AFRINIC and the economics of transfer market architecture

The market behind the registry

AFRINIC is usually introduced as Africa's Regional Internet Registry, a nonprofit and member-based organisation registered in Mauritius and entrusted with distributing and managing internet number resources for Africa and parts of the Indian Ocean. The description is correct, but it understates the economic problem now facing the institution. AFRINIC does not merely run a contact database. It records who may use globally unique IPv4 addresses, IPv6 prefixes and autonomous system numbers; it maintains the processes through which scarce resources are allocated, transferred, reserved or returned; and it operates in a region where IPv4 scarcity, litigation and contested governance have made registry decisions commercially consequential.

The central issue is the architecture of the IPv4 transfer market. A transfer market is not just a brokered sale between two firms. It is a rule system: who may sell or transfer a block, who may receive it, what evidence of need is required, whether the registry must approve the recipient's use case, whether resources can cross regional borders, how legacy status is treated, and whether a disputed holder is allowed to transact. Each design choice affects price, liquidity, incentives and confidence in the ledger. In AFRINIC, those choices have become unusually visible because the registry has been rationing a scarce pool while members and market participants have treated IPv4 as an operational and financial input.

The old mental model of an RIR was administrative. A network justified need; the registry allocated numbers; the numbers were entered into a registry database; the network routed them or delegated them downstream. In that world the registry's authority looked narrow, technical and largely clerical. IPv4 exhaustion changed the model. The relevant block is no longer just an identifier assigned for immediate use. It can be leased, acquired through a transfer, pledged in a business plan, counted as a strategic operating asset and defended in court. The registry record does not by itself make packets flow, but it is the institutional record that allows other networks, counterparties and customers to treat the address holder's claim as legitimate.

That is why the distinction between neutral registration and policy rationing matters. Neutral registration records a valid state of affairs and protects the uniqueness of identifiers. Policy rationing goes further: it decides whether a claimant deserves access to a scarce input, whether a later business use remains permissible, and whether a transfer should be blocked for reasons beyond record accuracy. AFRINIC's own Consolidated Policy Manual contains both logics. It says the internet registry system must preserve uniqueness and registration, but it also says conservation requires distribution according to actual need and immediate use. Scarcity turns those two imperatives into an institutional conflict.

The argument here is not that AFRINIC should abandon policy enforcement. A registry that ignores fraud, stale records, inaccurate contact data or resource hoarding undermines the very ledger it exists to maintain. Nor is the argument that IPv4 addresses are ordinary private property. The standard RIR doctrine that internet number resources are public resources, not conventional property, remains important. The problem is more precise. A transfer market needs rules that preserve registry integrity without letting a registry become an industrial-policy gatekeeper over the business models of address holders. If that boundary is vague, every transfer request can become a policy trial and every policy dispute can become a fight over economic control.

AFRINIC is now the sharpest case study because the region's registry has experienced nearly every stress a transfer architecture can face. Public reporting has described allegations about historical address-record manipulation. A major member dispute produced litigation and a bank-account freeze. The Supreme Court of Mauritius appointed a receiver, and the Number Resource Organization said the receiver was expected to preserve continuity while elections and executive appointment were restored. The Register has tracked attempted elections, annulment, later board formation, renewed litigation, ICANN intervention and the slow revision of the recognition framework for RIRs. Those episodes are not merely governance colour. They are market-architecture facts, because a transfer market depends on the credibility of the institution that approves, records and explains transfers.

The economics are not mysterious. IPv4 is finite; IPv6 has not removed near-term IPv4 dependence; operators still need IPv4 reachability for customers, hosting, anti-abuse systems, access networks and compatibility with the still-large IPv4 internet. When a registry retains administrative control over a scarce input, its procedures become part of the price of using that input. Delay is a cost. Ambiguity is a cost. Litigation risk is a cost. A credible ledger lowers transaction costs because buyers, sellers, lessors and networks can rely on a public record. An unpredictable gatekeeper raises them because every transaction must price the chance that the registry, a court, a receiver, a board dispute or a policy fight will interrupt the chain.

The question for AFRINIC is therefore institutional rather than ideological: can the registry design a transfer architecture that treats scarcity honestly, enforces the ledger rigorously, and restrains discretionary control over lawful commercial reliance? The answer will shape more than a few large holders. It will affect African access providers that need incremental IPv4 capacity, data-centre operators balancing IPv4 and IPv6 transition, enterprises that depend on address reputation, and the credibility of the RIR model when administrative resources become valuable enough to litigate.

