A National Internet Registry sounds, at first, like a modest administrative convenience. A regional registry is far away; a local registry is closer. A member can speak in a familiar language, pay through a familiar banking channel, submit company documents in a familiar form, and deal with staff who understand the domestic telecom market. In a region as large and uneven as Africa, where one registry incorporated in Mauritius serves operators across many legal systems, currencies, languages, licensing regimes and levels of administrative capacity, the attraction is obvious.
The economics are less comfortable. A national layer between AFRINIC and resource holders does not merely reduce distance. It creates another place where fees can be collected, documents interpreted, transfers slowed, national policy smuggled into regional rules, incumbent operators given influence, and members left unsure who is responsible when records, reverse DNS, routing-security services or transfer approvals are delayed. The local office that makes access cheaper can also become the local gate.
This is not an argument that AFRINIC already has a mature National Internet Registry system comparable to the best-known Asia-Pacific arrangements. The public record reviewed for this piece does not support treating such a structure as an established AFRINIC fact. The question is prospective and institutional. What would happen if national registry relationships, national validation partners or national administrative intermediaries became part of the African number-resource bargain?
APNIC supplies the obvious reference point, but only as a reference point. Examples historically or publicly associated with APNIC-region NIR arrangements include local institutions in Indonesia, China, India, Japan, Korea, Taiwan and Vietnam. They are not APNIC branch offices. They operate under local law, serve local communities, use local languages, maintain local relationships and sit inside a regional registry system. Their existence shows why national intermediation can be useful in a diverse region. It also shows why NIRs are not merely help desks. They are part of market structure.
The African question is therefore not whether national support is desirable in the abstract. It often is. The question is whether AFRINIC can use national intermediaries without losing ledger neutrality. A regional internet registry has value because it maintains a common record across borders. The scarcer IPv4 becomes, the more that record resembles settlement infrastructure. Buyers, sellers, lessors, lenders, data centres, telecom groups, public networks and courts all need to know who is recognised for which resources and which registry-controlled services follow from that recognition. If the common record becomes dependent on national discretion, the region gains proximity but loses uniformity. If it refuses every national layer, it may preserve formal uniformity while remaining practically remote from many operators.
AFRINIC's recent history makes the trade-off harder. Public reporting and institutional statements have described the Cloud Innovation dispute, litigation in Mauritius, receivership, bank-account restrictions, attempted elections, concern about powers of attorney and voter documentation, the annulment of a June 2025 election, a later board election, ICANN involvement and continuing arguments over the registry's future. Those facts do not prove misconduct by any current actor. They do prove that AFRINIC's governance environment is not a high-trust background condition. In such an environment, every new layer is interpreted through suspicion: who appoints it, who pays it, who audits it, whose policy it applies, whose records prevail, and whether it quietly gives a ministry, incumbent carrier, industry association or organised slate a new way to influence scarce-number governance.
The institutional-economics problem is simple to state. Local service can lower transaction costs. Local authority can raise them. A National Internet Registry is useful only if it reduces the cost of accurate recognition without becoming a national sovereign over the regional ledger.
A national layer solves one problem by creating another
The basic case for a national internet registry is transaction-cost reduction. Many resource holders are not global platforms with lawyers, regulatory staff and registry specialists. They are access providers, universities, data-centre operators, public-sector networks, enterprise service providers, wireless ISPs, IXPs, regional carriers, hosting firms and small companies whose administrative office may be one person with a router console, a billing system and an overloaded inbox. For them, dealing directly with a regional registry can be expensive even when the published fee is modest.
The costs are familiar. A member must understand AFRINIC's forms, policy vocabulary, account relationship, billing cycle, corporate-authority expectations, transfer rules, reverse-DNS procedures, RPKI dependencies and support channels. It may need to translate company documents. It may need to explain local licensing categories to a reviewer outside the country. It may need board resolutions, ministry letters or public-institution approvals in a form that matches a foreign registry's expectations. It may need to pay in foreign currency. It may need to follow English-language policy discussions even if its operating language is French, Arabic, Portuguese or another language. None of these costs is exotic. Together they decide whether a formally open registry is practically accessible.
An NIR can reduce many of these frictions. It can validate a domestic company's existence more cheaply than a regional office can. It can understand local telecom licences, company registries, public-sector signatory rules and customary corporate documents. It can speak the local language. It can collect fees through domestic banking channels. It can answer questions before they become formal tickets. It can explain regional policy in operational examples that local networks recognise. It can maintain relationships with national operator groups and regulators. It can help small networks avoid mistakes that would otherwise become delays, disputes or failed requests.
But intermediation has its own costs. The same layer that helps validate can also filter. It can become another fee collector. It can impose local forms on top of regional forms. It can delay a transfer while deciding whether a domestic policy concern exists. It can favour well-known incumbents whose staff sit on national committees. It can treat resource movement as a matter of national economic policy rather than record accuracy. It can keep downstream members dependent on a national institution without giving them the same direct corporate voice in the regional registry. It can make members unsure whether a rule comes from AFRINIC, from the NIR, from a telecom regulator, from tax law, from currency controls or from local politics.
The design question is therefore not "centralisation or decentralisation". It is what rights and duties move to the national layer. A local-language help desk is one thing. A domestic verifier of corporate documents is another. A local fee collector is another. A national approval authority for transfers is another. A state-linked body that can block outbound movement of IPv4 resources in the name of national interest is something else entirely. Each step adds convenience for some users and discretion for some institution.
An NIR model is therefore neither cure nor disease. It is a trade. It buys local knowledge by selling some simplicity. In AFRINIC's contested environment, that price is acceptable only when the regional ledger remains authoritative, rules are mapped clearly, fees are transparent, and national discretion is tightly bounded.
