RIPE NCC is usually described in institutional language: a Regional Internet Registry, a membership association, a secretariat for the RIPE community and the operator of registry services for Europe, the Middle East and parts of Central Asia. The description is accurate as far as it goes. It is also incomplete. The more revealing description is fiscal. RIPE NCC is a club wrapped around a monopoly ledger for scarce number resources. It finances itself through member contributions, maintains a large reserve, runs a wider set of community and public-good services, and asks members to decide how the bill should be divided.
That turns a technical institution into a problem in political economy. The question is not whether RIPE NCC performs useful work. It plainly does. The question is how a member-funded monopoly ledger should behave once the resource beneath the ledger is scarce, transferable and commercially strategic. Fees, reserves and charging schemes are not back-office details. They decide who pays for the authoritative record, who subsidises broader activity, who benefits from financial resilience, who bears fixed costs, and how much discretion the institution can exercise before members impose discipline.
The public documents provide unusually concrete exhibits. The 2026 charging scheme keeps the annual contribution at EUR 1,800 per Local Internet Registry account, adds EUR 75 for each independent Internet number resource assignment and EUR 50 for each ASN assignment, and charges a EUR 1,000 sign-up fee. The 2026 Activity Plan and Budget expects income of EUR 41.140m and costs of EUR 41.125m, with a tiny operating result and an overall surplus once financial income is included. It budgets for 202.1 FTEs. The 2025 Financial Report records a Clearing House reserve of about EUR 33.6m at year end and reports a capital expense ratio of 86%.
None of those numbers is shocking by itself. A critical registry should not be run on a shoestring. Its database, RPKI service, reverse DNS, member portal, security systems, account support, audit controls, legal functions and meeting obligations all require professional capacity. The analytical mistake is to stop there. A registry is a club good with monopoly features. The service is non-rival within reasonable limits: one more accurate record does not consume the ledger for everyone else. But access to the recognised ledger is excludable and regionally exclusive. A holder of RIPE-region resources cannot simply choose a rival RIPE NCC for the same administrative function. The fee is therefore not an ordinary subscription. It is closer to a mandatory levy on participation in the recognised numbering system.
Lu Heng's public cost-structure notes press this point in deliberately sharp form. They argue that the narrow technical task is maintaining the registration database and core security services, while many registry costs have grown around meetings, training, outreach and institutional self-maintenance. His note on RIPE NCC's cost structure uses the 2024 budget to argue that core registry services accounted for roughly EUR 9.6m of a projected EUR 38.2m budget; even allowing for legal, HR, facilities and support, he argues, the essential functions could be run for far less than the full institutional budget. That argument comes from a market participant with strong views about registry power, and it should be read with that interest in mind. But the economic question it raises is not partisan: when a fee is effectively compulsory, should it finance only the ledger and its direct safeguards, or also the surrounding apparatus of community, governance, data platforms and institutional presence?
The question matters more because IPv4 exhaustion has changed what registry membership means. During the allocation era, a registry was a rationing body. It assessed need, issued scarce addresses from a pool, maintained records and preserved uniqueness. In the post-exhaustion era, the free-pool function has diminished. The continuing power lies in recognition: who appears in the registry, who can update records, who can transfer resources, who can issue routing-security assertions, who remains in good standing and whose counterparties can trust the public record. The invoice funds a ledger. The ledger supports capital.
This is where club economics becomes institutional politics. A golf club can charge dues and decide whether to build a clubhouse. A professional association can charge members and decide whether to hold conferences. A registry club can do those things too, but it sits above operational resources that networks, customers, lenders and buyers treat as valuable. If the club underprices large members, small members subsidise them. If it overprices small members, entry and survival become harder. If it builds reserves, it becomes more resilient and harder to discipline. If it uses the ledger fee to pay for a broad public role, it turns the registry bill into a cross-subsidy. If it relies on member voting, the voting incentives of heterogeneous members become the core governance problem.
RIPE NCC's financial model is not obviously abusive. Nor is it obviously neutral. It is a mature example of a harder institutional problem: how to fund a monopoly ledger without allowing the ledger operator to become a fiscal state in miniature.
A club with limited exit
Ordinary clubs have meaningful exit. If a tennis club raises dues to build an elaborate restaurant, members can resign and play elsewhere. If a trade association becomes too expensive or too political, firms can leave and form another one. Registry clubs are different. A network can leave RIPE NCC membership only by giving up, transferring, reorganising or otherwise changing the administrative relationship around resources on which operations may depend. Exit is possible in legal and operational theory. It is not comparable to switching vendors.
Limited exit changes the meaning of equality. RIPE NCC's long-standing model has been one LIR account, one annual fee. The 2026 fee is EUR 1,800 per LIR account. That has the elegance of club equality. Each account pays the same base contribution. A small new network and a large incumbent with a far bigger resource footprint face the same account charge, apart from separate fees for certain independent resources and ASNs. The model is easy to understand, easy to administer and resistant to accusations that RIPE NCC is taxing the asset value of IPv4 holdings. It also creates an incidence problem.
The burden of a fixed fee does not fall equally. EUR 1,800 is immaterial for a large carrier, cloud provider, hosting group or multinational enterprise. It is more meaningful for a small ISP, community network, regional hoster or operator in a low-income or high-risk jurisdiction. Even when two members receive the same invoice, the economic burden is not the same. A flat fee is equal at the invoice line and unequal at the balance sheet.
RIPE NCC's own 2025 Financial Report shows the scale and diversity of the club. It started 2025 with 20,991 active LIR accounts and ended with 20,647. It opened 874 new LIR accounts, including 67 from countries classified by its banks as Ultra High-Risk, and closed 1,218. In membership terms, it started the year with 19,993 members and ended with 19,863. This is not a small private association dividing costs among similar firms. It is a large infrastructure club whose fee design affects thousands of organisations in sharply different economies.
