An African network operator watching AFRINIC's crisis did not need to be persuaded that a regional internet registry should hold cash. The practical question was sharper. If a court order can freeze the registry's bank accounts, if legal bills can consume money that would otherwise pay technical staff and maintain systems, if a receiver has to keep the organisation alive while elections are disputed, and if RPKI, reverse DNS, WHOIS, RDAP and allocation records still have to function while the corporate shell fights for legitimacy, what exactly are reserves for?
The answer cannot be "for AFRINIC" in the abstract. A registry reserve is legitimate when it protects the narrow ledger service that members cannot replace: resource records, authentication, publication, ticket handling, contractual standing, audit trails, secure systems and orderly transition. It becomes dangerous when the same cash lets an institution continue open-ended litigation, discretionary enforcement, mandate drift or governance conflict without effective member discipline. In one form, reserves are continuity insurance. In the other, they are institutional insulation.
AFRINIC made that distinction concrete. In 2021, the Internet Governance Project reported that the Supreme Court of Mauritius had provisionally frozen up to $50 million of AFRINIC funds in the dispute with Cloud Innovation. The dispute followed AFRINIC's attempt to terminate Cloud Innovation's membership and reclaim IPv4 resources, with Cloud Innovation seeking court protection and damages after challenging AFRINIC's conduct. The merits, motives and legal positions have been contested for years. Reserve policy does not need to decide every point in that fight. It does need to ask what happens when member-funded continuity cash becomes both the fuel and the target of a scarcity-era legal war.
That is the mechanism here. Registry reserves are a shock absorber only if the shock is defined. They are defensible when bounded to payroll for essential registry staff, critical vendors, secure publication, escrow, court-required preservation, audit, cyber response, member notices, RPKI and reverse-DNS continuity, and enough lawful authority to keep the ledger stable while governance is repaired. They become a moral-hazard buffer when they fund a strategy that would otherwise be disciplined by budget pressure, member confidence, settlement incentives or the collapse of ordinary authority.
AFRINIC's public reserve policy started from a sensible premise. A 2008 board resolution said the organisation should try to build a reserve fund sufficient to cover two years of operational expenses. A later public explanation clarified that the reserve should mean actual cash set aside from AFRINIC's own holdings into designated bank accounts; that ordinary operational accounts should not count as reserve funds; that withdrawals should be tightly controlled and subject to approval; and that the two-year horizon assumes a major event in which no or limited income may flow while operations continue, wind down or are reshaped. That is a real continuity idea. A registry should not be run hand to mouth.
But AFRINIC's crisis also shows why a two-year target is incomplete. A reserve ratio without permitted-use discipline can become a war chest. A cash balance without legal-spend classification can hide the difference between defending the ledger and defending an expansive institutional theory. A drawdown rule that assumes an ordinary board is weak when the board itself is absent, contested or being reconstructed by a receiver. A reserve target that ignores replenishment can turn today's litigation into tomorrow's quasi-tax on members. A registry reserve therefore needs more than size. It needs a constitution.
This is not an article about invoice incidence, generic member rights or ordinary board oversight. The reserve question is narrower: who absorbs institutional failure, what cash is protected for the ledger, what uses are forbidden, how legal risk is classified, when emergency drawdowns are allowed, how depletion is replenished, and whether scarce IPv4 recognition lets a registry turn member-funded insurance into institutional leverage.
AFRINIC is the hardest case because the evidence cuts both ways. It needed reserves. The record includes reported address-record corruption, a high-value dispute with Cloud Innovation, a bank-account freeze, years of board failure, receivership, an annulled election, later efforts to restore a board, continuing winding-up and bylaw disputes, legal-budget overhang, and a region where scarce IPv4 recognition has economic value beyond an ordinary administrative file. A brittle registry with no reserve would expose members to immediate service risk. A registry with reserves and weak reserve discipline can also keep funding conflict after confidence has collapsed.
The reserve should therefore be understood as a continuity firewall. It must stop institutional fire from reaching the registry ledger. It must not stop member discipline from reaching the institution.
The reserve is continuity insurance, not institutional comfort
A regional internet registry has an unusual balance sheet. Its most important outputs are not conventional goods. They are entries in a public numbering ledger, route-origin and certification material, reverse-DNS delegations, WHOIS and RDAP data, member-account status, allocation and transfer records, and operational trust. Members pay fees because the registry is the recognised settlement layer for number resources in its region. It does not own the internet and it does not route every packet, but its records shape who can prove recognised control over resources that networks, customers, brokers, lawyers and counterparties treat as economically meaningful.
That role creates a continuity problem. A normal company can cut services, enter insolvency, merge, sell assets or close units. A registry cannot simply stop publishing accurate records because a lawsuit becomes expensive or an election fails. If RPKI repositories drift, if reverse DNS stalls, if WHOIS or RDAP records become unreliable, if allocation and transfer requests disappear into uncertainty, or if staff cannot confirm member authority, the harm is not contained within the registry's office. It moves to operators that did not choose the dispute and cannot shop for a different regional ledger.
Reserves are supposed to prevent that migration of harm. They are a firewall between corporate distress and ledger continuity. They exist so that a bank dispute, delayed revenue, litigation injunction, failed board, disaster, cyber incident or temporary income collapse does not immediately impair essential functions. The firewall buys time for a narrow purpose: to keep services alive while lawful governance is restored, while a receiver or emergency authority preserves the institution, or while an orderly transition is arranged.
AFRINIC's own reserve explanation fits this model as a factual exhibit. It describes cash set aside from the organisation's own holdings, not merely an accumulated surplus. It distinguishes designated reserve accounts from normal operating bank accounts. It contemplates a major event that limits income and requires operations for up to two years. It recognises that crisis costs may differ from ordinary costs: some obligations continue, some activities are reduced, some legal or labour obligations arise, and the operating plan may need to be narrowed. That is not the language of a public-relations cushion. It is a service-continuity instrument.
The difficulty is that a continuity firewall must have direction. It should stop institutional fire from reaching the ledger. It should not stop discipline from reaching the institution. If management overreaches, if elections are defective, if legal spending explodes, if bylaw authority is unclear, or if scarce-resource policies become economically damaging, reserves should not let the registry ignore the members whose fees created the buffer. A reserve that pays essential staff during receivership protects members. A reserve that lets a weak institution litigate indefinitely against members may protect the institution from members.
This distinction matters because regional registries are monopoly utilities for their service regions. An AFRINIC member cannot move its African resource relationship to ARIN, RIPE NCC, APNIC or LACNIC because it dislikes the budget or distrusts the litigation strategy. Exit is structurally limited. That makes internal discipline more important. Fees and reserves are not ordinary customer revenue; they are compulsory support for a recognised ledger. The larger the reserve, the greater the need for rules showing when member-funded cash may be used and when it must be preserved.
The protected service should be defined before the protected institution. Core service categories deserve priority: registration database integrity, RPKI publication, reverse DNS, RDAP and WHOIS availability, ticket triage for essential changes, billing continuity where it preserves standing, cyber security, audit logs, critical vendors, staff needed to operate those functions, and legally required preservation of records. Non-core activities should be suspended before the reserve is consumed. Conferences, broad advocacy, discretionary outreach, political campaigns, expansion projects, reputation management and non-essential travel do not belong ahead of ledger continuity during a reserve drawdown.
