,240 may be a major carrier, cloud platform, content network, enterprise portfolio or address-rich institution. Its fee may be large as an invoice but small relative to address value, revenue, staff capacity, financing access and the cost of a single commercial dispute. It can maintain specialised registry staff. It can hire counsel. It can model fee changes. It can participate in consultations. It can spread the charge over many customers, products or internal units. It may also benefit from scarcity because a large address base gives it customer flexibility and strategic optionality.
A 3X-Small holder paying $275 may be a small network whose address use is modest but mission-critical. It may serve a rural market with low customer density, a specialist hosting niche, a school network, a municipal broadband deployment, a small Caribbean island, a regional enterprise or a community service. The dollar amount is lower, but the fixed administrative relationship can be more burdensome. The operator may lack dedicated registry staff. Its margins may be thin. Its customers may be price sensitive. A single missed notice, stale contact or misunderstood fee rule can take more organisational effort to fix than the invoice itself.
The tier boundary is another incidence point. A network growing across a threshold may face a step change while still lacking the scale advantages of incumbents. Growth can be punished before it is fully monetised. A small hoster adding customers may need address capacity, security tooling, support and billing capacity at the same time. If registry fees rise as address holdings grow, and if transfer-market acquisition costs are also high, the combined burden can make the middle of the market more difficult than either the tiny edge or the giant incumbent tier.
The address-rich incumbent sees the same table differently. It may treat registry fees as a carrying cost on a valuable stock. If the annual charge is low relative to the market value of IPv4, the holder can retain optionality cheaply. It may sell later, lease capacity, support customer growth, use addresses defensively or keep them as strategic inventory. A fee table that charges large holders more can still allow incumbents to carry scarcity advantage at a cost far below the value of that advantage.
That is the core regressivity tension in post-exhaustion registry fees. A low registry charge relative to address value benefits address-rich holders. A high charge relative to margin hurts smaller or address-poor operators. A purely service-cost model may be fair if it funds narrow registry work. A value-based model risks becoming rent on a scarcity asset that the registry records but did not create. ARIN must be clear which theory it is using.
The tiered schedule also interacts with IPv6 and dual-stack reality. An operator may be deploying IPv6 while still needing IPv4 for customers, legacy applications, payment systems, mail reputation, VPN access, law-enforcement traceability, enterprise contracts and cloud interoperability. The fee burden cannot be dismissed by saying IPv6 is the future. For many networks, dual-stack operation means paying the cost of transition while still needing the scarce IPv4 layer. The smaller the network, the harder it is to carry both.
A tiered schedule is therefore a necessary but insufficient fairness device. It answers the easy question: do larger holders pay more? It does not answer the harder ones: do smaller operators pay more relative to margin and capacity, do large incumbents pay too little relative to scarcity advantage, do tier jumps discourage growth, and do annual fees fund registry work rather than institutional expansion? Those are incidence questions, and they require more evidence than a table.
The increase cap constrains the slope, not the burden
ARIN's fee materials include an annual-increase constraint: Registration Services Plan fees may increase by no more than 5% as approved by the Board. A cap of that kind matters. It prevents sudden uncapped increases under ordinary fee-setting mechanics. It gives payers a planning signal. It also makes fee growth a governance decision rather than an entirely automatic staff action.
But a cap is not a cost standard. It limits the slope of the fee path; it does not prove that the starting base is right, that the burden is well distributed, or that the cost drivers are legitimate. A 5% increase on a necessary monopoly-like service can still compound into a meaningful burden. For a large holder, compounding may be tolerable. For a small operator with stagnant customer revenue, increased power costs and hardware inflation, the same percentage change can arrive as one more fixed charge that cannot be postponed.
Board approval also solves only part of the accountability problem. It identifies the route through which fee policy becomes binding. It does not by itself show incidence analysis. Trustees may approve an increase because operating costs rise, reserves need rebuilding, staff costs increase, systems need investment, security services expand, legal expenses grow, or broad programmes continue. Each reason has a different distributional meaning. A fee increase for RDAP reliability, account security and transfer processing has one incidence story. A fee increase for institutional expansion, travel-heavy engagement, broad communications or legal posture has another.