The regional ledger cannot be nationalised by convenience
The first principle is that the regional ledger must remain regional. AFRINIC exists because internet number resources require common recognition across borders. An IPv4 address, IPv6 prefix or autonomous system number is not useful because a national ministry likes it. It is useful because networks, registries, security systems, routing databases, reverse-DNS delegations, transfer counterparties and public accountability tools can rely on a shared record that avoids duplicate claims and identifies operational responsibility.
That record is not a national asset register. It is a coordination ledger. Its job is to make a globally unique resource legible to a region and to the wider internet. A national institution may help keep the record accurate, but it should not convert the record into a national permission system. This distinction matters because the language of local service can easily slide into the language of local sovereignty. The local registry knows the market; therefore it should approve requests. It collects the fee; therefore it should decide standing. It sees domestic need; therefore it should determine whether resources may leave. It hears from the regulator; therefore it should align registry recognition with telecom policy. Each step is plausible. Combined, they nationalise a regional ledger by administrative drift.
The Asia-Pacific comparison is useful precisely because it shows that national registry institutions can exist without formally replacing the regional registry. APNIC-region NIRs sit inside a regional system. Their legitimacy depends on connecting national service to regional recognition. They do not make IP addresses the sovereign property of a state merely because a domestic institution helps administer member service. The hard part is not the existence of a local office. It is the boundary between local administration and regional authority.
For AFRINIC, that boundary would need to be explicit before any national model became operational. If a national partner validates a company certificate, what is the legal effect of that validation? If it collects a fee, does non-payment affect only local service or AFRINIC standing? If it maintains a local account file, is the AFRINIC database still the authoritative record? If it receives a transfer request, can it reject the request or only advise AFRINIC? If a local regulator asks it to delay a transfer, must it obey? If a court in the local jurisdiction orders a freeze, how does that interact with a Mauritian court, AFRINIC bylaws and global registry expectations? If the local record and the regional record diverge, which one do counterparties rely on?
These questions are not theoretical. They are the questions that markets ask before pricing a resource. A buyer does not want to discover after signing that a national intermediary can delay recognition for reasons not visible in AFRINIC's policy manual. A seller does not want to learn that a domestic body can impose a local capital-control view on a globally routable resource. A bank or acquirer does not want to diligence two inconsistent ledgers. A small operator does not want to pay a local fee and then find that the regional account is not in good standing. A court does not want a registry that cannot explain who had authority to change the record.
Ledger neutrality does not mean AFRINIC must ignore national facts. A domestic court order, corporate dissolution, telecom licence cancellation, public-sector signatory rule or fraud report may be relevant evidence. Evidence is not the same as veto. The regional registry can accept national evidence while still applying regional rules. It can recognise that a local company no longer exists, that a merger is legally effective, that a representative lacks authority, or that a disputed claim requires a hold. It should not let a national intermediary turn every request into a test of domestic industrial policy.
The regional ledger's value comes from being narrower than national politics. It should answer questions such as: who is the recognised holder, who is authorised to act, what resource is involved, what records are current, what policy applies, whether there is a dispute, and which registry-controlled services follow. It should not answer whether a domestic market has enough local hosting, whether a foreign buyer is politically attractive, whether a national champion deserves priority, or whether address capital should be retained for strategic reasons unless a clear, lawful and regionally adopted policy says so.
If AFRINIC ever relies on national intermediaries, the design should begin with this sentence: the NIR assists the ledger; it does not own the ledger.
Delegation is not sovereignty transfer
The political temptation around National Internet Registries is to treat delegation as sovereignty transfer. A national institution administers number-resource requests; therefore the state, or a state-adjacent body, may feel it has acquired a voice over the resources associated with its territory. That assumption is understandable. It is also dangerous.
Governments increasingly see digital infrastructure as strategic. Telecom networks carry emergency communications, public services, mobile money, banking, health systems, education platforms, cloud adoption, government identity systems and national-security traffic. IPv4 scarcity, even during IPv6 transition, makes addresses feel like part of a national capacity stock. A telecom regulator may ask why addresses assigned to domestic operators should be sold abroad. A ministry may ask why a regional registry in Mauritius should decide whether a national provider can obtain more capacity. A competition authority may worry that large carriers, hosting firms or foreign platforms can outbid smaller local networks. A finance ministry may notice address transfers as cross-border value movement. Once the resource looks scarce and valuable, state interest follows.
That interest should not be dismissed. Public authorities have legitimate concerns about telecom licensing, consumer continuity, fraud, critical infrastructure, tax compliance, sanctions, insolvency, public-sector networks and national connectivity. A registry model that treats national regulators as irrelevant will fail in practice. But the answer cannot be to convert each national concern into registry discretion. Internet number resources are globally coordinated identifiers, not domestic spectrum licences. They are administered regionally because their uniqueness and recognition require a common system. The fact that a holder is based in a country does not make the resource subject to separate national registry sovereignty unless the regional framework has defined such a role.
Delegation is an agency arrangement. The national layer may perform services: intake, translation, validation, fee collection, local education, document mapping and support. It may feed local evidence into AFRINIC's decision. It may help maintain data quality. But unless the system is willing to abandon the regional ledger, the national layer should not acquire a sovereign right to decide how recognised resources move.
The distinction is most important for transfers. A national regulator may think outbound transfers reduce local digital capacity. It may ask the NIR to block them. It may be tempted to treat IPv4 addresses like foreign currency or strategic equipment. Yet a regional transfer wall can reduce the value of resources held by local networks and discourage inbound supply. If a small African provider has unused address space, the ability to sell or lease it may finance equipment, fibre, security or IPv6 transition. If a foreign holder fears that imported space will be trapped in a national or regional policy cage, it will demand a discount or avoid the market. A protectionist instinct can therefore make the protected market poorer.