AFRINIC's crisis shows why the line must be explicit. The organisation's public materials describe many valuable activities: training, community support, research, policy participation, internet-development programmes and stakeholder engagement. These may matter in ordinary periods. They should not be confused with the essential registry during a crisis. When a bank-account freeze, receivership or court dispute threatens viability, the reserve should not be treated as a general fund for preserving every institutional habit. It should protect the functions members cannot replace.
The first rule of reserve discipline is therefore simple: define the narrow ledger before authorising the broad institution. If the reserve is justified by RPKI, reverse DNS, WHOIS, RDAP and registration continuity, drawdown rules should say so. If the reserve is justified by payroll, it should be payroll for essential functions during a defined emergency, not an indefinite guarantee of pre-crisis staffing levels. If the reserve is justified by legal necessity, the legal matter should be classified by whether it preserves ledger continuity, defends against asset seizure, supports court-supervised recovery or pursues a discretionary institutional position. A registry that cannot make those distinctions does not have a reserve constitution. It has cash.
AFRINIC's two-year reserve target answered only the easy question
On paper, AFRINIC's two-year reserve ambition was conservative in the best sense. A small, specialised registry serving a region with uneven payment capacity, scarce IPv4 supply and dependence on global technical trust should not run with a few months of cash. Its services are too important and too hard to replace quickly. A two-year target recognises that institutional repair can be slow, especially when courts, member elections, international coordination and technical continuity all intersect.
Public financial exhibits show that AFRINIC had taken the idea seriously before the crisis fully escalated. The financial-reserves page recorded revenue reserves of about $4.76 million at the end of 2018 and strategic cash reserves of about $3.20 million by the end of the third quarter of 2019. Later finance updates reported cash holdings of about $11.35 million in late 2021 and $12.88 million in early 2022, with roughly $6.37 million included in strategic cash reserves. These figures are not a current audited position, but they show meaningful liquidity relative to ordinary annual revenue.
The 2022 budget gives the scale. AFRINIC projected revenue of about $6.22 million and operating expenses of roughly $6.14 million including contingency. Staff costs were the largest single cost line. Legal and consulting fees were budgeted at $247,000, including $140,000 for legal expenses. In normal conditions this looked like nonprofit prudence: revenue close to expenses, a small contingency, and a strategic cash reserve for stress.
Then the stress arrived in a form that ordinary budgets rarely price. The 2022 finance page later displayed legal fees totalling about $1.25 million, including more than $1.08 million to one law firm, plus other legal and related entries. That amount is not just a line item. It is a warning about legal-budget convexity. A registry can budget legal fees as if they are controllable. A scarcity dispute can make them non-linear. Once resource recognition, member status, bank accounts, board authority, receivership, bylaw validity and winding-up risk are all contested, legal spending stops being a support function and becomes a central operating reality.
This is where a reserve target without legal classification becomes dangerous. If a registry holds a two-year reserve and legal conflict consumes a material share of annual revenue, members need to know what kind of legal spending is being financed. There is a difference between defending the availability of essential registry services and pursuing a broad theory of institutional control over a member's commercial use of IPv4 resources. There is a difference between complying with a court order and initiating aggressive enforcement that predictably causes litigation. There is a difference between defending staff from improper personal exposure and funding a reputational campaign. There is a difference between preserving the ledger during winding-up proceedings and resisting every settlement that would narrow discretion.
AFRINIC's crisis contains all the grey zones. The original Cloud Innovation dispute arose after AFRINIC expressed concerns about resource use and regional policy compliance, against the background of earlier record-integrity scandals that had damaged confidence. AFRINIC had reasons to care about abuse, fraud, accuracy and the meaning of justified need. Cloud Innovation had reasons to treat reclamation as existential to its business and its customers. Courts were asked to preserve rights, restrain actions, consider damages, supervise governance and later handle receivership and winding-up arguments. For reserve policy, the key point is not that one side was wholly right. It is that a member-funded reserve can be pulled into every stage unless the rules separate ledger defence from policy war.
The 2021 bank-account freeze makes this more than a budget story. A reserve is protective only if it is accessible under the conditions it was built to survive. If all cash sits in bank structures equally vulnerable to attachment or freeze, the reserve may be real in accounting and fragile in law. If a litigant can immobilise funds used to maintain essential public records for unrelated members, continuity has not been adequately insulated. If the registry can label any fund "continuity" after the fact, creditors, courts and members will distrust the label. The reserve needs legal and operational legibility before the emergency, not rhetorical explanation after it.
That raises a design question every RIR should answer. Should strategic reserves be held in a legally segregated structure for essential registry continuity? Should there be a trust-like arrangement, escrow, restricted account, dual approval, emergency court protocol or inter-RIR continuity facility? Should litigants be able to immobilise the same cash that pays essential registry staff and maintains public-resource records? The answer cannot be that members should simply trust the board, because AFRINIC's board itself failed. Nor can it be that courts should ignore claims against the registry. The answer is a reserve constitution that lets courts, members and auditors see which funds are dedicated to public ledger continuity and which funds are ordinary corporate assets.
AFRINIC's published policy already distinguishes designated reserve accounts from operational accounts and requires controlled withdrawals. The crisis suggests that this distinction needs a harder shell. It should state whether reserves may be attached, pledged, spent on non-core disputes, used as collateral or drawn for litigation. It should explain what happens if the board cannot lawfully approve a withdrawal. It should identify who certifies that a drawdown protects ledger continuity rather than institutional preference. Without those answers, a two-year reserve can disappear into the first dispute that knows where to pressure the bank account.
The target answered the easy question: how much cash might a registry need to survive a severe interruption? The harder questions are what that cash may buy, who can release it, what must be preserved, what must be cut first, and who pays to rebuild it after the loss. AFRINIC's crisis priced the harder questions.
Scarce IPv4 turns cash into policy leverage
Reserve policy would be simpler if AFRINIC were only an address-book operator in a world of abundant supply. It is not. IPv4 scarcity changes every financial rule around a registry because the ledger supports positions with market value. The registry's fee revenue comes from members whose number-resource holdings can sustain access networks, hosting businesses, leasing arrangements, enterprise customers and acquisition valuations. Its policies can affect whether resources are reviewed, reclaimed, transferred, assigned, leased or left in place. Its legal fights can change the perceived risk of resources recorded in its region.
AFRINIC's exhaustion materials place the region in Phase 2 of IPv4 exhaustion, with a minimum allocation or assignment size of /24 and a maximum of /22. The same public materials describe services such as WHOIS, reverse DNS, resource certification, IRR and member support. These are narrow factual exhibits, not reasons to defer to institutional conclusions. They show the operating context: new IPv4 supply is rationed, existing recognition matters more, and a resource holder's relationship with the registry can carry substantial economic consequences.