Scarcity made transfer rules unavoidable

IPv4 scarcity is the condition that makes a transfer market necessary. AFRINIC's public exhaustion page states that it has managed internet number resources since 2005 and that the African internet community supported a Soft Landing policy in 2011 to address scarcity during the transition to IPv6. It records that AFRINIC announced the start of Phase 1 of its IPv4 exhaustion process on 31 March 2017 and entered Soft Landing Phase 2 on 13 January 2020. Under Phase 2, the minimum allocation or assignment size is /24 and the maximum is /22 per allocation or assignment. That is not the language of abundance. It is the language of rationing.

Scarcity has two different meanings in the AFRINIC context. The first is technical and administrative: the free pool is finite, and a registry must stretch it for late requesters while encouraging IPv6 deployment. The second is economic: when new supply is constrained, existing usable IPv4 space acquires a market price. The Internet Governance Project's 2021 analysis reported that IPv4 transfer-market prices had risen from about $8 per address in 2017 to about $30 by 2021, using a /16 block as an example of a resource worth millions of dollars. The exact price at any moment varies by reputation, size, routing history, contract terms and transaction risk. The institutional point is durable: allocation at administrative fees and transfer at market prices create different incentive systems.

AFRINIC's fee schedule shows the gap between registry service pricing and market value. The schedule sets annual membership fees by category and resource size, and it charges allocation or assignment fees when resources are approved. It also says that, for transfers between two existing resource members, AFRINIC does not currently charge IP and ASN transfer fees, though recipient categories can be recalculated for future renewals and accounts must be in good standing. These are service fees, not market-clearing prices. That is appropriate for a nonprofit registry, but it means the registry cannot pretend its allocation and transfer decisions have no market consequences. The price signal is outside the registry; the approval gate remains inside it.

Under abundance, a restrictive needs rule is less costly because rejected applicants can usually return with better documentation or wait for another allocation cycle. Under scarcity, rejection pushes the applicant toward the transfer market, leasing, workarounds or delayed deployment. Under abundance, an address holder that changes business use may see the issue as a compliance matter. Under scarcity, the same review can threaten a valuable stream of service revenue. The rule does not change its text; scarcity changes its stakes.

The Soft Landing policy embodies this tension. AFRINIC's policy manual says its purpose is to manage the remaining IPv4 pool after exhaustion and to help maintain IPv4 networks while IPv6 networks are deployed. It requires high utilisation of previous allocations for additional requests and says AFRINIC resources are for the AFRINIC service region, with any use outside the region to be solely in support of connectivity back to that region. To conservation-minded participants, those clauses protect a small regional pool from being drained into global demand. To market-minded participants, they risk turning the registry into a continuing judge of business geography, customer location and commercial model. Both readings follow from the same scarcity architecture.

AFRINIC's regional position heightens the issue. IGP's 2021 analysis argued that AFRINIC held only a small share of global IPv4 space and came late to the RIR system, while a large share of IPv4 was distributed before the modern regional registry model existed. Whatever one thinks of that comparison, the asymmetry is real enough to matter. African networks need IPv4 during the transition just as networks elsewhere do, but the global resource endowment was not distributed according to future African demand. A policy that tries to reserve local supply for regional development is understandable. A policy that traps resources in low-liquidity regional rationing can also become economically inefficient if addresses are stuck with holders that value them less than networks that need them more.

The policy answer cannot be a fantasy that IPv4 commercialisation will disappear. AFRINIC's own February 2026 public context, reported by The Register, included a senior AFRINIC executive saying the registry still had 773,376 unallocated IPv4 addresses and expressing eagerness to move the conversation to IPv6 when that pool reaches zero. IPv6 deployment is essential, but it does not erase present IPv4 reliance. A registry can promote IPv6 while still needing a coherent IPv4 transfer architecture for the remaining transition period. If it treats the transfer market as a moral failure rather than as a scarcity response, it will drive more activity into opaque contractual arrangements and litigation.

The better framing is that transfer rules are the market's constitution. They need not maximise trading volume at all costs. A well-designed transfer system should prevent fraudulent claims, ensure current records, verify that recipients can use and manage the resources, preserve routing and registration integrity, and protect the community against gaming. But it should also permit resources to move toward higher-value operational use when the registry itself can no longer satisfy demand. Scarcity made that movement unavoidable. The remaining question is whether AFRINIC can supervise it as a neutral registry rather than as a discretionary rationing authority.

What AFRINIC's transfer policy actually builds

AFRINIC's Consolidated Policy Manual contains the clearest public view of its formal transfer architecture. Section 5.7, added in the 2017 version history, is titled "IPv4 Resources transfer within the AFRINIC Region." It explains that a transfer policy is needed because AFRINIC will exhaust its IPv4 pool and late requesters will still need resources. The policy applies to organisations with a justified need for IPv4 resources that AFRINIC cannot satisfy. That formulation is important: the policy treats transfers as a scarcity relief valve, not as an ordinary property market.