Assignments carry the same risk. A state-backed NIR could be tempted to prioritise licensed telecom operators over hosting firms, enterprise networks, research networks or smaller access providers. It might see national champions as more deserving of scarce administrative attention. It might treat resource requests as part of telecom-sector planning rather than registry proof. That would import competition policy into registry administration without the safeguards that competition law normally requires.
AFRINIC's legitimacy depends on resisting this slide. The registry can cooperate with states without becoming their instrument. It can ask whether an applicant exists, whether it is authorised, whether it has accurate contacts, whether it accepts responsibility, whether a transfer is fraudulent or disputed, and whether policy conditions are met. It should not ask whether a national development plan would prefer the resource to be held by another domestic actor, whether an incumbent's market share makes its request more respectable, or whether a regulator's discomfort is enough to override a valid transaction.
If a state wants to regulate address-related transactions as a matter of domestic law, it can do so openly through legislation, regulation and courts, subject to the consequences. The registry should not hide that choice inside NIR practice. Hidden sovereignty is worse than explicit law because members cannot price it, challenge it or design around it. A registry that cannot say whether a delay came from regional policy or national pressure will lose trust from both markets and governments.
The rule should be plain: national delegation may improve evidence, language and payment. It must not transfer sovereignty over the regional ledger by implication.
Local service is a real economic good
The case against overbroad national authority should not obscure the case for local service. In Africa, local knowledge is not a luxury. It is often the difference between a registry record that reflects operational reality and one that reflects only the ability of a member to navigate a remote bureaucracy.
Corporate evidence varies widely. A private ISP, a state-owned telecom, a university network, a public ministry, a bank, a data-centre company and a non-profit IXP may have different authority chains. Some can produce board resolutions quickly. Some need ministerial approval. Some have directors whose names differ across languages or scripts. Some have legacy records tied to older company names. Some rely on licences issued by telecom regulators rather than general corporate registries. Some public institutions cannot sign commercial forms without procurement or treasury approval. A regional reviewer can learn these patterns, but a credible local institution may know them faster and with less friction.
Payment is another real service. Foreign exchange, correspondent banking, invoices, taxes, public-sector procurement and currency volatility can make regional billing costly for smaller networks. A national intermediary may collect in local currency, smooth payment timing, issue invoices that fit domestic accounting, and reduce the risk that a member falls out of standing because the banking path failed. This is not a minor issue. If good standing affects transfer eligibility, support access or confidence in registry recognition, payment friction becomes operational risk.
Local validation can also improve fraud control. A domestic institution may know whether a company is defunct, whether a signatory is plausible, whether a regulator has revoked a licence, whether a merger is publicly known, whether a court-appointed administrator exists, or whether a claimed representative is an outsider. It can detect problems a remote office might miss. In a scarcity market, where stale or dormant records can become valuable targets, this is a genuine ledger benefit.
The question is how to capture the benefit without creating a local monopoly over member voice. The answer is to treat local service as evidence production, not as final judgment. A national intermediary can certify that a document is authentic, that a company exists, that a signatory matches domestic records, that a licence category means what the applicant says it means, or that a translation is accurate. AFRINIC can then apply regional policy to that evidence. This preserves the value of local knowledge while keeping the decision path reviewable.
Local service should also be optional where possible. If a member wants to deal directly with AFRINIC and can meet regional requirements, the national layer should not be a compulsory toll unless the system has a strong reason. Compulsion changes the economics. It allows the NIR to collect rents, creates dependence on local politics, and forces members into a relationship they may not trust. In some APNIC-region arrangements, organisations have historically been able to choose between direct APNIC service and the local NIR path, though not to obtain resources through both at the same time. The useful lesson is not the exact operational detail; it is that choice can limit the power of the intermediary.
The best version of local service is boring: better forms, better translation, faster document checks, clearer invoices, lower payment risk, local examples, and support staff who understand domestic reality. The worst version is a national checkpoint with a help-desk label. AFRINIC's challenge would be to institutionalise the first and block the second.
Fee collection becomes fiscal federalism
NIR relationships turn registry fees into a form of fiscal federalism. Money flows from local networks to a national institution, from that institution to the regional registry, and sometimes from both into local programmes, reserves, taxes, support costs, training, meetings, exchange-rate buffers and administrative overhead. The published regional fee is no longer the price the operator experiences. The operator experiences a composite bill.
APNIC-region materials illustrate the issue without settling how AFRINIC should act. They describe local NIR fee schedules, local currency treatment, regional charges and formulas by which NIR-related members are counted or charged. The details are specific to the APNIC system and may change. The economic point is more durable. A downstream member in an NIR environment may not know how much of its bill reflects regional registry cost, how much reflects local registry operation, how much reflects domestic tax, how much reflects currency hedging, how much reflects development programmes, and how much reflects cross-subsidy among local members. The regional registry may know what it charges the NIR, but not how the charge is passed through. Other members may not know whether the NIR arrangement is cost recovery, political compromise, risk pricing or hidden subsidy.
For AFRINIC, this would be combustible. Fee politics are already sensitive because IPv4 scarcity makes membership standing economically meaningful. A bill that once looked like association cost can now affect transfer confidence, account status, service continuity and member trust. If an NIR adds another layer of fees or local consequences, the question "what am I paying for?" becomes harder to answer.
The opacity has several forms. There is pass-through opacity: the member cannot tell which part of the local fee comes from AFRINIC. There is function opacity: the member cannot tell whether the fee funds core registry service, local programmes, legal risk, training, reserves or national policy work. There is standing opacity: the member cannot tell whether local non-payment affects only local service or the regional account. There is timing opacity: the member cannot tell whether the NIR has paid AFRINIC even after the member has paid the NIR. There is subsidy opacity: the member cannot tell whether it is subsidising smaller networks, larger incumbents, public-sector bodies or the NIR itself.