The cap can even obscure the base question by making every increase sound moderate. A 5% ceiling has the language of restraint. Yet the economic issue is not only whether a single annual adjustment is moderate. It is whether the mandatory fee base is funding only functions that captive users should be required to fund. A small operator may accept a fee increase more readily if it can see that the money protects registry uptime, reduces transfer delays, strengthens fraud controls and improves support. It may push back if the same increase appears to finance broader programmes whose benefits are diffuse, reputational or more useful to repeat participants than to ordinary service-dependent networks.
An incidence-aware fee increase would therefore include a public explanation that separates cost drivers. How much comes from core record systems? How much from routing-security operations? How much from support capacity? How much from transfer processing? How much from corporate overhead? How much from governance meetings, outreach or communications? How much from legal work? How much from reserve targets? Which tiers carry the increase? Which operators are likely to move tiers? What small-operator burden analysis was done? What alternatives were considered?
Without that separation, fee policy becomes a single number attached to a story of organisational need. That is not enough for a captive registry service. Members and service users need to know whether they are paying for uniqueness and continuity or for the expansion of the institution that controls the unique record.
The cap also interacts with legacy expectations. A holder that entered a pre-2024 Legacy Registration Services Agreement under a capped-fee bargain faces one cost path. A later signer faces another. A transfer recipient of legacy resources may face the ordinary RSP structure. A large incumbent with historical holdings may experience fee growth as a manageable carrying cost. A smaller holder deciding whether to enter an agreement for service access may experience future fee uncertainty as part of the price of modernisation.
A capped increase may therefore be equitable in aggregate and still uneven at the edges. The correct question is not whether any fee increase is allowed. It is whether ARIN can show that the increase tracks registry work and that the burden on smaller, later-entering, address-poor or service-dependent networks has been considered before the increase becomes operational fact.
Transfer fees move through buyers, sellers and timing risk
Transfers reveal the difference between a charge for record work and a toll on scarcity. ARIN's 2026 fee mechanics include source-side request charges for major transfer paths and recipient-side processing charges scaled by aggregate IPv4 transfer size. Local materials record $500 non-refundable source fees for merger, acquisition and reorganisation transfers, specified-recipient transfers within the ARIN region and inter-RIR transfers out of the region. Recipient processing charges begin at $187.50 for a /24, rise to $375 for larger than /24 through /22, $750 for larger than /22 through /20, and continue upward by size to much larger charges for very large blocks. The schedule also records that outstanding annual fees must be paid before evaluation or completion.
Some transfer fee is defensible. Transfers require work. ARIN must verify the current registered holder, source authority, points of contact, corporate history, transaction documents, agreement status, recipient qualification, dispute posture and service-transition concerns. Staff must prevent forged transfers, stale-contact abuse, double claims and ambiguous record updates. A registry that performs no diligence would make the market less safe.
The incidence question is where the fee and the process cost finally land. The source may legally pay one fee. The recipient may legally pay another. The seller may accept a lower price if the source-side path is complicated. The buyer may pay higher counsel fees, broker fees, escrow costs or financing charges. A lender may require holdbacks until recognition is complete. A small buyer may lose a customer because a documentation round takes too long. A broker may earn more because the registry process is hard to navigate. Downstream customers may pay through slower onboarding or address rationing.
Fee currentness is a particularly strong incidence channel. Requiring payment of outstanding annual fees before evaluation or completion protects ARIN's revenue base and prevents parties from using a transfer to escape unpaid obligations. That is a legitimate institutional interest. It can also turn billing into a settlement gate. If a seller has neglected fees, the buyer may inherit delay. If the unpaid amount is small relative to deal value, the parties may treat it as a closing nuisance. If the block is small or the parties are thinly staffed, the same issue can derail timing. The economic burden may be paid by the buyer even if the legal arrear belongs to the seller.
Recipient qualification adds another layer. A specified-recipient transfer is not only a record update between willing parties. The recipient must satisfy ARIN policy. Needs-based logic can prevent abuse, sham demand and pure warehousing, but it also asks the registry to evaluate demand that the market is already pricing. Buyers with specialist staff can prepare better evidence. Large incumbents can present growth plans in the vocabulary the system expects. Smaller operators may have real demand but weaker documentation capacity. The result is a transfer market in which ability to satisfy process can matter alongside operational need.