The service region magnifies the point. RIPE NCC covers rich digital economies, mature European telecoms, cloud infrastructure, national research networks, datacentres, Middle Eastern operators, networks in conflict-affected or sanctions-exposed environments, and small providers whose customers may have little ability to absorb extra cost. A uniform euro-denominated fee is administratively neat. It is not economically uniform.
The 2026 budget shows how central member fees are to the institution. It expects EUR 36.0m from service fees paid by existing members, EUR 540,000 from service fees paid by new members, EUR 3.075m from independent resource and ASN fees, and EUR 600,000 from sign-up fees. Member fees, broadly defined, account for EUR 39.64m of EUR 41.14m in total income. RIPE NCC is overwhelmingly dues-funded. Members are not merely users of a service. They are the tax base of the registry club.
That tax base funds more than the narrow act of recording number resources. It finances personnel, IT, legal review, community engagement, training, RIPE Atlas, RIPEstat, the LIR Portal, RPKI, the RIPE Chair team, external coordination and organisational sustainability. Some of these services are direct operational necessities. Some are public goods for the wider Internet. Some are club goods for active participants. Some are institutional presence. They are all bundled into the same fiscal relationship.
The problem is familiar from local government and homeowner associations. A club that funds many services through one compulsory charge creates winners and losers with every spending decision. Those who use the pool subsidise those who use the gym; those who want security subsidise those who want landscaping; those who never attend meetings subsidise those who do. The difference is that RIPE NCC's underlying asset layer is not a leisure amenity. It is the recognised address registry.
That is why the fee debate should not be reduced to whether EUR 1,800 is high or low. The better question is what theory of cost allocation justifies the invoice. Is the annual contribution a charge for the narrow registry function? A membership due for a technical community? A pooled contribution to regional Internet development? A levy on the value of resources recorded? An insurance premium for legal, geopolitical and operational shocks? RIPE NCC's charging scheme contains elements of all these ideas, but they point in different directions.
If the fee is for the ledger, cost should be as low and direct as possible. If it is for community development, redistribution and participation should be explicit. If it is for resilience, reserve targets and legal buffers should be separated and explained. If it is partly value-sensitive, the flat model looks underdifferentiated. Good fee design begins by choosing the theory. Otherwise the invoice becomes a compromise whose politics are hidden by administrative language.
Fixed fees and the politics of incidence
The 2027 charging-scheme vote made the incidence question visible. In May 2026, members were asked to choose between two models. Option A preserved the one LIR account-one fee design, with an annual fee of EUR 1,894, an increase of EUR 94 from 2026. Option B introduced a category model based on PA IPv4 and IPv6 resources held in each LIR account. The Board said both models aimed at the same EUR 42.5m income budget, based on a 3.3% inflation increase over the 2026 income budget, assumed 20,000 active LIR accounts, maintained current services, included IT investment, and contained a commitment to an overall cost reduction of 1.5%.
The Board recommended Option B. Its case was that differentiation responded to members who wanted a wider gap between the lowest and highest fees. Under the proposed category model, the base fee would have been EUR 500 for LIR accounts with no PA IPv4 or with a /29 or less of IPv6 PA space, while the largest current holder would have paid more than EUR 30,000. The Board said almost 75% of LIR accounts would pay less than under the existing model.
Members chose Option A. The vote was close enough to be institutionally revealing: 3,049 votes were cast; Option A received 1,547 votes, or 51.12%; Option B received 1,479, or 48.88%; 23 abstained. On a question about cost incidence, the membership split almost exactly in half.
That split is the economics of RIPE NCC in miniature. A flat model is defensible as a membership principle. It treats each LIR account as a unit of participation rather than as a proxy for resource wealth. It is simple and predictable. It avoids turning RIPE NCC into a resource-value tax authority. It prevents annual fees from becoming an indirect claim on the market value of IPv4 holdings. Large resource holders can reasonably say that the registry provides a service, not ownership, and that the cost of maintaining a record does not rise linearly with the number of addresses recorded.
The category model is defensible as burden allocation. It recognises that larger holders receive more economically significant recognition from the same registry. It reduces entry and survival costs for smaller accounts. It reduces the regressive effect of a flat fee. It also answers a common political objection: that under one-account fees, the smallest participants help finance a system whose private value is greatest for larger incumbents.
Neither model is obviously correct. The choice depends on what RIPE NCC thinks it is charging for. If the registry is charging for membership in a mutual association, one account-one fee has a civic logic. If it is charging for a service tied to the scale of resources administered, differentiation has a cost-allocation logic. If it is charging for the value of scarce resource recognition, differentiation should be stronger still. If it wants to avoid asset taxation, differentiation should be cautious and capped.
The vote also reveals a problem of incentive compatibility. Members with small or empty accounts have reason to favour a category model if it reduces their fee. Members with large PA holdings have reason to prefer a flat model. Multiple-LIR structures complicate matters because account boundaries do not always correspond neatly to organisational scale. Legacy and PI treatment add further complexity. A charging system can induce behaviour: opening accounts, closing accounts, consolidating accounts, distributing resources, resisting transfers, or lobbying for definitions that reduce exposure.
The Board's own explanations acknowledged complexity around multiple LIRs, PI and legacy resources. That is not a technical nuisance. It is the heart of the fiscal design. When fees depend on categories, members optimise categories. When fees depend on accounts, members optimise accounts. When separate charges apply to ASNs and independent resources, members examine resource packaging. Every charging scheme creates incentives at the margin.
This is why "fairness" is not enough. A fair-looking fee can be gamed. A simple fee can be regressive. A differentiated fee can overreach. A low base fee can increase dependence on high-end contributors, giving large holders more de facto veto power over future budgets. A high flat fee can discourage small entrants or push marginal operators into dependence on upstream providers. The question is not which group deserves relief in a moral sense. It is which design raises sufficient revenue while minimising distortions, preserving the narrow ledger function and keeping member consent robust.
The 2026 result says that consent is fragile. A 51.12% vote for the one-fee model is a lawful decision, not a settlement. The losing side was large enough to keep the issue alive. If costs keep rising, if reserves are contested, if small operators remain under pressure, or if large holders appear to benefit disproportionately from flat pricing, differentiated charging will return. The fiscal constitution of the club has not been resolved. It has been extended for another cycle.