Currency risk makes the issue sharper in Africa. A local member may pay in domestic currency while AFRINIC's costs, reserves or invoices are denominated partly in foreign currency. If the currency moves, who bears the loss: the NIR, the member, AFRINIC or future fee schedules? If capital controls or banking restrictions delay payment, is the member treated as delinquent? If a public-sector network cannot complete foreign payment within the billing window, does local payment to the NIR preserve standing? These are not accounting trivia. They decide whether fees become operational risk.
The answer is visible mapping. Any AFRINIC-related national registry arrangement should publish a fee map: what AFRINIC charges, what the NIR charges, which local taxes or mandatory charges apply, what currency conversion rule is used, what reserves are held, what happens if the NIR fails to remit, what services are funded locally, what hardship or payment-plan options exist, and what standing consequences apply at each stage. Downstream members should not need forensic accounting to understand whether they are paying for regional registry service, local administration, domestic policy, currency risk or institutional overhead.
The fiscal principle should be the same as the ledger principle: local collection may assist the regional system, but it must not turn the registry relationship into an opaque tax.
Transfer authority is where the model becomes expensive
Transfers are the hardest test for any NIR relationship. Routine support can be localised without much systemic risk. Transfer authority cannot. Once IPv4 is scarce and priced, the ability to approve, delay, condition or block a transfer changes economic value. A national intermediary involved in transfers is therefore not merely a service provider. It becomes part of settlement.
The APNIC region shows the complexity at a high level. Transfer paths can involve regional accounts, NIR-related accounts, transfers between national environments, transfers between an NIR environment and the regional registry, and inter-RIR transfers involving another regional registry. Each path requires proof of holder authority, recipient eligibility, resource status, policy compatibility, documentation, record updates and operational continuity. In the best case, the national layer helps by explaining local documents and confirming domestic authority. In the worst case, the path becomes a toll road whose rules are known only to specialists.
For AFRINIC, where transfer policy is already politically charged, NIR involvement would require exceptional precision. The first question is whether the NIR can approve a transfer or only recommend. If it can approve, then regional policy has been partly delegated. If it can only recommend, the member needs to know whether AFRINIC will normally defer to the NIR. Informal deference can be as powerful as formal authority. A national body that knows AFRINIC rarely contradicts it effectively holds a veto even if the documents say otherwise.
The second question is whether NIR handling changes the standard. A direct AFRINIC member and an NIR-mediated member should not face materially different transferability unless the difference is published and justified. If a block held through a national path is harder to move, it will trade at a discount. If it is easier to move because local insiders can navigate the process, other members will see unfairness. If the national layer adds special domestic approval for outbound movement, the resource begins to carry a national lock.
The third question is timing. Transfer markets operate on commercial time. Escrow conditions expire. Buyers need capacity for customers. Sellers need liquidity. Acquisitions have closing dates. A national layer can speed evidence collection, but it can also add another queue. If a transaction needs local NIR review, AFRINIC review and perhaps another RIR's review, uncertainty compounds. The market will price the longest uncertain stage, not the shortest published step.
The fourth question is status continuity. A transfer is not complete merely because a database row changes. Associated records and services matter: RDAP, WHOIS, reverse DNS, route objects where relevant, RPKI certificates or ROAs, abuse contacts, account access and public transfer logs. If an NIR maintains some local data while AFRINIC maintains the regional record, the handoff must be explicit. A buyer should know which records will be deleted, recreated, migrated or left unchanged. A seller should know when responsibility ends. A lessee or downstream user should know whom to contact for abuse and routing-security issues. Operational delivery is part of transfer quality.
The fifth question is appeal. If the NIR says no, can the member appeal to AFRINIC? If AFRINIC says no based on NIR advice, can the member inspect the reason? If a national regulator influenced the decision, is that disclosed? If the NIR delays without a formal denial, when can the member escalate? A transfer market cannot function if denial can hide inside waiting.
The safest design is a mapped settlement path. For each transfer category, the system should state the roles: source authority proof, recipient eligibility, NIR evidence, AFRINIC decision, other-RIR coordination, service-transition steps, timelines, pause reasons, appeal routes and public reporting. The more the path depends on specialist memory, the more value moves from resource holders to brokers, insiders and repeat players. That may be profitable for intermediaries. It is not good registry design.
NIRs can reduce transfer cost when they supply local evidence and predictable steps. They increase transfer cost when they add discretion and uncertainty. AFRINIC would need to judge them by that economic outcome, not by the attractiveness of decentralisation language.
Local validation can become protectionism
The strongest practical argument for an NIR is local validation. The strongest political risk is that validation becomes protectionism. The boundary is thin.
Validation asks factual questions. Does this company exist? Is this person authorised? Is this licence current? Does this court document mean what the applicant says it means? Is this merger legally effective? Is this public institution able to sign? Is this local address or tax identifier real? Has this resource holder changed its name? Are the records stale because of normal domestic practice or because someone is trying to obscure control? These questions improve the ledger when answered well.
Protectionism asks different questions. Should this block leave the country? Should this buyer be allowed to acquire capacity when a local incumbent wants it? Should a foreign-owned data-centre company receive resources before a domestically owned ISP? Should addresses be reserved for networks that fit a national industrial plan? Should a holder be allowed to lease addresses to customers abroad? Should a cross-border cloud business count as local development? Should an operator with a disputed political reputation receive recognition? These questions may be important to governments, but they are not ordinary registry validation.
The danger is that protectionist questions can be disguised as validation defects. A local body may request more documents, question use plans, ask for regulator letters, delay confirmation, insist on domestic customer evidence, or treat outbound transfers as suspicious. Each request can be defended as prudence. The cumulative effect is capital control. The resource remains formally transferable, but only through a channel whose cost and uncertainty favour local incumbents or politically preferred actors.