The small-block market is especially exposed. A /24 buyer may need a modest amount of space for hosting, VPN services, customer growth, mail operations or network independence. The nominal recipient fee may look small. The total burden may not be. The buyer must find a seller, assess address reputation, pay a broker or adviser, prepare documentation, clear payment timing, manage RPKI and reverse-DNS transition, and coordinate customers. Fixed process costs consume a larger share of a /24 than of a /16. A fee schedule that scales by block size recognises some of that difference, but staff time and counsel do not scale down as neatly.
For large transfers, the incidence shifts. A higher processing charge may be small relative to deal value, but a large buyer may incorporate it into the capital cost of the block. Sellers may price expected registry friction into negotiations. Escrow providers may shape release mechanics around ARIN milestones. The registry fee is not the largest number, but it becomes part of settlement architecture. If the charge reflects registry work, it is a transaction cost. If it reflects the value of scarce addresses, it is a toll.
The distinction between anti-fraud work and quasi-tolling matters. Anti-fraud work verifies authority, prevents theft, preserves record accuracy and reduces market uncertainty. Quasi-tolling uses a necessary record update as an opportunity to extract value from the transaction because the parties cannot settle elsewhere. ARIN should want its transfer fees to be understood as the first, not the second. That requires a cost map: staff time, systems cost, review complexity, denial and withdrawal rates, documentation rounds, fee-related holds and service-transition work by transfer type and size band.
The market does not need private sale prices published. It does need enough evidence to know whether transfer charges track work performed. In a scarce-address economy, silence on that distinction invites suspicion that the registry is collecting not only the cost of settlement but a share of the scarcity it administers.
Legacy treatment creates several cost curves
Legacy resources make ARIN fee incidence unusually complex because not all holders face the same bargain. An organisation with historical resources outside a modern agreement, a pre-2024 LRSA holder, a later signer, a transfer recipient and an address-rich incumbent all sit on different cost curves.
The historical starting point is distinctive. Early North American resources were allocated before ARIN's modern contract structure. ARIN took on administration of many of those records when it was formed. Public legacy guidance records that legacy holders not under an ARIN agreement can maintain unique registration in Whois and RDAP, update public data, manage reverse DNS, maintain records through ARIN Online and access DNSSEC. It also records that RPKI and Internet Routing Registry services require resources to be covered under an ARIN agreement.
That service boundary is an incidence device. A legacy holder outside an agreement may pay no ordinary RSP fee for those resources and may value the historical independence of that position. But as routing security and routing-registry hygiene become more important in ordinary network operation, staying outside the agreement perimeter can impose a service cost. The holder may need external arrangements, manual workarounds or a different risk posture. If customers, counterparties or buyers increasingly expect RPKI and clean routing-registry data, the practical price of remaining outside the agreement rises even if the invoice does not.
The pre-2024 LRSA fee-cap distinction adds another curve. Local materials record that the legacy fee cap expired at the end of 2023, while organisations with an active LRSA entered before January 1, 2024 continue to have fees limited for legacy resources already covered before that date. Legacy resources covered under agreement after that point are subject to ordinary Registration Services Plan fees. That distinction is factual and important. It creates different incentives for older signers, later signers and holders still deciding whether to enter the agreement perimeter.
For a pre-2024 LRSA holder, the capped fee may look like a settled transitional bargain. For a later signer, the cost of modern service access includes ordinary RSP exposure. For a transfer recipient, legacy history may not deliver the same fee treatment. For an address-rich incumbent, the annual fee may be manageable relative to resource value. For a smaller legacy holder, especially a university, enterprise, public body or older network with limited staff, the decision to sign may feel like trading historical comfort for clearer services but more fee exposure.
None of this requires portraying legacy holders as victims or free riders. The economics are more practical. A registry inherited records that now support scarce, valuable resources. It must maintain accuracy, prevent hijacking and support modern services. Holders inherited positions under older arrangements. They need predictability. The fee system must bridge the two without making service access feel like coercive migration and without making historical status a permanent exemption from reasonable registry costs.
The incidence risk arises when ARIN does not separate cost from leverage. If RPKI and routing-registry services require an agreement because of liability, authentication, security and operational cost, the explanation should be explicit. If later signers pay ordinary RSP fees because capped treatment no longer covers the cost of modern service delivery, the cost logic should be visible. If transfer recipients face a different fee curve because a resource is entering ordinary market circulation, that transition should be described in economic terms rather than left as a technical contract consequence.