When recognition becomes more valuable, disputes over recognition become more expensive. When disputes become more expensive, reserves become more tempting. This is the second moral-hazard problem. A registry that controls scarce-resource recognition may believe that its mission justifies aggressive legal and enforcement spending. It may argue that without strong defence, remaining resources will be drained, misused or commercialised in ways that undermine regional goals. That concern is not imaginary. KrebsOnSecurity's 2019 reporting on alleged AFRINIC-related address-record manipulation described claims that dormant African resources had been commandeered and sold through companies linked to a former AFRINIC policy coordinator, with an estimated market value above $50 million. In a scarcity market, weak controls invite fraud.
But scarcity can also be used to justify overreach. If every commercial use of IPv4 is treated as suspicious, if every out-of-region customer is framed as a threat, if every transfer or lease is treated as a political loss, then the registry drifts from ledger discipline into economic planning. The cost of that drift is then paid from the same member base that depends on predictable records. Reserves become a buffer that lets the registry pursue a contested theory longer than ordinary member discipline would allow.
The Cloud Innovation dispute sits on this fault line. IGP reported that Cloud Innovation had received rights to millions of IPv4 numbers from AFRINIC and used part of that resource base outside Africa, including through customers in China, while AFRINIC later challenged compliance with policy and membership terms. Cloud Innovation and associated voices argued that AFRINIC was changing the economic meaning of earlier allocations and using administrative power in a way that threatened resource holders. AFRINIC argued that it had to defend policy compliance and the region's address space. The litigation then pulled resource policy, commercial dependence, public-interest language and bank-account risk into one conflict.
For reserves, the question is not whether IPv4 should be property, whether leasing should be encouraged, or whether every scarcity-market behaviour is acceptable. The question is whether cash raised for registry continuity may be used to reshape the scarcity market. A reserve justified as ledger insurance should not silently become a subsidy for market intervention. If AFRINIC wants to spend heavily on enforcement in a scarcity dispute, the reserve policy should require a narrow ledger test: does the spending correct demonstrable record corruption, preserve uniqueness, prevent fraud, comply with a court order or protect unrelated members from immediate service harm? Or does it fund a broader institutional preference about how IPv4 should be used?
This distinction protects both sides. It prevents a powerful holder from using litigation pressure to exhaust the registry and capture the ledger. It also prevents the registry from using member-funded reserves to pursue a discretionary theory without pricing the cost. A reserve constitution should not assume that large resource holders are innocent or that the registry is self-serving. It should force each drawdown to identify the protected ledger function.
Scarce IPv4 also changes replenishment. If a high-value dispute drains reserves, future fees may refill them. That means small access networks, universities, data centres and public networks may pay for a fight centred on large address positions and policy authority. The same scarcity that makes enforcement important also makes incidence unfair if costs are not classified. A small member with a /24 or legacy dependency should not have to guess whether tomorrow's reserve replenishment funds RPKI continuity, receiver costs, a legal appeal, or an institutional campaign against a large holder.
The reserve ratio should therefore be linked to scarcity risk. During abundance, a two-year operating reserve may be enough. During scarcity, the reserve should include a legal-risk layer, a transition floor and a proportionality rule for high-value enforcement. It should ask whether the possible litigation cost is proportionate to the ledger harm being prevented. It should require settlement analysis before scarce-resource disputes consume the core continuity layer. It should treat address-market dependence as a reason for tighter discipline, not as a blank cheque.
AFRINIC's crisis should make registries more serious about fraud, audit trails and accurate records. It should also make them more careful about using reserves to finance discretion. Scarcity makes continuity more valuable. It also makes institutional power more dangerous.
Legal budgets need their own reserve constitution
Legal spending is the hardest reserve category because it can be both essential and self-serving. A registry under attack must be able to defend itself. It must respond to injunctions, preserve evidence, protect staff from improper claims, defend bank access needed for payroll and infrastructure, explain the public nature of numbering resources, and participate where court outcomes could affect continuity. A registry that cannot afford lawyers can be captured or dismantled by whoever can litigate longest.
At the same time, legal spending is the easiest way for a registry to avoid governance discipline. Privilege can hide strategy. Ongoing proceedings can justify limited disclosure. Complex cases can be presented as too sensitive for member debate. Each new motion can be described as unavoidable. Over time, members see only the invoice and the rhetoric: the institution says the fight is necessary; opponents say it is abuse; lawyers continue to be paid. The reserve becomes a legal oxygen supply.
AFRINIC's financial exhibits show the scale of the problem. A budgeted legal-and-consulting line of $247,000 in 2022 sat beside later disclosed legal fees of about $1.25 million for that year. That gap is not proof of misconduct. Litigation can genuinely surprise a budget. But it is proof that legal spending must have a different governance category from ordinary operating costs. A registry facing high-stakes scarcity litigation cannot treat legal fees as just another administrative expense.
The first classification should separate mandatory legal defence from discretionary legal strategy. Mandatory defence includes responding to court orders, preserving essential registry services, protecting bank access, defending the legal separation between numbering resources and corporate assets in winding-up proceedings, and maintaining compliance with the law of incorporation. Discretionary strategy includes initiating claims, expanding disputes, resisting settlement on policy grounds, public communications designed to influence members, and litigation whose main purpose is to preserve broad institutional discretion rather than core ledger continuity. The line will not always be clean, but drawing it changes incentives.
The second classification should separate ledger-defence legal spending from institution-defence legal spending. Ledger-defence spending protects the integrity, availability and neutrality of the number-resource record. Institution-defence spending protects the current corporate body, board, management, policy preference or reputation. Sometimes they coincide. If a winding-up petition would treat numbering resources as corporate assets, defending the ledger and defending the institution may overlap. If a member challenges a flawed election, defending the institution may conflict with member discipline. If a resource holder seeks an injunction to prevent immediate reclamation while a contractual dispute is heard, resisting all relief may be institution-defence even if the registry presents it as policy enforcement.
The third classification should separate emergency spending from campaign spending. Emergency spending is time-limited and court-driven. Campaign spending is a continuing choice. AFRINIC's crisis has produced repeated public claims from the registry, Cloud Innovation, Larus-linked voices, member groups, ICANN, ISPA, Smart Africa, operator associations and commentators. Each side has an interest. A reserve rule should prevent members from funding a public campaign under the cover of emergency continuity.
The fourth classification should be settlement utility. A registry is not merely a litigant; it is a settlement utility under stress. Its legal strategy should seek outcomes that reduce uncertainty for the ledger and members, not merely outcomes that win institutional claims. Settlement can be valuable even when the registry believes it is right. A proportional settlement that protects records, preserves service continuity, confines a disputed resource position and avoids future fee shocks may be better than a courtroom victory that consumes reserves and deepens legitimacy loss.
Settlement principles need not reveal privileged tactics. They can state that any settlement must preserve uniqueness, maintain accurate records, protect third-party network continuity, avoid preferential treatment, respect court orders, and disclose material budget consequences. They can also state what the registry will not spend reserves to obtain: punitive leverage unrelated to ledger service, retroactive policy expansion without member approval, or public messaging victories. Members do not need every legal memo to know whether their reserves are being used with discipline.
AFRINIC's receiver-era experience reinforces the point. The NRO's 2023 statement described the receiver's role as maintaining the status quo of AFRINIC's assets, preserving business value, overseeing elections, facilitating a proper board and appointing a chief executive. That statement is useful here as a narrow description of the receiver's assigned role, not as authority for any broader conclusion. During such a period, reserve spending should be especially narrow. The receiver's task is continuity and restoration. It should not become a route through which contested policy ambitions are financed while ordinary control is absent.