Africa's telecom structure makes this risk material. Many national markets have strong incumbents, state-linked operators, licence bottlenecks, limited wholesale competition or regulator relationships that matter for market entry. A national registry hosted by, dependent on, or heavily influenced by the dominant telecom circle could unintentionally reproduce that structure inside number-resource administration. A small ISP that already struggles for interconnection, spectrum, ducts or wholesale access should not also have to seek number-resource approval from a body socially aligned with its competitors.
National champions create the same problem. Governments may view large domestic carriers, state-backed digital platforms or favoured infrastructure companies as strategic vehicles. They may expect scarce addresses to support those champions first. A registry that serves the public internet cannot simply ignore smaller operators, hosting firms, enterprise networks, IXPs, universities or independent data centres because they are less politically central. Number-resource governance should not become an instrument for choosing industrial winners unless a lawful public policy says so openly and accepts the cost.
The antidote is purpose limitation. Local validation should be tied to specified evidence categories. A national intermediary may confirm identity, authority, document meaning, contactability, licence status where relevant, and local-language accuracy. It should not decide commercial merit, national strategic priority, customer geography or acceptable business model unless regional policy directly and narrowly requires such a finding. If a regulator supplies information, the category should be visible: legal prohibition, court order, licence fact, tax issue, national-security order, or non-binding policy preference. Only the first few categories may justify registry consequences.
Publication also matters. AFRINIC and any NIR should report aggregate delay reasons: missing authority, translation defect, unpaid fee, disputed resource, court hold, policy ineligibility, regulator inquiry, fraud investigation, member non-response, or other. If too many cases sit in vague "local review", members will correctly infer hidden discretion. If denials and delays are categorised, protectionism becomes harder to hide.
Good validation produces cleaner records. Bad validation produces a local permission market. AFRINIC's task would be to make the first cheaper than the second.
Regulators will see scarce addresses as industrial capacity
AFRINIC cannot design NIR relationships as if telecom regulators do not exist. In most African markets, connectivity is a regulated sector. Operators hold licences. Regulators manage numbering, spectrum, interconnection, quality of service, consumer protection, lawful-interception duties, SIM registration, emergency communications, universal-service funds and sometimes data-centre or gateway matters. Even when IP addresses are not assigned by the telecom regulator, they sit near regulated activities.
Scarcity changes the regulator's perception. When IPv4 looked like technical administration, it could be left to engineers. When it becomes priced, transferable, leased and embedded in national connectivity, it looks like capacity. A regulator may see an address transfer as equivalent to a sale of scarce national input. It may worry that foreign buyers will drain local networks. It may worry that unlicensed operators can obtain capacity. It may worry about abuse, fraud, cybersecurity, data localisation, law-enforcement contactability or the ability of public services to remain reachable. These concerns are not imaginary.
The problem is instrument choice. A telecom regulator has tools: licences, reporting duties, merger review, spectrum conditions, consumer-protection rules, cybersecurity obligations and court processes. A regional internet registry has different tools: record accuracy, resource delegation, transfer recording, contact data, reverse DNS, RPKI and member standing. Mixing them can create overreach. If an NIR becomes the bridge, it may be pressured to enforce telecom policy through registry tools.
Capital-control instincts are especially likely around outbound transfers. A country short of IPv4 capacity may want to keep addresses inside national borders. Yet address retention is not the same as connectivity development. A block held by an inefficient incumbent but barred from sale may not serve consumers better than a block sold, leased or financed into a more productive network. A foreign buyer may pay a price that lets a domestic operator invest in fibre or equipment. A domestic lessee may need temporary capacity that an outright retention policy would not create. The registry should not assume that immobility equals national benefit.
The same applies to inbound transfers and leasing. If a national system is seen as sticky or politicised, address capital will hesitate to enter. If the official path is too rigid, actors will use informal arrangements with weaker abuse, lessee and routing-security information. A registry can support regulator accountability through accurate contacts, clear responsibility and cooperation when legal orders exist; it does not need to turn every movement of address space into national economic review.
AFRINIC's role should be to translate regulator concerns into ledger-relevant categories. Fraud is ledger-relevant. Legal prohibition is ledger-relevant. Contactability is ledger-relevant. A valid court order is ledger-relevant. A general preference that resources stay with national champions is not ledger-relevant unless embodied in lawful, reviewable rules. A registry that cannot make that distinction will become a quiet arm of national economic policy without the legitimacy, expertise or liability of a state.
The healthier relationship is one of structured cooperation. Regulators can provide factual evidence and receive accurate information. NIRs can explain registry rules to domestic authorities. AFRINIC can maintain the common ledger. None of them should pretend that local administrative convenience has created a national property regime over globally coordinated identifiers.
National champions turn registry architecture into competition policy
Every NIR model must answer a competition question: who controls the national intermediary? In theory, the answer is a neutral local institution. In practice, national internet communities are not flat. Incumbent telecom operators, mobile groups, state-owned carriers, large ISPs, government agencies, IXPs, universities, data-centre groups and industry associations have different resources and different incentives. Whoever dominates the local institution can shape how national intermediation is experienced.
The risk is subtle. A dominant operator does not need to seize the registry. It only needs to make the national layer more comfortable for actors like itself. Documentation expectations can mirror large-company practice. Meeting times and committees can favour firms with policy staff. Fee structures can be painless for large resource holders but regressive for small networks. Transfer review can be strict for new entrants and routine for familiar incumbents. Local validation can treat unusual business models as suspicious. Regulator relationships can privilege licensed carriers over hosting, cloud, campus, community or enterprise networks.