What the invoice buys
The phrase "registry service" hides a large bundle. In the 2026 Activity Plan and Budget, the narrow "Registry" activity is budgeted at EUR 5.665m and 43.4 FTEs. It includes Registration Services at EUR 2.7m, Member Services at EUR 1.45m and Registry Monitoring at EUR 1.515m. RPKI is budgeted separately under Information Services, at EUR 1.16m and 7.1 FTEs. The LIR Portal is EUR 2.88m. The RIPE Database is EUR 700,000. DNS and K-root are EUR 840,000. RIPE Atlas is EUR 1.7m, RIPEstat EUR 550,000, RIS EUR 950,000 and IT Support EUR 4.02m.
One can draw the boundary of the core in several ways. A minimalist definition would include Registration Services, Member Services, Registry Monitoring, the RIPE Database, RPKI, reverse DNS and enough IT and security support to operate them reliably. A broader technical definition would include the LIR Portal, K-root, RIS, RIPEstat and RIPE Atlas, because they support measurement, transparency, routing analysis and operational usefulness. A community definition would include training, engagement, policy support, meetings and coordination. An institutional definition would include legal, finance, HR, facilities, information security, risk, compliance, the Managing Director's office and the RIPE Chair.
The budget funds all definitions at once. External Engagement and Community is EUR 9.8m. Community Building and Member Engagement is EUR 5.95m. Learning and Development is EUR 1.9m. Coordination and Collaboration is EUR 1.95m. Organisational Sustainability is EUR 11.945m, including Facilities at EUR 2.4m, HR at EUR 1.2m, Legal at EUR 1.3m, Finance at EUR 1.865m, Information Security, Risk and Compliance at EUR 2.8m, the Managing Director's office at EUR 2.0m and the RIPE Chair at EUR 380,000.
These figures do not prove waste. A registry with roughly 20,000 LIR accounts, sanctions exposure, legal complexity, public accountability, international coordination and a large meeting culture will incur overhead. A serious criticism should not pretend that a few servers and a spreadsheet could replace the institution tomorrow. The question is whether the mandatory fee should finance the full institutional bundle or only the functions members cannot obtain elsewhere.
This is the cross-subsidy problem. A small access provider may need accurate registry records, RPKI, reverse DNS, account support and predictable transfers. It may not value international policy coordination, RIPE meetings, community outreach, training events or measurement platforms at the same level. A large multinational may value RIPE Atlas, RIPEstat, policy engagement and government relations. A regulator may value training and regional coordination. A researcher may value open data. A member that never attends a meeting still pays for the meeting infrastructure.
In public-finance terms, the ledger is a club good with important public-good spillovers: everyone benefits from uniqueness and accurate records. RPKI has a security public-good element: a more secure routing environment benefits more than the certificate holder. Meetings and training are mixed goods. Community engagement can be a legitimacy good, but also an institutional-expansion good. RIPE Atlas and RIPEstat are valuable public data services, but their funding from mandatory registry fees is a policy choice, not a technical necessity.
Lu Heng's RIPE cost note takes the separation argument to its logical end: compulsory fees should fund core registry services and RPKI, while value-added services such as RIPE Atlas, RIPEstat, training and meetings should rely more on voluntary funding, sponsorships, donations or usage-based charges. That fiscal minimalism may understate the practical benefits of neutral shared measurement and community infrastructure. Yet it identifies the essential test: if an activity is valuable, can it attract explicit support from those who value it most? If it cannot, why should resource recognition be the mechanism that funds it?
RIPE NCC's defenders have a serious reply. Some services are valuable precisely because they are neutral, widely available and not dependent on sponsorship by large players. If RIPE Atlas or training were funded mainly by big operators, the service might become less independent or less evenly available. If meetings were funded only by attendance fees, smaller participants might be excluded. If policy support were stripped down, the legitimacy of registry decisions might weaken. Mandatory cross-subsidy can protect openness.
That reply strengthens the case for clarity rather than ending the debate. If the fee is being used as a fiscal instrument to sustain a particular kind of Internet community, members should be told that explicitly. They should be shown which public goods are subsidised, by whom, for whose benefit and with what evidence of effectiveness. "Cost recovery" is too narrow a description for a budget that pays for the ledger, the club, the public data platforms and the institution's voice.
The closing of the Community Projects Fund in 2025 is a useful signal because it shows that non-core initiatives can be reviewed and stopped when effectiveness or resource use is questioned. That discipline should not be exceptional. A mature registry should classify spending into at least three buckets: unavoidable ledger and security functions; supporting functions needed for reliable operation; and discretionary public-good or community activities. The first bucket has the strongest claim on mandatory fees. The third should face the strongest burden of proof.
Without that classification, every activity borrows legitimacy from the ledger. Every line supports trust; every meeting supports community; every overhead supports resilience; every expansion supports stability. The language may be partly true, but it is too elastic. Fiscal accountability requires narrower words.
Reserves as insurance and bargaining power
No responsible member should want RIPE NCC to be thinly capitalised. A registry that cannot absorb shocks is a risk to everyone. It must survive payment disruption, sanctions complications, legal surprises, market losses, security incidents, vendor failures, office moves, staff disruption, emergency technical work and periods of member anger. Weak reserves would make the ledger more fragile, not more accountable.
The question is what reserves do to incentives. RIPE NCC's 2025 Financial Report records a Clearing House reserve of EUR 33.6m as of 31 December 2025. Based on total expenses of EUR 38.952m, it reports a capital expense ratio of 86%, down from 91% in 2024. The 2026 budget projects reserves of EUR 34.427m at year end if there is a redistribution in 2026, against expenses of EUR 41.125m, or 84% of expenses. Without redistribution, projected capital would be EUR 36.831m, or 90% of expenses.