The legal-budget rule should be simple enough for members to understand: every major legal drawdown should disclose the case category, the protected function, the approved cost band, the funding source, the authority that approved it, the expected operational risk if the case is lost, and the replenishment plan if reserves are used. Privilege can protect tactics. It should not hide the economic classification. A monopoly ledger owes its users that much.
Reserve authority when ordinary governance fails
AFRINIC's reserve explanation makes approval central to withdrawals and trigger events. That is normal in an ordinary nonprofit. It is insufficient for a registry whose ordinary governance can fail. AFRINIC operated for years without a functioning board after litigation and governance disputes left it unable to elect or appoint normal leadership. Courts in Mauritius appointed a receiver to keep the organisation alive and move it back toward functional governance. A June 2025 election was suspended and annulled after allegations involving voter documentation and powers of attorney. A later election restored directors, but legal and legitimacy questions continued.
This history matters for reserve policy because a board-only control assumes the very institution that failed. If the reserve can be drawn only by board resolution, what happens when there is no lawful board? If a receiver may draw it, under what member-visible standard? If staff need emergency funds to keep RPKI and reverse DNS alive while the board is absent, who authorises payment? If a newly elected board is under challenge, should it be able to approve major litigation spending from reserves? If a board is accused of capture by one faction, should it control the cash buffer without additional safeguards?
The answer is not to remove the board from reserve policy. A board is necessary. But reserve control should have layers. The first layer is the ordinary board, which approves the reserve policy, annual target, investment rules and non-emergency use. The second layer is member disclosure and, for extraordinary drawdowns, member approval or later ratification. The third layer is an emergency continuity authority for periods when the board cannot lawfully act. The fourth layer is independent audit and public reporting after the emergency. The fifth layer is external coordination only for ledger continuity, not for ordinary policy choice.
AFRINIC's receivership shows why the emergency layer must be designed before the emergency. The receiver had to preserve assets, supervise elections and keep the registry functioning. That is a sensible court-supervised bridge. But if the reserve constitution does not define what a receiver may spend and why, the bridge becomes discretionary. The receiver may need legal advice, election vendors, staff payroll, technical vendors and court filings. Some of these are continuity expenses. Some are governance-restoration expenses. Some may be litigation expenses. Members should not have to infer the categories after the money has gone.
Governance failure also creates a replenishment problem. Suppose a receiver uses reserves to sustain essential services and organise elections. Replenishment through future member fees is legitimate because the reserve was spent to restore member control over the ledger. Suppose a contested board uses reserves to fight bylaw changes, resist disclosure or extend a resource-policy conflict. Replenishment is less legitimate because the reserve was spent while discipline was weak. The same future charge can therefore be either an insurance premium or a forced contribution to unresolved governance.
The June 2025 election episode sharpens the point. Public reporting said voting was suspended after questions about powers of attorney and voter documentation, then annulled by the receiver after stakeholder concerns. ISPA alleged votes had been cast without proper authority; ICANN demanded explanations and warned of possible compliance review. The reserve question is whether member-funded cash should finance election repair, legal responses and communications under rules those members can inspect.
Election legitimacy and reserve legitimacy are connected because reserve authority depends on the same member map. If members cannot trust who can vote, they cannot trust the process that approves reserve replenishment or extraordinary drawdown. If powers of attorney are disputed, if resource-member status is unclear under Mauritian law, if bylaw reforms may change participation rights, then a board resolution approving legal spending may be formally valid but economically suspect. Reserve discipline therefore depends on member-register discipline. A registry cannot claim strong cash authority while its member-authority map is contested.
The later restoration of a board did not erase that need. A functioning board can prepare budgets, appoint interim management and plan strategy. AFRINIC's public posture in 2026 included signs of renewal: interim management, budget planning, work toward a multi-year strategy and efforts to restore ordinary operations. Those signs matter. But the first reserve test for a restored board is not optimism. It is whether it narrows the conditions under which member-funded reserves can be spent. If the new board treats reserves as proof of institutional health while legal overhang remains opaque, the governance failure has merely been refinanced.
A reserve constitution should therefore include a governance-failure clause. In the absence of a lawfully functioning board, reserves may be drawn only for specified essential services, court-ordered preservation, election restoration, member-register verification, minimal legal defence of continuity and verified staff obligations. Discretionary policy litigation, new strategic initiatives and non-essential programmes should be prohibited unless an independent court or member-approved emergency process authorises them. Post-restoration disclosure should follow within a fixed period. A crisis reserve that depends on non-crisis governance is not a crisis reserve.
The mandate firewall: what reserves may not finance
Every reserve policy needs negative space. It is not enough to say what reserves may fund; it must say what they may not fund. In a registry, the prohibited uses are where institutional economics matters most, because the institution's incentives are not neutral. A registry wants continuity, legitimacy, policy authority, staff security, international recognition and freedom from disruptive litigation. Members want continuity too, but they also need discipline over registry discretion. A reserve that funds everything the registry calls necessary weakens that discipline.
The first prohibited use should be open-ended litigation without defined reserve exposure. AFRINIC's experience shows how litigation can become the operating environment rather than a temporary event. A reserve may need to fund emergency legal defence. It should not fund an indefinite legal campaign unless members receive cost bands, triggers, settlement principles and replenishment consequences. If a case is expected to cost more than a defined share of annual operating expenses or strategic reserves, it should require a higher approval threshold. A registry that can spend two years of operating reserve on a dispute has effectively found a way to spend future member fees today.
The second prohibited use should be discretionary enforcement that lacks proportionality review. AFRINIC had real reasons to worry about record integrity after the reported address heist. But the leap from fraud control to broad reclamation risk is where reserves can distort incentives. If management believes a reserve will support years of litigation after a severe enforcement action, it may underestimate the cost of choosing the most aggressive remedy. Reserve policy should require high-value enforcement actions to include a proportionality assessment before reserves can support legal defence. The question should be whether the remedy is tailored to ledger integrity or whether it attempts to control a member's business model.
The third prohibited use should be mandate expansion during crisis. A registry in distress may seek broader authority: more control over transfers, more say over leasing, greater power over resource use, stronger bylaw changes, wider government support, or external recognition as a special public body. Some reforms may be necessary. But reserves justified by continuity should not finance mandate expansion while members are distracted by crisis. A mandate firewall should say that reserve drawdowns cannot be used to create or defend new discretionary powers unless those powers are directly necessary for essential service continuity and have received proper review.
The fourth prohibited use should be reputation management detached from service facts. Public communications during a crisis are necessary. Members need to know whether services continue, what court orders mean, whether invoices are valid, whether elections are scheduled and how resource requests are handled. But a reserve should not fund broad campaigns to demonise critics, influence elections, frame commercial actors as enemies or substitute narrative for disclosure. AFRINIC's crisis has been surrounded by statements from the registry, Cloud Innovation, NRS, ICANN, ISPA, Smart Africa, operator groups and commentators. A registry's reserve should pay for factual member notices, not factional persuasion.