This matters because number resources can affect competition. IPv4 capacity supports customer acquisition, hosting density, dedicated services, enterprise contracts, security products and migration strategies. A network that can obtain or move address space cheaply has more options. A network that must wait, lease informally, use more NAT, or depend on an upstream's addresses has fewer options. If national intermediation raises fixed costs, large incumbents benefit even when no rule says "favour incumbents".
National champions also have stronger political narratives. They can say they serve national development, rural coverage, government connectivity, emergency services or strategic digital infrastructure. Those claims may be true. They should not automatically outweigh the claims of smaller operators that create competition, local hosting, regional resilience or specialised services. A registry's job is not to rank business virtue. It is to maintain accurate, neutral recognition under policy.
An NIR hosted by an industry association may face a related problem. Associations are useful because they aggregate member voice. They are risky because their active members are not always representative. Large companies attend more meetings, pay larger dues, supply committee members and understand procedure. Smaller firms may rely on the association for help while lacking the capacity to monitor whether the association's registry role serves them. Representation becomes bundled with administration.
Conflict rules are therefore essential. Any AFRINIC-related national intermediary should disclose governance composition, funding sources, voting rules, staff conflicts, links to dominant operators, regulator ties and decision roles. Staff or board members connected to a transfer party should be recused. Aggregate service metrics should be public by request type and member class. Appeals should go to AFRINIC or an independent regional mechanism, not only to the same national committee. Fee waivers and hardship treatment should follow criteria, not relationships.
Market participants will look for signs of capture. Do small operators receive timely support? Are transfer delays concentrated among new entrants? Are local fees more burdensome for modest holders? Are outbound transfers stigmatised? Are leasing arrangements treated differently depending on who provides them? Are national champions overrepresented in governance? Are regulator letters required in cases where policy does not demand them? These patterns matter more than mission statements.
The point is not that national institutions are doomed to capture. Some may be more responsive than a remote regional office. The point is that national institutions concentrate local political economy. AFRINIC should not assume that because an intermediary is closer to members, it is closer to all members equally.
Data quality is the price of intermediation
The technical risk of an NIR model is duplicated truth. A regional registry maintains the authoritative database. A national intermediary maintains local member records, local billing records, document files, support history and perhaps local assignment details. If the two are perfectly synchronised, intermediation improves service. If they diverge, every party must ask which record is real.
Data quality is not only a technical housekeeping issue. It is market infrastructure. A buyer wants to know whether the seller is recognised. An abuse desk wants the correct contact. A network operator wants reverse DNS to point where expected. A route-origin system depends on resource certification tied to recognised control. A regulator wants to know who is accountable. A court wants to know which record reflects authority. A small ISP wants its account status to survive a local payment or name-change update. If the local and regional records disagree, confidence falls.
The risk has several forms. A member pays the NIR, but the regional account remains unpaid. A local contact is updated, but the AFRINIC record remains stale. A local assignment is recorded, but regional visibility is absent. A transfer is approved locally, but AFRINIC has not recognised it. A local court or regulator letter is stored by the NIR, but no dispute flag appears in the regional record. A local staff member knows the real operator, but public RDAP data points to an old entity. A resource holder believes a lease or delegation is locally accepted, but the regional policy treats it differently.
Each divergence creates a different cost. Some costs are operational: mail reputation, reverse DNS, routing-security errors, abuse misrouting, failed automation. Some are transactional: delayed transfer, escrow extension, extra warranties, lower price. Some are legal: uncertainty over authority, reliance, member standing or dispute status. Some are political: claims that the NIR favours certain actors or hides information from AFRINIC.
AFRINIC would need data rules as strict as fee and transfer rules. The regional database should remain the authoritative public record. NIR-maintained local records should have defined status: supporting evidence, local service record, assignment detail, billing file, or public registry data. Synchronisation intervals should be stated. Critical changes should have confirmation receipts. Members should be able to see what the NIR has sent to AFRINIC and what AFRINIC has accepted. If the NIR rejects or delays a change before it reaches AFRINIC, the member should know that the delay is local, not regional.
Operational services require special care. RPKI, reverse DNS and routing-registry data should not be vulnerable to local administrative ambiguity unless the ambiguity directly concerns authority over the resource. A fee dispute with the NIR should not casually disrupt security assertions used by relying networks. A local corporate-name update should not break reverse DNS. A disputed transfer should be isolated from unrelated resources. A registry's discipline is shown by how well it separates the risky item from the stable record.
Auditability is the key. AFRINIC and any NIR should maintain a traceable chain for resource changes: request, evidence, local validation, regional decision, public record update, service transition and notification. This does not require publishing private documents. It requires knowing, and where appropriate reporting, which institution did what and when. Without that chain, disputes become narrative contests.
Data quality is the cost that NIR enthusiasts often understate. Local service is useful only if it produces better data than direct regional service would have produced. If it produces more ambiguity, the market will price the ambiguity and members will pay for decentralisation through discounts, delays and mistrust.
Member trust depends on knowing who is accountable
A member relationship is a bundle of rights, duties and expectations. The member pays fees, keeps records current, follows policy, receives services, participates in governance and relies on continuity. An NIR changes the bundle because the member may interact daily with a national body while depending ultimately on AFRINIC recognition. Trust depends on knowing which institution is accountable for each part.
The principal question is simple. If something goes wrong, whom does the member call, and who has power to fix it? If a fee is misapplied, is the NIR responsible or AFRINIC? If a transfer is delayed, is the delay local validation, regional review, legal hold or missing evidence? If a staff member gives incorrect advice, is the member protected? If an NIR loses documents, does AFRINIC accept resubmission? If local records are stale, can the member update directly with AFRINIC? If a local committee has a conflict, who hears the complaint? If a national regulator pressures the NIR, is the member notified?