That is a substantial cushion. It is not excessive in the crude sense; many infrastructure bodies would like something close to a year's operating coverage. But a reserve near one year's total expenses changes bargaining behaviour. It gives the institution time. It can absorb a deficit. It can withstand delayed payments. It can finance legal advice. It can survive a difficult vote cycle. It can resist immediate pressure to cut. Resilience for the function is also autonomy for management.
The dual character should be explicit. Reserves are insurance for members when they preserve the ledger. They are institutional power when they allow the organisation to pursue contested strategies without immediate fiscal constraint. The same euro can keep RPKI running during a crisis or finance a legal posture that many members dislike. It can protect service continuity or soften the budget discipline that member dues are meant to impose.
RIPE NCC has real reasons to hold reserves. The 2025 Financial Report says it manages an uninvoiced exposure of EUR 5.2m relating to members in countries classified by its banks as Ultra High-Risk, covering part of 2021 and the full years 2022-2025. It says regular payment routes are unavailable for some members, anti-money-laundering requirements complicate collection, and the amount is now disclosed outside the balance sheet rather than recognised as an asset and liability. The report also says RIPE NCC continues to serve these members and that the legal obligation to pay remains.
That is a fiscal resilience problem with a geopolitical cause. Members in some countries can still depend on registry services even when payment channels fail. Other members effectively finance continuity during the delay. Bad debt is budgeted at EUR 375,000 in 2026, and the 2025 report says bad debt expense reached EUR 496,000, reflecting provisions for doubtful debts and payment extensions. A narrow commercial firm might cut off service faster. A critical registry cannot treat disconnection as an ordinary collection tool without risking the stability and legitimacy of the ledger.
This is a good reason to hold reserves. It is also a reason to ask who bears the burden. When banks classify countries as high-risk, the cost is not borne only by those countries. It is spread through the membership. When RIPE NCC chooses continued service, it protects Internet stability and its own public legitimacy. It also turns the fee base into a mutual insurance pool.
The Joint RIR Stability Fund adds another layer. The 2026 budget notes that RIPE NCC is committed to a maximum contribution of EUR 1m through that fund, coming out of reserves if activated, to mitigate unforeseen regional and global disruptions or threats and safeguard the stability of the RIR system. This may be prudent. The turmoil in another regional registry has shown why registry continuity can matter beyond one service region. But the incentive issue remains: RIPE NCC members may be asked, through reserves, to insure not only their own ledger but the wider RIR system.
The insurance logic should be made more granular. Members can reasonably support a continuity reserve for core services: registry records, RDAP, WHOIS, reverse DNS, RPKI, security operations, data escrow, disaster recovery and essential staff. They may also support a payment-risk reserve for cases where service should continue despite blocked payment routes. They may support a system-stability reserve for coordinated RIR continuity. They may support a legal-contingency reserve. Each purpose has different legitimacy. A single Clearing House reserve does not show which risk is being insured.
The interaction with redistribution makes this more than accounting. RIPE NCC's charging scheme says members vote each year at the General Meeting on returning excess paid fees or shortage of fees through redistribution to the membership. In 2025, the operational surplus before redistribution was EUR 2.8m, higher than the budgeted surplus of EUR 1.1m, and the fiscal surplus was redistributed following the General Meeting decision. After redistribution and the financial result, EUR 622,000 was added to reserves.
The mechanism is a useful check. It prevents the association from automatically accumulating every surplus. But it also creates a recurring tension between fee relief and reserve strengthening. Members who prefer lower invoices vote for redistribution. Members who value resilience vote for accumulation. Management may prefer stability. Small members may prefer cash relief. Large members may be less sensitive to the same amount. The reserve level is therefore not just a financial variable. It is a political settlement.
That settlement should not depend on annual intuition alone. A reserve target expressed as a percentage of core operating expense, rather than only total institutional expense, would sharpen the debate. If members are told reserves are 86% of total expenses, they should also know how many months of narrow ledger-continuity costs are covered. If the core registry plus RPKI, database, DNS and essential IT costs far less than the full budget, reserves may cover multiple years of core continuity even while covering less than one year of the whole institution. That distinction is central to the difference between protecting the ledger and protecting the full institutional bundle.
Legal contingency and the price of discretion
Legal spending is small in percentage terms and large in institutional meaning. The 2026 budget sets Legal at EUR 1.3m and 6.0 FTEs, up from a 2025 budget of EUR 1.2m and 5.0 FTEs. It says RIPE NCC aims to maintain a robust, consistent and compliant legal framework, review new legislation, support policy proposals, support projects and defend the accountability of the self-regulatory system and existing Internet governance structures. Apart from staff costs, the main expense is legal consultancy.
None of that is strange. RIPE NCC operates under Dutch law, serves members across many jurisdictions, deals with sanctions, banking risk, data protection, contracts, number-resource policies, disputes, corporate governance, RPKI assurance, a Dubai subsidiary and international coordination. A legal department is not optional.
But legal capacity also shapes behaviour. An institution with in-house lawyers, external counsel and reserves can enforce rules, defend interpretations, negotiate with governments, respond to disputes, design contracts and resist challenges. That is part of resilience. It is also part of power. The more valuable the resources beneath the ledger, the more each legal interpretation matters.
Public critiques of registry power often frame this as a gap between control and liability. If a registry can affect the recognised status of resources worth far more than the fees paid, but limits its own downside through contract and legal structure, discretion becomes economically asymmetric. A member may face existential harm from registry action; the registry may face litigation cost, reputational risk and limited damages exposure. Even if RIPE NCC's processes are generally careful, the structure deserves scrutiny.
Lu Heng's 2025 phishing warning captured a softer version of the same phenomenon. Scammers could exploit fear of RIPE NCC because some members treat the registry as if it were a sovereign regulator capable of threatening a business overnight. The warning correctly observed that RIPE NCC is a private membership entity, not a government or Internet police, and that legitimate processes such as Assisted Registry Checks are procedural and cooperative. But the fear is not imaginary. It comes from dependence on a registry relationship that members often experience as more than a vendor contract.