The fifth prohibited use should be cross-subsidising non-core programmes when core risk is unresolved. Training, research, community support, meetings and stakeholder engagement can be valuable in ordinary times. The 2026 coverage of AFRINIC's troubles noted the registry's concern that delays and soaring legal costs obstructed initiatives such as training and research. That complaint may be true. It also reveals the priority conflict. If legal costs and continuity risk are high, non-core programmes should not be sustained from reserves merely to show that the institution is alive. Members need the ledger first.
The sixth prohibited use should be replenishment without diagnosis. After a reserve drawdown, a registry may ask members to rebuild the buffer. The replenishment plan should identify why the drawdown occurred and what controls changed. If reserves were depleted because bank accounts were frozen, the plan should address legal segregation and banking structure. If they were depleted by legal fees, it should address legal-budget approval and settlement policy. If they were depleted by governance failure, it should address emergency authority and election controls. Replenishment without diagnosis creates a moral-hazard loop: members pay, the buffer refills, and the next crisis repeats.
AFRINIC's reserve page already contains the seed of a mandate firewall when it says reserves are for operational activity after a major event causing no or limited income, and that withdrawals are strictly controlled. But the policy needs to be more explicit because the crisis has become more complex than a revenue interruption. AFRINIC did not merely face a short-term fee shock. It faced litigation over resources, bank accounts, board legitimacy, elections, receivership, bylaw questions, winding-up and the treatment of IPv4 commercialisation. A policy written for operational interruption must be updated for institutional conflict.
The best formulation is blunt: reserves are not a substitute for member consent. They may preserve the registry while consent is restored. They may not be used to evade the consequences of lost consent. If courts question governance, if elections fail, if litigation exposes unclear authority, or if members reject a course of action, the reserve should keep the ledger alive while the institution narrows its mandate. It should not keep the institution comfortable while the ledger's users remain uncertain.
Transparency is the price of member-funded insurance
Insurance works only if the insured understand the coverage. Members fund AFRINIC's reserves through fees. They therefore deserve a reserve statement that reads less like an accounting note and more like an institutional contract. It should show the target, current balance, restricted balance, unrestricted cash, permitted uses, prohibited uses, drawdown history, replenishment plan, legal exposure and service-continuity mapping. It should not require members to reconstruct the position from scattered budgets, finance updates, board resolutions and crisis statements.
AFRINIC's public finance materials provide partial visibility. The finance page describes an annual budgeting process, including management review, Board Finance Committee evaluation, board approval, website publication, presentation at the annual member meeting, quarterly financial reports and mid-term review. The financial-reserves page explains the two-year target and the distinction between revenue reserves and strategic cash reserves. The 2021 and 2022 finance updates disclose cash holdings and strategic cash reserves. The 2022 finance page discloses legal fees. These are useful exhibits.
They are not enough. Crisis reserve transparency requires comparability and timeliness. Public finance documents seen in mid-2026 appeared to cover years only through 2022. That may reflect the disruption of receivership and board failure. It also demonstrates the problem. If the most important financial period in the registry's modern history is precisely the period for which ordinary reporting is thin, members cannot evaluate the reserve bargain. A registry cannot rely on reserves during a crisis and postpone reserve reporting until after trust is restored. Reporting is part of restoration.
A mature reserve report would separate five columns. First, essential registry operations: payroll and vendors for RPKI, reverse DNS, WHOIS, RDAP, IRR, database integrity, cyber security and member support. Second, governance restoration: elections, receiver costs, audit, member-register verification, bylaw review and required court-supervised processes. Third, legal defence of continuity: bank-access protection, winding-up response, preservation of numbering resources as non-distributable public coordination records, and mandatory compliance with court orders. Fourth, discretionary legal or policy positions: enforcement claims, appeals, transfer-policy defence, resource-use theory and institutional reputation matters. Fifth, non-core programmes: outreach, training, meetings, research and stakeholder engagement.
The report should show which columns have used reserves and which used operating revenue. It should identify material deviations from budget. It should disclose whether legal costs have delayed technical investment or service improvements. It should state whether reserve funds are legally segregated from operating accounts and whether any were restrained, pledged or subject to court orders. It should identify the authority for each drawdown. It should publish a replenishment path that states whether future fees, cost cuts, external support or reduced programmes will rebuild the reserve.
Transparency also disciplines legal advisers and executives. If lawyers know that reserve-funded work will be classified as continuity defence, governance restoration or discretionary strategy, they will structure advice differently. Management will have to justify why a litigation step belongs in the protected category. Board members will have to vote with visible cost consequences. Members will be able to ask whether the next reserve draw protects the ledger or prolongs institutional conflict. That is precisely the discipline a monopoly ledger needs.
AFRINIC's legal-cost overhang makes this urgent. A public budget that expected $140,000 in legal expenses and a later disclosed legal-fee total above $1.25 million cannot be treated as a normal variance. It should trigger a legal-reserve review. What cases drove the difference? Which were court-mandated? Which were discretionary? Which protected essential services? Which could have been settled? Which resulted from prior enforcement choices? What reserve portion was used? What future fee impact followed? Without answers, members cannot separate unavoidable defence from avoidable escalation.
Transparency should also apply to the relationship between financial reserves and address reserves. AFRINIC's Soft Landing policy and exhaustion materials show that IPv4 allocation remains rationed and that remaining pools are small. When cash reserves are depleted, pressure may rise to use fees, allocation rules, transfer restrictions or policy changes to shore up institutional position. Members should be able to see that the cash reserve and the address reserve are not being mixed rhetorically. A cash problem should not be solved by manipulating address mobility. An address-policy dispute should not be hidden as a cash-continuity issue.
The guiding principle is simple: a registry may ask members to fund insurance, but insurance cannot be secret. Members should know what shock is covered, who can claim against the fund, what exclusions apply, how claims are audited, and how premiums are reset after a loss.
Reserves should absorb shocks without dulling discipline
Institutional economics treats buffers with suspicion as well as respect. A buffer can prevent panic. It can also reduce the cost of bad behaviour. A bank capital buffer protects depositors and the payment system, but if management expects rescue without discipline it may take more risk. A university endowment protects research, but it can also insulate administrators from students and donors. A registry reserve protects the ledger, but it can also let directors and executives continue contested strategies after ordinary accountability would have stopped them.
AFRINIC's history contains the necessary ingredients for moral hazard. It holds a monopoly function for a region. Members have limited exit. IPv4 scarcity makes registry recognition economically valuable. Legal liability for wrong decisions may be slow relative to the losses members suffer. Official continuity concerns from ICANN, the NRO and peer registries can strengthen the institution's claim that it must be preserved. Reserves can finance preservation. The danger is that "preserve the registry" becomes too easily confused with "preserve the current institutional theory."
The reported address-record corruption episode shows one side of the moral hazard. If a registry does not invest in controls, staff oversight, audit trails and member notification, valuable resources can be misdirected. Reserves can fund the necessary repair: forensic review, database security, independent audit, legal recovery of improperly moved resources and stronger verification. Spending reserves for those purposes may be entirely appropriate because it protects the ledger from corruption.