Without answers, NIR intermediation creates accountability gaps. The national body may say AFRINIC policy required the result. AFRINIC may say the national body handled local evidence. The member may be stuck between two institutions, each able to point to the other. This is common in federated systems. It is especially harmful when the underlying resource is scarce and operationally important.
Downstream voice is another problem. If an operator works through an NIR, does it have the same practical voice in AFRINIC governance as a direct member? Can it vote directly? Can it comment on fee changes that affect NIR pass-through? Can it appeal regional policy effects? Does the NIR represent its members in AFRINIC meetings, and if so, how is that representation authorised? If the NIR's position conflicts with a downstream member's position, can the member be heard independently?
These questions are not abstract democracy concerns. They affect pricing and compliance. A downstream operator that cannot influence regional policy may treat APNIC-style or AFRINIC-style charges as taxes rather than member fees. A member that cannot appeal local interpretation may route around the official system. A buyer that cannot tell whether an NIR-mediated holder has regional standing may demand warranties. A small network that depends on the NIR for voice may not object to rules that hurt it because doing so risks the local relationship.
A credible model would publish a responsibility matrix. It would identify who handles membership intake, identity validation, fee invoicing, fee remittance, contact updates, resource requests, transfer intake, reverse DNS, RPKI, abuse-contact updates, dispute flags, voting notices, appeals, termination and reinstatement. For each function it would name the responsible institution, the decision standard, the evidence burden, the timeline, the escalation path and the public record created.
Service metrics and appeals should follow the same logic. AFRINIC and any NIR should report aggregate timing, return reasons, local-delay categories, remittance problems, transfer escalations and appeal outcomes. If the NIR is the only forum, local politics can dominate. If AFRINIC is the only forum, local evidence may be misunderstood. The appeal path should combine local fact review, regional ledger review and independent review for high-consequence decisions, while preserving the last verified operational state where possible.
The member should never be left asking whether it is a customer of the NIR, a member of AFRINIC, a subject of a regulator or a dependent of an informal committee. It may be all of these in different ways. The system must say which role matters for each decision.
Why AFRINIC is the harder test
If AFRINIC were a quiet, well-capitalised registry with stable board legitimacy, uncontested elections, predictable transfer rules and high member confidence, NIR relationships would still be difficult. They would be ordinary institutional engineering. AFRINIC is not in that position. Its recent history makes every design choice more heavily loaded.
The public record shows a registry that has been under extreme stress. Reporting and institutional statements have described the Cloud Innovation dispute, litigation in Mauritius, bank-account restraints, years without ordinary board continuity, the appointment of a receiver, attempted elections, concerns around powers of attorney and voter documentation, annulment of a June 2025 election, a later board election, continued litigation and ICANN involvement in proceedings concerning AFRINIC's future. Those facts do not prove misconduct by any particular current actor. They prove that AFRINIC's governance environment is not a high-trust background condition.
In that environment, a national intermediary can be interpreted in two opposite ways. Supporters may describe it as a way to reconnect AFRINIC with real operators after years of central crisis. Critics may describe it as a way to distribute gatekeeping to local allies, regulators or incumbents. Both readings will find plausible examples. That is why design must do more work than rhetoric.
Scarcity magnifies the stakes. AFRINIC's IPv4 free-pool position and soft-landing rules mean that many future needs will not be satisfied by large new allocations. Transfers, leasing, reallocation, better utilisation, IPv6 transition and business restructuring will matter. Any national layer that affects those paths affects economic value. It is not enough to say an NIR would help members. The question is which members, in which transactions, at what fee, with what appeal, and under whose influence.
The crisis also makes national political involvement more likely. Governments and regional bodies may see a troubled registry as too important to leave to a small technical community. They may want national channels to protect continuity, prevent perceived asset flight, support public networks or ensure domestic representation. Some involvement may be constructive. But a registry recovering from legitimacy disputes should be cautious about creating new authority points before old ones are trusted.
AFRINIC's service region is also more fragmented than a map suggests. Countries differ in language, corporate law, banking capacity, telecom-market structure, regulator strength, public-sector procurement, currency stability and operator scale. A single national model will not fit all. A large market with a capable neutral technical institution may handle NIR functions better than a small market where one incumbent dominates industry bodies. A country with severe currency controls may need payment intermediation more than transfer discretion. A Francophone market may need language support more than fee collection. A public-sector-heavy market may need authority templates more than local policy committees.
This argues for modular design rather than a grand NIR scheme. AFRINIC could recognise local validation partners without creating full NIRs. It could offer local-language support contracts. It could allow domestic fee-payment agents under strict remittance rules. It could sign regulator information protocols for factual notices only. It could publish country-specific document guides. It could pilot local assistance for company authority without delegating transfer approval. Each module should be judged by whether it lowers registry transaction cost without adding hidden discretion.
The hardest part is sequencing. A low-trust institution should first rebuild core service metrics, transfer transparency, appeal paths, fee clarity and board accountability. National layers should not be used to compensate for the absence of those foundations. If the central ledger is not trusted, decentralising touchpoints may spread distrust rather than cure it.
AFRINIC is therefore the harder test because national intermediation would arrive not as a neutral efficiency reform, but as a choice inside a contested scarcity regime. That does not make it impossible. It makes narrowness mandatory.
A credible AFRINIC model would start narrow
If AFRINIC ever moves toward national registry relationships, the safest model would begin with narrow service functions rather than resource authority. The order matters. Start with tasks that improve evidence and member support. Delay tasks that can move economic value until trust, metrics and appeals are mature.
The first module is document mapping. AFRINIC can publish country-specific guidance explaining which corporate documents, regulator records, public-institution authorities, merger papers and translations are typically acceptable. Local experts can help prepare these guides without acquiring decision power. This would lower support cost and make evidence standards more predictable.