Legal capacity should therefore be tied to due process. A registry must have enough legal strength to prevent fraud, maintain accurate records, comply with sanctions and defend continuity. It should not use legal strength to expand discretion beyond the ledger function without explicit member consent and clear limits. Legal spending should be reported by category: ordinary corporate governance, compliance and sanctions, member disputes, policy development, litigation, external legal opinions, inter-RIR work and new-entity support. The more a line relates to discretionary enforcement or institutional expansion, the more detail members should receive.
The Dubai entity illustrates the point. RIPE NCC Middle East FZ-LLC began operating in 2025 after work on banking, employment, office space, tax, legislation and creating a legal entity fully owned by a Dutch non-profit association. The 2026 Activity Plan says the entity enhances RIPE NCC's presence in the Middle East and helps support members locally. The 2025 Financial Report notes new currency and reporting risk, including an AED 5m loan and a negative exchange result on AED of EUR 60,000 in 2025. This may be a sensible regional adaptation. It is also institutional expansion. In a fee-funded monopoly ledger, such expansion should face a fiscal test.
The same applies to external governance. RIPE NCC's role in Internet governance, government engagement, NRO costs, ICANN-related work and the RIPE Chair function may be important. The 2026 budget includes contributions such as NRO shared costs, an ICANN contribution, an IETF Endowment contribution and ISOC Platinum membership. These are not ledger entries in the narrow sense. They are part of the institutional ecosystem that sustains the RIR model. Members may value that ecosystem. But the invoice should not hide the cross-subsidy.
The legal lesson is simple. A registry that wants trust should be strongest at defending the ledger and weakest at discretionary overreach. Its legal budget should reflect that ordering. The more it spends on protecting accuracy, continuity, security, due process and external compliance, the stronger its legitimacy. The more it spends on preserving institutional prestige, expanding mandate or insulating discretion, the more members should worry.
IPv4 scarcity and budget politics
IPv4 scarcity did not make RIPE NCC poorer. It changed the source of its relevance. The free pool is no longer the main economic story. The stock of existing resources, the transfers between holders, the accuracy of the registry, the security metadata around routes and the contractual relationship with members are now more important than fresh allocation. A registry after exhaustion is not a distributor of abundance. It is an administrator of scarcity.
That changes budget politics in three ways.
First, the membership base becomes the revenue base. The 2026 budget assumes 20,000 LIRs and member fees of EUR 39.64m. Existing-member service fees alone account for EUR 36m. New-member service fees and sign-up fees are much smaller. RIPE NCC's fiscal stability depends less on growth in new allocations than on keeping a broad base of existing accounts paying.
Second, resource value and service cost diverge. The cost of recording a large address block is not proportional to the market value of that block. Yet the value of accurate recognition may be much higher for a large holder. This is the underlying argument for differentiated fees. It is also the underlying argument against turning the registry into a rent collector. If RIPE NCC says it charges for service, it should not justify fees by reference to the market value of IPv4. If it says larger holders should pay more because they receive more valuable recognition, it should admit that the charging scheme is partly value-sensitive.
Third, administrative rules become economically active. Rules about transfers, documentation, sanctions, legacy resources, LIR accounts, RPKI and account closure are no longer merely procedural. They affect liquidity, risk premiums, counterparty confidence and the cost of capital. A registry can insist that it does not own number resources while still exercising power over the conditions under which those resources are recognised and transacted.
This is why the fee model cannot be isolated from transfer-market politics. A large holder's willingness to accept a higher annual fee may depend on whether it sees RIPE NCC as a neutral record-keeper or an expanding gatekeeper. A small operator's willingness to pay a flat fee may depend on whether it receives efficient support and predictable rights. A member's view of reserves may depend on whether reserves protect service continuity or fund institutional positions that constrain market movement.
RIPE NCC is in a stronger position than some registries because its public processes are comparatively orderly, its financial reporting is detailed and its member votes are real. But orderliness should not be mistaken for incentive neutrality. A well-run club can still have a fiscal structure that favours some members over others.
The 2027 vote shows how IPv4 scarcity shapes member coalitions. The category model would have charged by PA resource holdings, included a /29 of IPv6 PA space in the base fee, included one ASN if policy requirements were met, and lowered the fee for about 75% of LIR accounts. It also would have charged the largest current holder more than EUR 30,000. That is modest relative to the market value of large IPv4 holdings, but politically significant as a signal: the ledger would begin charging more explicitly by resource scale.
Large holders could see such a model as the beginning of asset-based taxation. Small holders could see it as overdue fairness. The Board could see it as a way to preserve total income while lowering the barrier for smaller networks. Each view is rational. That is why the vote was close. IPv4 scarcity turned a budget line into a constitutional choice.
The scarcity context also affects multiple LIR accounts. When address space is scarce and account fees matter, members may hold multiple accounts for historical, operational or strategic reasons. A one-fee-per-account model can penalise organisational structures with many accounts, or reward holders that consolidate efficiently. A category model can penalise resource concentration within one account, or create incentives to distribute resources across accounts if not designed carefully. Fee models cannot avoid strategic behaviour; they can only choose which behaviour they encourage.
The broader political question is whether RIPE NCC wants its charging scheme to express membership equality or resource incidence. "Equality among members" sounds attractive, but it is not the same as equality of burden. "Resource-based differentiation" sounds economically rational, but it risks confusing service cost with asset value. The membership's narrow vote for Option A suggests a preference for continuity and simplicity, but nearly half the votes favoured a move toward differentiation. That is not a settled fiscal identity.
An incentive-compatible design would separate the issues more cleanly. Basic ledger service could have a low universal fee. Resource-scale charges, if any, could be justified as risk, support and recognition-cost charges rather than as stealth value taxes. Optional services could be funded separately. Member voting could be accompanied by distributional analysis showing which member types pay more, which pay less and how behaviour might change. Reserve targets could be tied to core continuity rather than total institutional spending. Transfer and resource policies could be reviewed for hidden fiscal incentives.
That would not remove politics. It would make the politics honest.