The Cloud Innovation dispute shows the other side. After a corruption scandal, a registry may overcorrect by treating aggressive review and reclamation as proof of seriousness. If it has reserves, it can sustain the overcorrection longer. IGP's 2021 analysis argued that AFRINIC's move against Cloud Innovation was an overreaction to past problems and that Cloud Innovation responded with excessive legal measures. Whether one accepts every part of that view is less important than the reserve lesson: a buffer can fund both cleanup and overcompensation. Reserve policy must distinguish them before the dispute becomes existential.
Member discipline is the counterweight. In an ordinary market, customers discipline a service provider by leaving. In a regional registry, members discipline through elections, budget scrutiny, policy processes, legal rights and public accountability. When ordinary governance fails and elections are disputed, that discipline weakens. When legal strategy is opaque, it weakens further. When reserves finance the institution through the period of weak discipline, the moral hazard peaks. The registry can keep operating, which is good; it can also keep resisting pressure to narrow its mandate, which may be bad.
The correct design is not to starve the registry. Starvation would harm the members discipline is meant to protect. The correct design is conditional liquidity. Reserves should be liquid for essential services and illiquid for discretionary conflict. They should be easy to draw for payroll of essential technical staff, critical vendors, security incidents, data escrow, member notices and court-required continuity. They should be harder to draw for new litigation, expansive enforcement, political communications or programmes unrelated to the ledger. They should be hardest to draw when ordinary authority is contested, the member register is uncertain, or the spending would affect member standing.
This conditionality should be automatic enough to operate in crisis. A rule that says "the board will decide prudently" is not enough. AFRINIC's board vacuum shows why. The rule should specify thresholds: ordinary operating reserves can cover a defined number of months of core services under management certification; emergency reserves beyond that require board or receiver approval and member notice; discretionary legal spending above a defined amount requires independent legal-budget review and disclosure; reserve depletion below a target ratio triggers an automatic replenishment and cost-reduction plan; use during receivership is limited to continuity and restoration.
The reserve ratio itself should be expressed in service terms, not only years of general expenses. Two years of ordinary operating expenses is too blunt. During crisis, some expenses should fall and others may rise. A better ratio would have layers: six months of immediately accessible core-service cash; twelve to twenty-four months of restricted continuity reserve based on a crisis operating model; a separate legal contingency with annual cap and approval thresholds; and a replenishment target tied to audited expenditure. AFRINIC's own reserve explanation already recognises adjusted operating costs under crisis. The next step is to formalise the layers.
Member discipline also requires after-action accounting. Each reserve use should produce a lessons report: what triggered the drawdown, whether the event matched the policy, what services were preserved, what spending was denied, what member approvals occurred, what controls changed and how the reserve will be rebuilt. A registry that spends reserves and then merely asks for patience has not learned. A registry that treats reserve use as an accountable claim against member-funded insurance strengthens legitimacy even if the crisis was painful.
AFRINIC's future credibility will depend partly on whether it can prove this discipline. If it rebuilds cash but not rules, members will see a restored war chest. If it rebuilds rules but not cash, members will see paper resilience. If it rebuilds both under visible constraints, the reserve becomes what it should have been all along: continuity insurance that does not dull accountability.
Receivership showed continuity can be separated from institutional victory
Receivership is often described as institutional failure. In AFRINIC's case it also demonstrated an important design possibility: registry continuity can be separated from ordinary corporate control. The receiver's publicly described role was to maintain the status quo, preserve business value, oversee elections, facilitate a proper board and appoint a chief executive. Some observers treated that as private-governance remediation through rule of law. Others raised concerns about legal process, outside pressure and the role of courts. The competing views matter. Reserve policy should focus on the structural lesson.
The ledger did not need every normal institutional function to continue at full scale. It needed enough lawful authority, staff, cash, systems and external recognition to preserve essential services while governance was repaired. That is exactly what a reserve should finance. It should buy a bridge from broken governance to restored governance. It should not be tied to proving that every prior institutional position was correct.
This matters because the rhetoric of continuity can be overused. An institution under stress often says any attack on its decisions is an attack on continuity. A member under stress often says any institutional defence is self-preservation. Receivership cuts through both claims by showing a third category. The institution can be constrained while the function continues. Corporate authority can be supervised while the ledger remains recognised. Elections can be organised by an emergency authority while staff maintain services. The registry can be preserved without giving management or any faction unlimited discretion.
Reserve policy should institutionalise that third category. It should define "continuity mode" as a temporary state with narrower permitted spending, enhanced reporting and limited powers. In continuity mode, the reserve is not available for ordinary institutional ambition. It is available for essential services, court compliance, election repair, member-register verification, independent audit, minimum communications and narrowly classified legal defence. It is not available for broad policy campaigns, new discretionary programmes, expansion initiatives, or legal steps whose main purpose is to vindicate past management choices.
AFRINIC's bank-freeze episode shows why continuity mode should include banking architecture. If ordinary accounts can be frozen, a continuity reserve may need a different legal and operational setup. That does not mean the reserve should be beyond all lawful claims. It means the reserve should be recognisable as restricted to essential public-coordinate functions, so a court can consider the systemic consequences of immobilising it. A member claiming damages should not automatically gain leverage over the cash used to keep unrelated members' records stable. Conversely, the registry should not be able to hide ordinary funds behind continuity labels. Segregation must be real and audited.
Receivership also clarifies external support. Public reporting indicated that AFRINIC's troubles led the other RIRs and the NRO to prepare possible financial support, and that the wider system reviewed the policy governing the lifecycle of RIRs, including assistance and possible derecognition. Such support can be useful if it protects the numbering system. It can be dangerous if it reduces local member discipline or allows a registry to avoid hard reforms. A reserve constitution should state how external support interacts with internal reserves: whether it is repayable, what conditions apply, whether it funds core services only, and how members are informed.
Receivership should also change how settlement is valued. A solvent, confident institution may pursue legal victory to establish precedent. A registry in continuity mode should put greater weight on settlement that reduces systemic risk. If a dispute can be resolved by preserving records, clarifying member obligations, limiting future review standards, protecting customers and avoiding further reserve depletion, the settlement may serve the registry's public function better than continued litigation. Victory is not the only form of continuity. Sometimes continuity is the ability to stop fighting without collapsing the ledger.
AFRINIC's crisis has not ended every question about receivership, board legitimacy or member rights. But it has shown that the ledger can be conceptually isolated from the institution's broader battles. A disciplined reserve policy should make that isolation financial. It should make the reserve available for continuity and restoration, not for proving that any faction deserved to win the crisis.
Replenishment is a quasi-tax unless drawdowns are classified
When reserves are depleted, they must either stay depleted or be rebuilt. In a competitive market, customers can decide whether to keep buying from a firm that charges more after a strategic mistake. In a regional internet registry, members have less choice. They pay to maintain recognition, receive services, preserve standing and participate in the only regional ledger available to them. That is why reserve replenishment is economically similar to taxation even when legally it is a membership fee.
The quasi-tax risk is not that every fee is illegitimate. AFRINIC needs revenue. Staff, systems, security, audit, legal compliance, member support and service continuity cost money. The risk is that members may be asked to refill reserves after spending they could not effectively control, did not benefit from equally, and may have opposed. A small ISP that needed a /24, a university network maintaining legacy arrangements, a data centre planning customer growth, or a national research network relying on reverse DNS may all pay for the aftermath of a dispute centred on large IPv4 holdings and institutional strategy.