The second module is language and training support. Local partners can explain registry concepts, billing cycles, contact updates, reverse DNS, RPKI, transfer basics and policy notices in domestic languages. The goal is operational literacy, not political representation. The partner should not be able to decide a member's rights merely because it provides education.
The third module is payment facilitation. AFRINIC may allow domestic payment agents or local-currency collection where banking friction is high, but the rules must be strict: member receipts, remittance deadlines, segregation of funds, currency conversion formula, standing protection after local payment, audit, and remedies if the agent fails. Payment facilitation without these rules becomes a fee-risk multiplier.
The fourth module is local validation. A partner may confirm document authenticity, signatory authority, licence facts or translation accuracy. AFRINIC should retain the policy decision. Validation reports should be visible to the member and challengeable. A local fact should be labelled as fact, not mixed with recommendation.
The fifth module is dispute notation. Local partners may alert AFRINIC to domestic disputes, court orders, suspected fraud or conflicting authority claims. AFRINIC should classify the notice: binding legal order, credible fraud report, corporate dispute, regulator inquiry, or non-binding concern. Each category should have different consequences. Not every local dispute should freeze all services.
Only after these modules work should AFRINIC consider deeper NIR functions such as assignment administration, full local membership management or transfer intake. Even then, authority should be bounded. The NIR can gather evidence, maintain local records and support the member. AFRINIC should preserve final recognition, transfer finality, regional policy interpretation and appeal.
A credible model would also include a prohibition on hidden local policy. If an NIR applies additional domestic requirements, those requirements must be published, tied to lawful authority, and mapped to regional consequences. If a regulator asks for a hold, the legal basis must be recorded. If a national fee is charged, the fee's function must be disclosed. If local review delays a transfer, the member must know the reason and escalation path.
Finally, the model should be reversible in modules. If a payment agent fails audit, payment facilitation can be suspended without ending language support. If a local validator develops conflicts, validation can be moved back to AFRINIC without disrupting records. If transfer intake creates delay, transfer intake can be centralised while document guides remain local. Full NIRs are hard to unwind; modular relationships are easier to correct.
The goal is not to build a federation for its own sake. It is to lower the cost of using a regional ledger. Narrow modules are less glamorous than national registries. They are also less likely to become national gates.
Watchpoints for national registry economics
The first watchpoint is language. If AFRINIC or African internet-governance actors describe national intermediaries as a legitimacy solution, the model is already drifting. Local presence may improve service, but legitimacy comes from auditable authority, not geography. The vocabulary should stay with service, evidence, records, fees, timelines, appeals and continuity. If it shifts toward national ownership, retention, strategic resource control, proper domestic use, industrial priority or protection from foreign buyers, the NIR relationship is becoming a political economy of address movement.
The second watchpoint is money and settlement authority. Any national layer that collects money should publish what is regional pass-through, what is local administration, what is tax, what is reserve, what is programme cost and what is subsidy. Members should receive receipts that protect regional standing when they have paid locally. Transfer roles should be equally visible: evidence gathering, recommendation, approval, rejection or transmission. Transfer authority is where service becomes market power, and market power will show itself in pricing, warranties, preferred counterparties and the use of brokers who sell navigation rather than capacity.
The third watchpoint is local political economy. Cooperation with telecom regulators is useful when it supplies facts or lawful orders; it is dangerous when it supplies policy preferences disguised as validation. National registry boards, committees, funding and staff should be examined for concentration among dominant telecom operators, state-linked carriers or industry groups with commercial stakes. The regional database should remain authoritative, members should know how to participate and appeal, and metrics should show whether small networks face lower or higher total cost after intermediation.
These watchpoints are not a checklist for blocking reform. They are a way of testing whether the national layer is a bridge to the ledger or a gate in front of it.
The conservative conclusion
AFRINIC's NIR question is not whether Africa needs more local capacity. It does. The region's operators would benefit from clearer documents, better language support, easier payment channels, faster validation, stronger regulator education and support paths that do not require every small network to become expert in regional registry procedure. A purely central model can be formally neutral while practically expensive.
The question is whether local capacity can be built without fragmenting the ledger. A National Internet Registry is attractive because it looks close to the member. It is risky because it is also close to national politics, local incumbents, currency frictions, regulator expectations and domestic ideas about scarce digital capacity. Proximity lowers one cost and raises another.
The economic line should be drawn at recognised control. Local intermediaries may help produce evidence. They may translate. They may collect money under audited rules. They may explain regional policy. They may validate domestic facts. They may improve data quality. They should not own the regional record, decide ordinary commercial merit, impose hidden transfer conditions, convert fees into opaque local taxes, or turn national development preferences into registry vetoes.
AFRINIC's crisis makes this discipline more important, not less. A registry emerging from litigation and receivership needs fewer ambiguous authority points, not more. If national layers are introduced before core regional processes are trusted, they will be read as factional instruments. If they are introduced narrowly, measured carefully and made appealable, they could reduce the very frictions that made AFRINIC remote to many members.
The best model is modest. Keep the regional ledger authoritative. Use national knowledge to improve facts. Publish fee maps. Separate payment from existential resource harm. Treat transfer approval as regional settlement, not national permission. Label regulator input by legal status. Audit local intermediaries. Protect small operators from incumbent capture. Preserve direct appeal. Report friction. Make every added layer prove that it lowers the cost of accurate recognition.
That is a conservative position in the institutional sense. It does not romanticise central control or local sovereignty. It asks a registry to do the boring thing well: maintain a neutral record that members can use without needing political favour, procedural insider status or national permission beyond what law actually requires.
AFRINIC is a test case because the temptation to localise is real and the danger of losing neutrality is also real. National internet registry relationships can make a regional registry more accessible. They can also make it less regional. The difference lies in whether the national layer is built as a bridge to the ledger or as a gate in front of it.