Legacy resources also interact with transfer prices. A buyer may discount a block if legacy status changes, agreement status is unclear, old contacts are stale or service access requires post-closing work. A seller may hesitate to transfer because the transaction changes future fee treatment or service status. A broker may earn a premium from navigating the path. The registry fee may be small compared with price, but the fee boundary shapes liquidity.
For public-sector and university legacy holders, the issue can be especially delicate. Some institutions hold older space because they were early network builders. Their internal budgets, procurement rules and public accountability may not fit a commercial transfer model. They may need modern security services but lack a commercial revenue base over which to spread new registry costs. A fee design that looks modest to a data-centre operator may create a different burden for a public university or agency.
The constructive rule is straightforward: legacy fee policy should be predictable, service-cost based and candid about transitions. Historical recognition should not become an excuse for stale records or weak security. Modern service requirements should not become a hidden method of extracting scarcity value. The fee curve should tell holders what part pays for records, what part pays for security and support, what part reflects agreement-related obligations, and what part, if any, reflects broader institutional cost.
Downstream pass-through is hidden but real
Most people who bear registry costs never receive an ARIN invoice. They are broadband customers, hosting customers, enterprise tenants, SaaS users, students, hospital staff, municipal residents, public agencies, small businesses, content publishers and developers whose service depends on networks that rely on ARIN-administered resources. They encounter the cost only after it has been translated into price, service quality, delay or constraint.
A small ISP can pass annual registry charges and address-acquisition friction into monthly rates if the local market allows it. In low-density or competitive markets, full pass-through may be impossible. Then the incidence moves into margin, deferred upgrades, slower customer activation, less redundancy, more NAT, fewer public addresses per customer, limited static-address options, weaker abuse tooling or reduced support capacity. The customer sees a service package. The registry cost is embedded.
Hosting firms and regional data centres see a different pass-through. IPv4 addresses remain part of the customer promise. Some customers need dedicated addresses for mail reputation, SSL compatibility, VPNs, legacy applications, payment systems, security appliances or regulatory separation. When addresses are expensive and registry processes consume time, hosters ration capacity, charge setup fees, push customers into shared-address models or prioritise larger customers. A small customer may not know ARIN exists. It still pays through product design.
Cloud users and enterprise customers may bear the burden through platform pricing and architecture. Large platforms can absorb registry fees easily, but address scarcity and transfer friction still influence how they price public IPv4, NAT gateways, load balancers, static addresses and outbound connectivity. When large providers charge more for scarce IPv4 features, some of the cost reflects market scarcity, not ARIN fees. But registry settlement and carrying costs are part of the environment that makes IPv4 capacity a priced feature rather than a neutral background input.
Universities and research networks face incidence through public mission trade-offs. A network budget that pays registry fees, counsel for address questions or staff time for legacy service decisions has less room for wireless upgrades, security monitoring, student services, research connectivity or redundancy. The invoice may be small, but the administrative burden can be high because public institutions move through procurement rules and internal governance. A commercial operator may pass the cost to customers. A university may pass it into slower improvement.
Public-sector networks and municipal projects face similar trade-offs. A city broadband network or public agency cannot always raise prices. It may absorb registry and address-market costs through general budgets. The public ultimately pays through taxes, reduced services or delayed infrastructure. When registry costs are tied to essential recognition and address access, they become part of the price of public digital services.
NAT complexity is another downstream channel. Address-poor operators often use carrier-grade NAT, shared addressing, complex logging and customer restrictions because public IPv4 is scarce and expensive. ARIN fees are not the sole cause of NAT. IPv4 scarcity is the larger driver. But if registry processes, transfer charges and agreement boundaries make acquiring or maintaining address space harder for small networks, they reinforce the move toward architectures that shift support and traceability costs downstream. Customers pay through troubleshooting, application limits, gaming issues, law-enforcement logging burdens and reduced network transparency.
The downstream burden is hardest to see because it is not labelled. A customer bill does not say: ARIN fee incidence, transfer delay premium, legacy service uncertainty, registry attention cost. It says broadband, hosting, cloud, support, static IP, setup fee or enterprise connectivity. That invisibility is why incidence analysis matters. It forces the registry to consider not only the account holder but the chain of users behind the account.