This is why reserve drawdown and replenishment must be paired. A drawdown decision should include a replenishment scenario before the money is spent. If the reserve is used for core continuity, replenishment can be broad because the benefit is broad. If it is used for litigation about a particular member, replenishment should identify whether costs can be recovered from the litigant, insured, budgeted over time, offset by cuts, or approved by members as a shared institutional defence. If it is used for discretionary policy enforcement, members should be told why they are bearing that risk.
AFRINIC's 2022 legal-fee disclosures show how quickly the quasi-tax problem can become real. A legal-fee total above $1.25 million is material relative to a $6 million annual revenue base. If such spending recurs, the reserve ratio changes, future budgets change, and member-funded programmes or services are displaced. AFRINIC itself later said soaring legal costs were obstructing attempts to give back to the community and strengthen membership through initiatives such as training and research. That statement is an admission of incidence: legal conflict consumes member-funded capacity.
The incidence is not only financial. Legal overhang can slow allocations, transfers, record updates, bylaw repair, elections, staff hiring and technical investment. Public reporting described AFRINIC as unable to elect a board or perform many functions from 2022 to 2025, and noted that the restored board still faced active critics and court risk. A member paying fees during such a period receives a weaker institutional product. If future fees rise to rebuild reserves, the member pays twice: once through degraded service and again through replenishment.
A reserve constitution should therefore include a member-impact statement for major drawdowns. It should answer: how much of the reserve is being used, how this affects the two-year target, which services are protected, which services are delayed, whether future fees may rise, whether small members are shielded, whether non-core spending is reduced first, and whether the spending was avoidable through settlement or narrower action. Without this statement, replenishment becomes a hidden tax.
The quasi-tax risk also changes how one should evaluate external calls for stronger registry action. When commentators urge AFRINIC to fight harder against a commercial holder, restrict transfers, defend regional resources or resist litigation, they should identify who pays if the fight consumes reserves. When member advocates urge winding-up, reconstruction or radical decentralisation, they should identify who pays for transition risk. Reserve discipline forces every side to price its preferred remedy. That is healthy. It shifts debate from slogans to incidence.
AFRINIC's situation is particularly sensitive because Africa's operators vary widely in size, resources and legal capacity. A large holder or international group can absorb legal and governance uncertainty better than a small regional network. A small operator has little ability to hedge registry risk. It cannot easily litigate, lobby or move its resource relationship. If reserve depletion leads to higher fees or weaker services, small members face the largest relative burden. Reserve policy should therefore include a small-member test: would this drawdown still be justified if its replenishment burden fell on the smallest compliant members?
That test does not make large disputes impossible. It makes them honest. A registry may decide that a legal fight is essential even if small members will help pay. But it should say what ledger function is protected and why cheaper alternatives are insufficient. It should show that non-core spending has been cut before small members are asked to rebuild the buffer. It should avoid using reserve replenishment to socialise the cost of discretionary institutional risk-taking.
Quasi-tax discipline is also a safeguard against capture by either side. A registry cannot quietly tax members for self-preservation. A powerful member cannot impose legal costs on the registry and then pretend the costs are irrelevant to others. Both AFRINIC and its antagonists should be judged by how their actions affect the common reserve. The reserve belongs economically to the continuity of the ledger and the member base, not to any faction's litigation theory.
Winding-up risk clarifies what the reserve is not
The winding-up proceedings reported in 2026 exposed another reserve boundary. ICANN intervened to tell the Mauritian court that numbering resources administered through AFRINIC are not AFRINIC assets available for distribution in a winding-up. That point is essential as a narrow factual exhibit. IP number resources in a registry ledger are not like office furniture, bank balances or receivables. They are part of a global coordination system. Treating them as corporate assets would threaten the distinction between the registry's legal shell and the public function it performs.
The same distinction should apply to reserves, but in a more nuanced way. Cash is a corporate asset in a way numbering resources are not. Yet some cash may be functionally dedicated to public ledger continuity. If AFRINIC is ever wound up, reconstructed, derecognised or transferred to another operating arrangement, the reserve should not be treated simply as surplus to be consumed by the old institution. It should follow the continuity obligation as far as law allows. The point of the reserve is not to enrich a corporate estate. It is to keep the registry function alive, wind it down safely, or transition it without harming members.
AFRINIC's own reserve explanation contemplates dissolution by saying that if a trigger event results in dissolution, all available funds shall be applied, not only built-up reserves. That statement needs a modern interpretation. Applied to what? If dissolution occurs, cash should first satisfy lawful obligations, but the reserve principle should prioritise orderly continuity: data preservation, member notices, technical transition, staff obligations, court-supervised handover, audit and protection of essential services. A reserve should not become a prize in a corporate fight any more than number resources should.
This is where reserve policy meets the wider RIR lifecycle debate. Public reporting described AFRINIC's troubles as pushing the RIR community to revise lifecycle rules to cover assistance and possible derecognition. Emergency registry support and derecognition are last-resort tools. They require finance. If the local registry's reserves are legally or practically unavailable for transition, the burden shifts to other RIRs, ICANN-related structures or members through future fees. A disciplined reserve should reduce the cost of emergency transition by setting aside continuity funds in a form that can be used for that purpose.
Winding-up risk also disciplines legal spending. If a registry spends reserves fighting dissolution but does not preserve funds for transition if it loses, it has privileged institutional survival over member continuity. It may be right to oppose winding-up. A regional registry should not be casually dissolved by a litigant, especially when numbering resources and unrelated members are at stake. But the legal defence should be paired with a contingency plan. What happens to services if the court orders restructuring? What funds remain? Who pays vendors? How are records transferred? How are member authorities preserved? What happens to pending resource requests and disputes?
The reserve constitution should therefore include a transition reserve within the reserve. It should be a floor below which funds cannot be drawn for ordinary litigation without extraordinary approval. The floor would cover a defined period of essential services plus transition costs. In normal conditions, this may seem excessive. AFRINIC's crisis proves it is not. A bank freeze, receivership, annulled election and winding-up application are not theoretical in this case. They are part of the recent record.
The transition reserve also prevents a registry from fighting to the last dollar. Without a floor, management and boards may rationally spend reserves on legal defence because collapse is unthinkable. With a floor, the institution must preserve enough cash for the members even if the institution loses. That changes incentives. It makes settlement more attractive before the floor is threatened. It makes external bodies more confident that emergency continuity can be funded. It reassures members that the reserve is theirs in functional terms, even if not in legal ownership.
Winding-up risk also clarifies language around ownership. AFRINIC and other RIRs often resist treating IP addresses as property. That caution is justified for coordination reasons. But cash reserves are not a reason to deny the economic reliance members have on registry recognition. The reserve exists precisely because recognition is important enough to require continuity funding. If the registry says members do not own addresses in the traditional sense, it should be even more careful with the member-funded cash that protects their recognised positions. Denying property language for addresses should not become a licence to use fees without accountability.