The source side of a transfer is constrained. The IPv4 resources must come from an existing AFRINIC member account or from a legacy resource holder in the AFRINIC service region. The source must be the current rights holder recognised by AFRINIC and must not be involved in a dispute as to the status of those resources. The source is barred from receiving further IPv4 allocations or assignments from AFRINIC for twelve months after transfer approval, and it must not have received a transfer, allocation or assignment from AFRINIC in the preceding twelve months, except in merger-and-acquisition contexts. This design tries to prevent round-tripping, speculative churn and the use of administrative allocation as immediate market inventory.

The recipient side is even more revealing. AFRINIC must approve the recipient's need for the IPv4 resources. The recipient must justify and demonstrate its initial or additional allocation or assignment usage according to policies in force. It must be an AFRINIC member and must sign the Registration Service Agreement for the resources being received. If the transferred resources are legacy resources, they cease to be treated as legacy once transferred. The effect is to bring transferred resources inside the same member, policy and contract architecture that governs ordinary AFRINIC resources.

This is a need-based transfer architecture. It is neither a fully open market nor a simple recording system. A purely recording system would ask whether the source has the right to transfer, whether the recipient exists and accepts responsibility, whether the block is cleanly identified, whether required notices have been made, and whether the database can be updated without conflict. AFRINIC's policy asks more. It makes the recipient justify need before the registry, importing allocation logic into the transfer market. The registry therefore remains an allocator even when no new addresses are being allocated from the free pool.

There are reasons for that design. Without recipient need review, a scarce regional pool could be converted into a speculative asset pipeline. A member could obtain resources under a regional policy bargain and later sell them to a party with no operational connection to regional network development. The twelve-month restrictions address that danger. The good-standing and dispute-status rules also protect ledger integrity: a disputed block should not move into another account while the registry lacks confidence in the source's claim. These are not arbitrary controls; they answer genuine risks.

The economic cost is lower liquidity and higher uncertainty. A buyer or recipient does not only negotiate with the source. It must satisfy AFRINIC's view of need, policy compliance and membership standing. A seller must worry that the registry may regard the resource status as disputed or may delay approval. A financier or customer evaluating the block must price the risk that the transfer cannot close even if commercial terms are agreed. If the registry applies the policy predictably, these costs may be acceptable. If the registry's interpretation shifts with board politics, litigation pressure or informal regional-use ideology, the costs become a market discount on AFRINIC-origin resources.

The treatment of legacy resources is particularly instructive. By converting transferred legacy resources into non-legacy resources subject to current AFRINIC policy and contract, the registry prevents old claims from escaping the contemporary rulebook once they enter the market. That may improve accountability and database coherence. It may also reduce the price a legacy holder can realise if buyers prefer fewer registry-side restrictions. Again, this is not merely a legal detail. It is market architecture: policy determines how value is converted when a resource changes hands.

The same pattern appears in the fee schedule. AFRINIC says transfers between existing members are not currently charged as IP and ASN transfers, but both parties must have accounts in good standing, and a new recipient organisation must apply for membership and pay the relevant allocation and membership fees. Good standing is sensible when membership fees support registry operations. It also gives the registry leverage over transactions. Payment, documentation, membership classification and policy review become prerequisites for moving a scarce input. In a stable institution, that leverage is administrative. In a contested one, it can become strategic.

AFRINIC's transfer policy therefore builds a hybrid structure. It acknowledges that transfers must occur after exhaustion, but it keeps transfers inside a needs-based, member-controlled, regionally bounded framework. The policy is coherent as conservation architecture. It is less obviously sufficient as market architecture, because it does not fully answer how the registry should treat commercial leasing, changed use after allocation, out-of-region customer demand, or address blocks whose operational use evolves over time. Those unresolved questions explain why the dispute over AFRINIC transfers is not a side battle. It is the point at which the registry's historical allocation model meets the economics of a secondary market.

Ledger integrity as the first market condition

Before a transfer market can be efficient, it must be trusted. Trust begins with the ledger. For IPv4, the ledger is not only the WHOIS or RDAP entry that names a holder. It includes the historical chain through which resources were allocated, assigned, transferred or returned; the status of associated reverse DNS, route and certification services; the contact records that allow abuse and operational issues to be handled; and the registry procedures by which a contested entry can be corrected. A transfer market without ledger integrity is a market in disputed claims.

AFRINIC's policy manual makes registration central to the internet registry system. It lists uniqueness, registration, aggregation and conservation as goals for IPv4 management. It states that every assignment and allocation of public internet address space must be registered in the AFRINIC WHOIS database, and that this is necessary both to ensure uniqueness and to provide troubleshooting information. It also says registration data must be correct at all times and that unregistered resources are invalid. These are not bureaucratic niceties. They are the conditions under which other networks can trust that an address block has a recognised holder.