An incidence-aware ARIN fee model would not try to micromanage every downstream price. It would ask whether its own charges and procedures increase avoidable downstream burden. Are small-block buyers paying too much process cost relative to value? Are transfer fees making small operators more dependent on leasing or NAT? Are annual increases harder for public and low-margin networks? Are service boundaries for legacy resources delaying routing-security adoption? Are fee-currentness rules designed with proportional cure paths? These questions belong in fee policy because fees do not stop at the invoice.
Participation cost is part of the bill
Fee accountability is often described through formal governance: Board approval, membership rights, consultations, elections and public materials. Those mechanisms matter. They are not free to use. Participation itself has incidence.
A fee consultation asks organisations to read materials, understand current charges, model future tiers, compare alternatives, estimate their own address growth, assess transfer plans, decide whether to comment and submit arguments in the right forum before the decision hardens. A large carrier or cloud platform can assign that work to staff. A small ISP may handle it after customer tickets. A university may need internal approval before taking a position. A public-sector network may not have a policy specialist. A small hoster may not know a fee change matters until the invoice arrives.
This is not a minor cost. In a specialised registry, vocabulary itself is a barrier. RSP tiers, transfer paths, LRSA distinctions, fee caps, agreement coverage, outstanding-fee requirements, source and recipient charges, waiting-list interactions, routing-security services and member categories are not ordinary small-business language. Understanding the fee relationship requires institutional literacy. Those who have it pay less attention cost. Those who lack it pay more.
Participation cost also affects whose concerns become visible. A large address-rich holder has reason to monitor fees because the absolute dollars are high and the stakes around legacy, transfers and services are significant. A small address-poor entrant may be more burdened relative to margin but less able to participate. Silence from small networks therefore cannot be treated as consent. It may be rational inattention, overload or lack of procedural fluency.
The same problem appears in membership accountability. Service-dependent organisations may not all be General Members with voting rights. Even those eligible to participate must maintain account status, contacts, deadlines and internal ownership. A fee issue may be decided by the visible subset that has time, vocabulary and governance habit. That visible subset is real, but it is not the entire affected economy.
Participation burden becomes regressive when the cost of contesting a fee is fixed. Drafting a useful comment takes roughly similar institutional effort whether the operator has 500 customers or 5 million. The large operator can spread the cost. The small operator cannot. A formally open consultation can therefore magnify repeat-player advantage. The more complex the fee model, the stronger the advantage.
ARIN can reduce that burden without weakening governance. It can publish plain-language incidence notes with every fee consultation. It can identify which tiers are affected, which organisations are likely to move tiers, how transfer fees map to work, what small-operator burden was considered, how legacy transitions are treated and what alternatives were rejected. It can provide calculators, examples and concise summaries separate from legal and policy detail. It can report the categories of commenters and whether small operators, public-sector networks, universities, Caribbean networks and hosters were heard.
The point is not to privilege the smallest payer automatically. A registry must remain financially sustainable. The point is to recognise that accountability is not only a right; it is a cost. A fee model that requires high participation effort to contest will naturally be shaped by those who can afford to pay that effort. If ARIN wants fee legitimacy, it should lower the cost of understanding and challenging fees before the schedule becomes an invoice.
Cross-subsidy should be tied to a narrow cost standard
Every fee schedule contains cross-subsidy. The question is whether the cross-subsidy is honest and defensible. A registry with fixed costs cannot charge every account exactly the cost of each ticket, database row, reverse-DNS delegation or support case. Some costs are shared: systems, security, staff, governance, publication, audit, continuity and disaster recovery. A tiered model spreads those costs across holders with different resource sizes. That is normal.
The problem begins when the cost standard becomes unclear. A narrow registry cost standard would fund ledger accuracy, account authority, public data, transfer processing, routing-security operations, reverse-DNS continuity, support, security, corporate compliance and enough governance to keep the institution accountable. A broader institutional standard may also fund education, conferences, fellowships, communications, global coordination, advocacy, legal strategy, reserve growth, expanded programmes and a larger organisational footprint. Some of those activities may be useful. Usefulness alone does not decide whether captive registry fees should fund them.