Voting incentives and the median-fee problem
RIPE NCC has a real democratic mechanism: members vote at General Meetings, including on charging schemes, financial reports, board seats and redistribution. The May 2026 General Meeting registered 3,421 members to vote and 3,049 cast ballots. That is a significant participation figure by the standards of member associations. It is also a reminder that the electorate is only a subset of the affected economy.
The affected economy includes downstream customers, non-member networks relying on members, counterparties in transfer transactions, organisations using RIPE Atlas and RIPEstat, governments, security researchers, route validators, legacy holders, sponsored resource holders and future entrants. Member voting is necessary. It is not sufficient to establish full legitimacy over every economic effect of registry decisions.
The internal electorate is heterogeneous. Some members are large incumbents with policy teams. Some are small operators with little time. Some are multiple-account holders. Some care deeply about RIPE community culture. Some care mainly about invoices. Some are active in mailing lists. Some appear only when fees change. Some operate in countries where banking and sanctions make payment difficult. Some can pass costs to customers; others cannot. One member-one vote and one account-one fee language can hide large differences in exposure.
In public-choice terms, the registry faces a median-fee problem. The chosen charging scheme may reflect the coalition that can assemble a majority, not the economically optimal allocation of costs. A model that lowers fees for many small accounts may lose if large or organised voters mobilise against it. A model that preserves simplicity may win even if it is regressive. A model that raises total income slightly may pass if the increase is dispersed. A surplus redistribution may win even if reserves would be prudent. A reserve build-up may pass if the benefits are vivid and the costs diffuse.
This is not a criticism unique to RIPE NCC. It is how associations behave. But because RIPE NCC controls a monopoly ledger, the usual association answer - "members voted" - should not end the analysis. Voting legitimises a decision procedurally. It does not prove the decision is efficient, fair or incentive-compatible.
The May 2026 charging vote is a good example. Option A won by 68 votes out of 3,026 non-abstaining votes on the charging resolution. The decision was procedurally valid. Yet a fiscal system for a EUR 42.5m income budget and 20,000-LIR assumption was decided by a margin smaller than many local clubs would see in a board election. That does not invalidate the outcome. It counsels humility in treating it as consensus.
Member voting also interacts with information. RIPE NCC publishes detailed documents, calculators and consultations. Still, the cost of understanding fee incidence is high. A member must evaluate the current fee, the proposed category model, resource holdings, ASNs, PI resources, legacy treatment, multiple accounts, future growth, redistribution, service value and institutional trajectory. Sophisticated members will model this. Many others will vote from principle, instinct or partial information.
The Board's role is therefore not neutral. By choosing which options to put forward, how to describe them, which model to recommend and how to frame the trade-off, it shapes the electorate's choice. In 2026, the Board recommended Option B but also offered Option A because members had asked for the current model to remain available. That was a reasonable compromise. It also made the vote a referendum on whether RIPE NCC should remain a flat-fee club or move toward resource-based fiscal differentiation.
The closeness of the result should lead to better deliberation, not a winner-takes-all pause. The next iteration should ask more granular questions. Should the base fee be lower? Should large holders pay more through a capped resource component? Should one ASN be included? Should PI and legacy resources be treated differently? Should multiple LIRs face anti-gaming rules? Should optional services be separated? Should the income target be challenged before the fee distribution is debated? Should reserve funding be separated from the ordinary service fee?
One useful reform would be a member-facing incidence statement for each charging proposal. It would show distributions by account size, region, member type, number of accounts, resource holdings and fee change. It would estimate behavioural responses. Another would be a cost-causation statement: which costs rise with member count, which with resource count, which with service complexity, which with legal risk, and which are fixed institutional overhead. Members should not have to infer these from dense budget documents.
Democracy in a registry club should be more than the right to approve a bill. It should be a right to see how the bill was constructed.
Resilience can become fiscal inertia
Resilience and accountability are both virtues. In registry governance they often pull against each other.
Resilience asks for reserves, professional staff, redundant systems, legal capacity, risk management, information security, international relationships, continuity funds, sanctions workarounds, local presence and the ability to respond quickly. Accountability asks for narrow mandate, cost discipline, member control, transparency, minimal cross-subsidy, due process, public metrics and limits on discretion. The danger is that resilience becomes the master word. Once everything is justified as resilience, accountability begins to sound like a threat.
RIPE NCC's 2026 budget is full of resilience work: registry accuracy, automation, LIR Portal modernisation, RPKI improvements, NRO RPKI programme work, ISAE 3000/SOC 2 Type 2 assurance for RPKI, ISO/IEC 27001 alignment, security measures, risk communication, legal framework review, EU legislation monitoring and organisational culture. Much of this is sensible. Critical infrastructure should be secure, audited and legally prepared.
But resilience has a fiscal appetite. It rarely says "enough" by itself. Every system can be more secure. Every legal framework can be more robust. Every community can be more engaged. Every office can be healthier. Every risk can be monitored more thoroughly. Every meeting can improve legitimacy. Without external discipline, resilience becomes a permanent reason to maintain or increase budgets.
The 2026 budget's total expense increase is modest in aggregate - 3% over the 2025 budget and 6% over the 2025 forecast - but the composition matters. Personnel is budgeted at EUR 25.55m, up 5% over the 2025 budget. Information Technology is EUR 4.25m, up 15%. Travel is EUR 1.2m, up 19%. CAPEX is EUR 930,000, up 16%. Legal grows to EUR 1.3m and 6 FTE. These increases may be justified. They also show that the institution's baseline is not shrinking after IPv4 exhaustion.
The cost-structure critiques argue that RIRs expanded sideways after their original allocation role declined, adding training, conferences, travel-heavy meetings and administrative structures. RIPE NCC can reply that post-exhaustion work is not decline but transformation: registry accuracy, RPKI, security, sanctions, transfers, automation and data services are harder than old allocation in some respects. Both claims contain truth. The decisive question is whether each activity is tested against the narrow ledger mission or allowed to inherit legitimacy from institutional history.