AFRINIC's winding-up disputes are therefore not separate from reserve policy. They are the extreme case that reveals the policy's purpose. A reserve is not the institution's comfort blanket, not a directors' defence fund, not an unrestricted surplus and not a substitute for legal clarity. It is a continuity asset attached morally, and perhaps eventually legally, to the ledger function.
A reserve constitution for AFRINIC
What would disciplined reserve policy look like for AFRINIC after this crisis? It would begin with a clean statement of purpose. The reserve exists to preserve the AFRINIC registry function for the benefit of resource members, affected network users and the global numbering system during severe revenue, governance, legal, operational or transition shocks. It does not exist to protect any particular board, management team, policy faction, commercial opponent or public narrative.
The second element is a layered target. AFRINIC's historic two-year operational-expense target is useful but too blunt. It should be split into ordinary liquidity, a core-service continuity layer for a reduced crisis operating model, a legal-contingency layer with caps and classification rules, and a transition floor preserved for receivership, handover, derecognition, restructuring or orderly wind-down.
The third element is permitted-use classification. Core-service continuity should be pre-approved within limits: technical infrastructure, RPKI, reverse DNS, WHOIS, RDAP, IRR, database operations, cyber security, critical vendors, essential staff, billing needed for member standing, and legally required record preservation. Governance-restoration spending should include elections, member-register verification, independent audit, receiver support and bylaw legal clarification. Legal spending should be divided into mandatory continuity defence, ledger-defence strategy and discretionary institutional litigation. Non-core programmes should be excluded during continuity mode unless separately funded.
The fourth element is prohibited use. Reserves should not fund open-ended litigation without cost bands; policy expansion during governance failure; discretionary enforcement without proportionality review; reputation campaigns detached from factual member notices; non-essential travel or conferences during reserve drawdown; replenishment without diagnosis; or legal steps that threaten the transition floor. If the board believes an exception is necessary, it should disclose the reason and seek member approval or independent emergency authorisation.
The fifth element is legal-spend discipline. Every major matter should have a litigation budget approved by the board or emergency authority, a classification, a protected function, a settlement principle and a reporting cadence. Privileged advice remains protected. The economic category does not. If legal spending exceeds a threshold, the board must explain why settlement, narrowing, mediation or external dispute resolution is insufficient. If legal spending delays core services or non-core programmes, members should see the trade-off.
The sixth element is crisis drawdown procedure. When a trigger occurs, AFRINIC should enter continuity mode. Triggers should include a bank freeze, revenue collapse, governance failure, receivership, court order, cyber incident, critical vendor loss, major legal threat or service-continuity risk. Management certifies the trigger; the board or receiver confirms authority; members receive a notice; spending shifts to permitted categories; non-core spending is suspended; and a public reserve-status report follows within a fixed period. If no lawful board exists, the receiver or a pre-defined emergency committee can authorise only core and restoration spending.
The seventh element is replenishment discipline. Drawdowns should be replenished according to their category. Core-service drawdowns can be rebuilt through general fees over time. Governance-restoration drawdowns may be general but should include reform milestones. Discretionary legal drawdowns should require a specific member-approved replenishment plan and identify recoveries or offsets. If a drawdown resulted from avoidable management or board decisions, replenishment should be paired with accountability measures, not merely higher charges.
The eighth element is segregated custody. Strategic reserves should be held in designated accounts or structures that are clearly separate from operating cash, with signatory controls, audit visibility and a statement of restricted purpose. AFRINIC's published policy already distinguishes reserves from normal operational accounts; the crisis suggests that distinction should be stronger and more legible to courts. The goal is not to evade lawful claims. It is to ensure that essential ledger-continuity cash is recognised as such before a freeze or garnishment threatens unrelated members.
The ninth element is member reporting. AFRINIC should publish an annual reserve report with the target, balance, restricted and unrestricted cash, drawdowns, legal classifications, replenishment status, stress tests and service-continuity mapping. During continuity mode, it should publish shorter updates. The report should be written for operators, not accountants alone. Members should be able to tell whether the reserve protects their services, finances litigation, funds restoration or waits for transition.
The tenth element is independent review. After the crisis period, an independent reviewer should assess whether reserve use matched the policy. The review should evaluate legal classifications, approvals, service outcomes, member notices, vendor continuity, replenishment and governance lessons. It should identify spending that should not recur. This is how reserves become discipline rather than memory loss.
The institutional lesson beyond AFRINIC
AFRINIC is not merely an African exception. It is a visible case of a general problem in scarce-resource ledgers. When a private, member-based technical institution administers records that support economic value, its reserves become part of governance. The old language of nonprofit prudence is too small. Cash reserves decide how long the institution can withstand shocks, how long it can resist critics, how much legal risk it can take, and how much future cost it can impose on members.
The general rule is that registry reserves should be tied to a narrow ledger mandate. The narrower the mandate, the easier it is to justify a strong reserve. Members can accept funding two years of continuity if they know the cash protects records, security, essential staff and lawful transition. They will be less willing, and should be less willing, to fund reserves that can be used for broad policy wars, discretionary market control or institutional self-defence. A strong reserve and a narrow mandate reinforce each other. A strong reserve and a broad mandate create moral hazard.
AFRINIC also shows that continuity and accountability are not opposites. Some official statements during the crisis understandably emphasised stability: the registry must keep operating, staff must be supported, members must receive services, and numbering resources must not be treated as corporate assets in a winding-up. Those are correct continuity concerns. But continuity without member discipline is brittle. It can preserve the shell while the legitimacy deficit widens. The goal is not to choose between continuity and accountability. The goal is to use reserves to buy time for accountability to be restored.
The settlement-utility model is the best guide. A registry under stress should act like a clearing house for recognised number-resource positions. It should preserve the ledger, process objective changes, flag disputes, comply with courts, protect third parties and minimise discretionary shocks. Its reserve should support that settlement role. It should not finance an ambition to decide the entire economic future of IPv4 in the region. It should not subsidise member capture. It should not protect corrupt records. It should not treat every legal challenge as sabotage or every commercial holder as illegitimate.
AFRINIC may still stabilise. A functioning board, credible finance reporting, clarified bylaws, disciplined legal budgets, verified member authority, reserve segregation and settlement-oriented litigation could reduce the risk premium attached to AFRINIC-administered resources. The region needs that outcome. African networks should not bear a permanent governance discount because their registry became a battlefield for scarcity, litigation and institutional legitimacy.
But stabilisation should be measured by harder signs than survival. Does AFRINIC publish current reserve balances and legal classifications? Does it identify a transition floor? Does it cap reserve-funded discretionary litigation? Does it separate core services from non-core programmes? Does it explain replenishment burdens? Does it protect small members from crisis costs they did not create? Does it show that anti-corruption repair is targeted at ledger integrity rather than broad discretionary control? Does it make settlement possible without rewarding litigation pressure? These are the tests of reserve policy discipline.
An operator asking whether AFRINIC's reserves protect the ledger or protect the institution from members is asking the right question. The answer should not depend on trust in a board, a receiver, a litigant, ICANN, the NRO, a public campaign or a court alone. It should be written into the reserve constitution. Cash held by a registry is not neutral. In a scarcity institution, cash is power over time. AFRINIC's crisis has shown that power can keep essential records alive. It can also keep unresolved authority alive. Reserve discipline is the difference.