The difference matters because a monopoly-like charge should be minimised around non-substitutable functions. If a small ISP must pay ARIN to remain in good standing for resources it cannot move elsewhere, the strongest claim on that money is record and service continuity. If the same fee base funds broad programmes whose benefits accrue mainly to participants with time to attend, large organisations with policy staff or the institution's reputation, the regressivity concern grows.
A registry should not be starved. Underfunded records, weak security, slow transfers, bad support and poor continuity would hurt smaller operators first. The choice is not between a robust registry and no fees. It is between cost-based funding for core functions and ambiguous funding that lets institutional expansion hide inside necessary charges.
Transfer fees deserve the same standard. If the recipient fee is meant to recover staff work, systems use and risk review, ARIN should explain the cost drivers. Do larger blocks actually require more work, or do they mainly carry more market value? Do small transfers consume more staff time per address? How often do source-side requests fail? What share of staff time goes to authority verification, recipient qualification, fee clearance, agreements, inter-RIR coordination or routing-security guidance? A charge that matches work is defensible. A charge that rises because the block is more valuable needs a different justification.
Reserves and legal spending sit close to this question. Reserves can protect continuity, and legal capacity can protect the ledger. But both can also insulate the institution from payer discipline if poorly explained. If fee increases are tied to reserve restoration, members should know what kind of continuity risk is being insured. If fees support legal work, payers should know whether the work is routine corporate counsel, contract management, transfer disputes, enforcement, governance or broad institutional defence. Category reporting can preserve confidentiality while making incidence visible.
The cost standard should also distinguish scarcity value from registry value. IPv4 address space has market value because it is scarce, operationally useful and embedded in networks. ARIN's record helps make that value more reliable. It does not mean ARIN created the entire value. If fees are justified by the market value of recorded resources, the registry begins to act like a tax authority over capital. If fees are justified by the cost of maintaining the record and services around those resources, the registry remains closer to a bookkeeper.
That distinction is ideological only in the sense that all institutional cost standards are ideological. The practical rule is simple: a bookkeeper should charge for bookkeeping, security, settlement and continuity, not for the value of the houses whose titles it records. If the bookkeeper wants to fund broader civic work, it should show why captive payers must finance it and why the burden is fair.
ARIN can strengthen legitimacy by adopting an explicit fee-cost map. Every major fee category should be tied to functions, service metrics and cost drivers. Every broader programme funded from mandatory fees should be justified as common registry benefit or separated into voluntary, sponsored or opt-in support where feasible. Cross-subsidy is acceptable when it protects the ledger and users. It becomes suspect when it protects office convenience or institutional ambition.
What incidence-aware fee policy would disclose
A credible ARIN fee model would not merely publish a schedule. It would publish an incidence explanation. That explanation need not reveal confidential account data or private transfer prices. It should show enough for affected parties to see how the burden moves.
The first disclosure should be cost drivers. ARIN should separate the cost of core registration systems, public data services, routing-security operations, reverse DNS, transfer processing, support, account security, software development, corporate overhead, governance, legal categories, reserve targets and broader programmes. The point is not to invite line-by-line micromanagement. It is to show whether mandatory fees are tied to necessary registry work.
The second should be tier-change effects. Payers should know how many organisations sit near each boundary, how annual increases affect each tier, how many organisations moved tiers in recent years and which types of holders are most exposed to threshold jumps. A tiered table without movement data hides growth penalties and middle-market pressure.
The third should be small-operator burden. ARIN should model not only dollar amounts but administrative effort. How many 3X-Small, 2X-Small, X-Small and Small organisations participate in consultations? How often do small holders face transfer fees, outstanding-fee holds or documentation rounds? What support is available for organisations without specialist staff? How do Caribbean, rural, municipal, public-sector and university networks experience billing and service access? Aggregate analysis would make the fee debate less dominated by large-account assumptions.
The fourth should be transfer-fee rationale. For each transfer path and size band, ARIN should explain what work the charge recovers. It should publish aggregate processing times, documentation rounds, withdrawal categories, denial categories, fee-related holds, agreement execution delays and inter-RIR bottlenecks. If larger transfers cost more because they require more work or create higher risk, show the pattern. If small transfers are more expensive per address because fixed review work dominates, acknowledge that too.