Accountability should therefore be functional. It should not ask only whether RIPE NCC is transparent; it should ask whether transparency is decision-useful. The Activity Plan and Budget is detailed, but members need a simpler map of the fiscal boundary between the ledger and the institution. The Financial Report is professional, but members need legal-spend categories, reserve-purpose categories and cost-causation analysis. The GM vote is real, but members need more than binary choices between large bundled models.
The phrase "protect the ledger, not the gatekeeper" is useful because it separates function from institution. For RIPE NCC, it should not be read as a call to weaken the registry. It should be read as a budgeting principle. The ledger deserves strong protection: records, uniqueness, RDAP, WHOIS, reverse DNS, RPKI, transfer integrity, support and security. The gatekeeper functions deserve narrower review: institutional expansion, broad policy positioning, discretionary interpretations, costly meetings, optional data services and external political engagement.
That distinction can make RIPE NCC stronger. If the membership can see that core continuity is ring-fenced, it may tolerate reserves and legal capacity more readily. If optional services are separately justified, members may support them voluntarily or through explicit votes. If large expansions are shown with long-term costs, opposition becomes less suspicious and more analytical. Accountability does not have to be anti-resilience. It can be the method by which resilience remains legitimate.
The alternative is slow distrust. Members may continue paying because they must, while privately treating the institution as too expensive, too broad or too self-protective. That is dangerous for a registry because its authority rests heavily on voluntary cooperation and perceived neutrality. A registry without sovereign force cannot rely indefinitely on inertia. It needs a belief that the fee is proportionate and the institution is constrained.
In that sense, cost discipline is not only a financial virtue. It is part of registry neutrality.
The neutral bookkeeper and the public institution
RIPE NCC's self-presentation combines two roles that are not identical. It is the neutral registry for Internet number resources in its service region. It is also a coordinator, educator, community convenor and participant in Internet governance. The first role is ledger neutrality. The second is institutional presence. The tension between them is growing.
Ledger neutrality means the registry should maintain accurate records without taking unnecessary positions on business models, resource value, political narratives or market outcomes. It should be predictable, procedural and modest. Institutional presence pulls in the other direction. It asks RIPE NCC to speak to governments, support policy communities, promote IPv6, advance routing security, attend forums, run training, develop data platforms, fund community work and represent technical expertise in governance debates.
These activities can support neutrality by making the registry better understood and more accountable. They can also threaten neutrality by making the institution a policy actor with its own interests. A registry that promotes itself, expands regional presence, builds public platforms and speaks in governance forums may still be neutral in resource administration. But it becomes harder to distinguish record-keeping from institutional agenda.
The cost structure shows the blending. External Engagement and Community is nearly EUR 10m. Community Building and Member Engagement alone is EUR 5.95m, with main expenses including outreach and PR, consultancy and travel. Coordination and Collaboration is EUR 1.95m. The Office of the Managing Director includes contributions and travel. The RIPE Chair function is separately supported. These lines are not improper. They are evidence that RIPE NCC is more than a database operator.
The problem is not breadth itself. The Internet has always relied on institutions that combine technical work, community process and governance translation. The problem is using the monopoly ledger fee to finance breadth without continuously renewing the case for each layer. A member who wants only neutral registry service may not want to fund a broad governance footprint. A member who values global Internet governance may see that footprint as essential. The charging scheme does not ask them separately.
RIPE NCC's public data services sharpen the issue. RIPE Atlas, RIPEstat and RIS provide value beyond paying members. They support operators, researchers, policymakers and the wider Internet. They are also not the same as maintaining the authoritative registry. If their social value is high, a case can be made for member funding. But that case should be made as a public-good subsidy, not as unavoidable registry cost.
The same is true of training. Training can improve routing security, registry accuracy, IPv6 deployment and operator capacity. In a diverse region, it may reduce long-term support burden and strengthen the network ecosystem. Yet a mandatory fee-funded training programme still creates distributional questions. Who attends? Which countries benefit? Are small operators actually reached? Could large operators, governments or sponsors fund more of it? What outcomes are measured?
Neutrality is also tested by IPv6. The 2027 Option B proposal included a /29 of IPv6 PA space in the base fee and described this as encouraging IPv6 take-up. That may be desirable. It may reduce barriers. But it is not neutral in the narrow ledger sense. It uses fiscal design to encourage a technical transition that has its own economic politics.
None of this means RIPE NCC should retreat into silence. It means institutional expansion should be unbundled and named. The registry should be able to say: here is what the ledger costs; here is what security costs; here is what public data costs; here is what community engagement costs; here is what government and governance work costs; here is who benefits; here is why mandatory member funding is justified.
The more precise the categories, the easier it is to preserve neutrality. The less precise they are, the more every cost becomes part of a self-protecting mythology: the registry is important, therefore the institution's breadth is important, therefore the fee is necessary. That chain is too easy.
The bookkeeper metaphor restores modesty. A bookkeeper may need good software, security, audit, counsel and a resilient office. It may publish useful statistics and train users. But it remains a bookkeeper because its authority comes from accurate service to reality, not from authorship of reality. RIPE NCC's fiscal design should make that humility visible.
Small operators and fixed-burden politics
The economics of fixed fees is harshest at the edge. A large operator can spread EUR 1,800 across millions of customers or large enterprise contracts. A small operator may spread it across hundreds or thousands. A start-up network may experience the sign-up fee, annual fee, administrative time and compliance learning curve as a barrier. A network in a lower-income country may face currency, banking and local-market constraints. A sanctioned or high-risk environment may face payment complexity even when the member wants to pay.
This is the small-operator penalty in a RIPE NCC setting. The visible annual fee is only part of the burden. The real burden includes legal review, staff time, documentation, participation, delayed support, understanding policy, maintaining account standing, implementing routing security and managing transfers or sponsorship. Large operators have compliance teams. Small operators have founders and engineers doing everything.
The 2025 Financial Report shows that even RIPE NCC faces collection and banking complexity. If the institution itself cannot easily invoice and collect from certain countries because banks classify them as high risk, small members in those countries face the same financial reality from the opposite side. Continued service is admirable. It also reveals that a uniform fee in euros is not uniform in economic effect.
Option B's proposed EUR 500 base fee was an attempt to lower barriers for smaller or low-resource accounts. It lost narrowly. That does not eliminate the underlying problem. If the association chooses a one-fee model, it should find other ways to mitigate regressive effects: hardship procedures, transparent payment extensions, crisis-country handling, lower entry costs for genuinely small operators, or targeted discounts tied to objective criteria. The challenge is avoiding abuse and complexity.
There is also a risk that small-operator relief becomes symbolic while the broader bundle remains expensive. A lower base fee funded by higher large-holder charges may help if it reduces real barriers. But if total costs keep rising, fee politics will return. The best pro-small-operator policy may be not only differentiated charging but scope discipline: keep mandatory costs as low as possible by separating optional services from essential ledger service.
Small operators also have weaker governance capacity. They are less likely to attend meetings, read every budget line, participate in consultations or model fee proposals. The members most sensitive to fees may be least able to influence them. Fee calculators help, but so would concise incidence summaries, country-level payment-risk explanations and plain-English cost-causation statements. The issue is reducing the attention tax that makes governance harder for members without policy specialists.
The same applies to broader services. If training, meetings and community engagement are justified partly as support for smaller or less-resourced networks, RIPE NCC should publish evidence of reach. How many small operators attend? Which countries benefit? Do training recipients improve registry accuracy, RPKI adoption or operational security? Are travel and meeting costs producing measurable inclusion, or mostly serving those already able to participate?
The small-operator problem is not solved by slogans about community. It is solved by measuring the total cost of membership, including attention, compliance and travel, and then designing fees and services to reduce that burden without creating new distortions.
Toward incentive-compatible registry finance
An incentive-compatible RIPE NCC financial model would begin with a narrow principle: mandatory money should follow mandatory functions. Members must rely on the recognised ledger. Therefore the ledger, security and minimum support functions have the strongest claim on compulsory funding. Other activities may be valuable, but their compulsory funding should be justified separately.
The first reform is cost classification. RIPE NCC should publish, alongside the Activity Plan and Budget, a core-continuity statement. It would identify the minimum costs required to maintain registry records, RDAP, WHOIS, RIPE Database functions, reverse DNS, RPKI, security, data escrow, essential IT, member support and legal compliance directly tied to those functions. It would also identify supporting costs and discretionary public-good costs. This would not force cuts. It would make cross-subsidy visible.
The second reform is reserve classification. The Clearing House reserve should be mapped to purposes: core service continuity, operating volatility, payment-risk absorption, legal contingency, system-stability commitments and strategic investment. Members should know how many months of core continuity are protected, not only what percentage of total institutional expenses the reserve covers. A reserve that covers 86% of total expenses may cover far more of the narrow ledger. That distinction changes the surplus-redistribution debate.
The third reform is legal-spend disclosure by category. Confidentiality can protect case details, but categories should be visible. Members should know how much legal spending goes to compliance, ordinary contracts, sanctions, governance, disputes, policy analysis, external institutional engagement and new legal structures. A fee-funded registry should not ask members to infer the price of discretion from one budget line.
The fourth reform is charging-scheme incidence analysis. Every proposal should show distributional effects across account types, resource holdings, member sizes and regions. It should show likely behavioural incentives, including multiple-LIR optimisation and resource movement. It should state clearly whether a charge is based on cost, risk, resource scale, ability to pay or policy preference. A fee justified by "fairness" should say which fairness principle it uses.
The fifth reform is optional-service funding experiments. RIPE NCC need not privatise or abandon RIPE Atlas, RIPEstat, training, meetings or community programmes. It could test partial sponsorship, opt-in contributions, grants, usage-based support for advanced features or ring-fenced public-good funds. If members are willing to fund these services voluntarily or through explicit votes, their legitimacy increases. If they are not, their mandatory funding should be reconsidered.
The sixth reform is a small-operator attention standard. Budget and fee documents should be usable by members without policy staff. A long budget is useful for experts. A two-page incidence and cost-causation summary is useful for democracy.
The seventh reform is a neutrality audit for institutional expansion. New offices, subsidiaries, government-relations efforts, large data initiatives and governance commitments should be assessed against the narrow registry mandate. The test should ask: does this protect the ledger, improve service, reduce risk, or expand institutional presence? If the answer is expansion, members may still approve it. But they should approve it knowingly.
These reforms would not decide the philosophical dispute over flat versus differentiated fees. They would make the dispute more honest. A one-fee model might remain if members knowingly prefer club equality and simplicity. A category model might return if members decide incidence matters more. A hybrid might emerge: a low base fee, modest resource-based supplement, separate charges for high-cost services and explicit funding for public goods.
The most important change would be conceptual. RIPE NCC should be evaluated not as a benevolent technical institution above economics, but as a club with monopoly characteristics and fiscal incentives. That does not diminish its importance. It clarifies the duties that come with importance.
The ledger needs resilience. Members need accountability. Small operators need lower fixed burdens. Large holders need predictable, non-extractive rules. The broader community needs some shared services. The institution needs enough money to operate without turning every challenge into a crisis. These goals conflict. A mature registry does not pretend they do not. It designs for the conflict.
RIPE NCC's great advantage is that it still has the tools to do so: published accounts, member votes, professional staff, a detailed budget, a reserve mechanism, public consultations and a culture that at least recognises the need for explanation. Its great risk is complacency born of the same stability. A stable club can drift into expensive breadth more quietly than a failing registry can. It can convert resilience into habit, habit into entitlement, and entitlement into fiscal inertia.
The economics of fees, reserves and incentives is therefore not a side issue. It is the live test of whether RIPE NCC can remain a neutral ledger institution while funding a much broader role. The answer will not be found in mission statements. It will be found in the invoice, the reserve policy, the legal budget, the next charging-scheme vote and the willingness to separate the cost of keeping the book from the ambitions of the bookkeeper.