The fifth should be legacy transition logic. Holders should understand exactly how pre-2024 LRSA fee-cap treatment differs from later agreement coverage, what services are available outside agreement, why RPKI and routing-registry access require agreement, how future fee exposure is calculated and how transfer recipients are treated. Legacy policy should not require a holder to infer economic consequences from scattered service and contract language.
The sixth should be service-cost mapping. If annual fees fund RPKI, RDAP, Whois, reverse DNS, support and account security, publish service metrics in ways that payers can use. Uptime, incident categories, support response, authority-recovery time, transfer completion, routing-security support and public-data reliability are part of the value purchased by the invoice.
The seventh should be hardship and payment data where safe. ARIN need not expose individual payers. It can publish aggregate late-payment patterns, payment-plan usage if any, fee-related transfer holds, account closures, cure outcomes and categories of support demand. Such data would show whether fee pressure is concentrated among smaller or particular classes of networks.
The eighth should be a clear statement of whether charges track registry work or address value. If ARIN believes size-based fees are the best proxy for service cost and risk, it should say so and support the claim. If the fees also reflect ability to pay or scarcity value, that should be debated openly. Hidden theories produce distrust.
The ninth should be a participation-cost report. Fee policy should show who commented, which groups were notified, what concerns were raised by small operators, what alternatives were considered, and why rejected options were rejected. A fee schedule approved after thin participation should not be presented as broad consent without that caveat.
None of these disclosures would prevent ARIN from charging fees. They would make the fee bargain more legible. They would also protect ARIN. A registry that can show cost drivers, burden analysis and service mapping is better defended against claims that it is taxing scarcity. A registry that asks users simply to trust the table invites the opposite inference.
The invoice question
The final question returns to the small invoice in the crowded budget. When ARIN sends a bill, what exactly is the recipient paying for?
Part of the answer should be easy. The recipient should be paying for uniqueness, accurate registration, public records, account authority, reverse-DNS continuity, routing-security services, transfer processing, support, security and institutional continuity. Those functions protect the Internet's numbering ledger and the users who rely on it. A fee for that work is not only legitimate; it is necessary.
Another part may also be defensible, but only with explanation. The recipient may be paying for governance, legal compliance, reserve targets, software renewal, data-centre resilience, outreach that improves participation, and policy systems that keep the registry accountable. These costs can support the core mission, but they need category clarity because they are easier to expand than a database or support queue.
The suspect part is the one that quietly prices a captive relationship. If a fee rises because the registry knows holders cannot leave, if a transfer charge tracks the market value of scarce addresses more than the cost of settlement, if service access pushes legacy holders into fee exposure without candid cost explanation, if small operators pay fixed attention costs that large incumbents can ignore, or if mandatory fees finance institutional ambition beyond the narrow ledger, the charge begins to look less like cost recovery and more like a regressive levy.
ARIN does not need crisis drama for this question to matter. Orderly institutions can distribute cost unfairly. Public schedules can hide burden. Tiered fees can still favour incumbents. Board-approved increases can still miss downstream incidence. Transfer charges can be legitimate anti-fraud costs or quiet tolls. Legacy distinctions can respect history or convert service dependence into leverage. Participation mechanisms can be open and still expensive to use.
The better model is narrower and stronger. Charge enough to keep the ledger accurate, secure, usable and continuous. Make transfer settlement predictable. Explain each fee by work, risk and service. Publish aggregate evidence of burden. Protect small operators from avoidable fixed process cost. Treat legacy transition as an economic bargain, not only a contract migration. Separate mandatory registry finance from optional institutional expansion wherever possible. Do not justify charges by the market value of resources merely because the registry records them.
When a small ISP, regional hoster, public network or university opens the ARIN invoice, the most important number may not be the dollar amount. It may be the explanation behind the amount. The payer should be able to see which part protects uniqueness and service continuity, which part pays for administration, which part funds broader institutional choices, and which part, if any, is charging for the privilege of having no practical exit.
That is the incidence test for ARIN after IPv4 exhaustion. The invoice names the payer. Legitimacy depends on whether ARIN can show who really bears the cost, why the burden is fair, and how much of the charge belongs to the ledger rather than to the power of the bookkeeper.