The 2019 reporting by KrebsOnSecurity shows why this matters. Krebs reported allegations, based on Ron Guilmette's investigation and related South African reporting, that African IPv4 address blocks had been commandeered from defunct or acquired organisations and sold through companies linked to a former AFRINIC policy coordinator. Krebs reported an estimated market value above $50 million for the addresses documented by Guilmette, said the executive resigned in October 2019 after the allegations became public, and quoted AFRINIC's then chief executive as saying the organisation was investigating. Those were reported allegations and responses, not a final public adjudication in that article. The market lesson is still clear: if registry records can be manipulated or left stale, scarcity turns bad records into high-value claims.

In a low-value environment, a stale contact or dormant holder creates operational nuisance. In a high-value environment, it creates an arbitrage opportunity. Someone who can alter records, revive a dormant entity, obtain a misleading authority document or exploit ambiguity in corporate succession may turn registry weakness into economic gain. Conversely, a legitimate holder may be wrongly suspected if the registry lacks transparent provenance. Market architecture therefore needs auditability. It must be possible to distinguish ordinary transfers, mergers, legacy updates, fraud remediation and disputed claims without relying on informal trust in staff or board discretion.

AFRINIC's formal transfer rule partly recognises this. It says the source must be the current rights holder recognised by AFRINIC and must not be involved in any dispute over the status of the resources. That protects buyers and recipients from inheriting unresolved title problems. But it also puts substantial interpretive power in the registry. What counts as a dispute? A court action? A member complaint? A staff suspicion? A public allegation? A conflict over corporate control? If this threshold is unclear, a party can block a transfer by manufacturing a dispute, while the registry can suspend liquidity by declaring uncertainty.

The right architecture would treat dispute status as a defined category with procedural consequences. Some disputes should freeze transfer approval because there is a real risk of transferring the wrong party's resource. Others should permit transfer with notation, escrow-like conditions, indemnities or delayed effect. Still others should be dismissed as unsupported. A registry that simply says "disputed" without showing the type, source, process and expected path to resolution creates a market-wide uncertainty tax. Buyers will discount blocks from that registry; sellers will prefer private leasing arrangements; members will litigate because no one knows which record is final.

Ledger integrity also requires separation between record accuracy and policy punishment. If a block is misregistered, the registry should correct the record. If a holder has breached payment obligations, the registry should enforce billing and membership rules. If a holder committed fraud at application, the registry may need recovery powers. If a holder's customer base changed after lawful allocation, the appropriate response may be different again. Collapsing these categories creates a dangerous market signal: the registry can use record control to enforce broad policy preferences. That is precisely the point at which neutral registration becomes gatekeeping.

The public record around AFRINIC shows how costly this boundary can be. IGP's 2021 analysis reported that AFRINIC questioned Cloud Innovation over discrepancies between registered usage and actual use, alleged inconsistency with the needs originally expressed, and regional-use issues. Cloud Innovation contested AFRINIC's interpretation. Whatever the legal merits, the economic issue is that a resource record became the instrument through which a business model was challenged. A registry with weakly defined review powers can find itself treating every changed use as a potential ledger defect and every ledger defect as a reason to threaten resource withdrawal.

A credible transfer market requires the opposite discipline. The ledger should identify who currently holds the resource and what public registry facts are attached to it. Separate procedures should decide whether the holder breached policy, whether a transfer can proceed, whether downstream users need protection and whether any remedy is proportionate. If AFRINIC wants its transfer market to support regional development rather than litigation, it must make ledger integrity boring again: accurate, auditable, predictable and narrow enough that market participants do not confuse record administration with discretionary economic control.

Allocative efficiency and policy rationing

The economic case for transfers is allocative efficiency. When a resource is scarce, and when some holders need it less than others, a transfer market can move that resource toward higher-value use. In IPv4, "higher value" need not mean speculative profit. It may mean a broadband provider serving new customers, a data centre that cannot renumber legacy clients, a bank maintaining compatibility with payment partners, a hosting firm with reputation-sensitive workloads, or an enterprise migration plan that still depends on IPv4 reachability. Transfers are a way of discovering who values the resource enough to bear the cost of obtaining it.

Policy rationing has a different logic. It asks the registry to decide who deserves the resource according to community rules. AFRINIC's needs-based transfer policy preserves that logic after exhaustion. The recipient must justify and demonstrate need before AFRINIC; the recipient must be a member; the resource becomes subject to current policy and the Registration Service Agreement. This model can be defended as a way to ensure that scarce African resources support real network operations rather than pure financial inventory. It also reduces the ability of an early holder to convert a community-allocated resource into unrestricted private gain.

Yet rationing can misallocate when rules lag behind operational reality. Network need is not static. A provider may move customers among countries, lease capacity to another operator, use addresses for cloud workloads, support multinational routing, or change architecture after acquisitions and customer churn. A policy that requires registry approval for each material business evolution will slow adaptation. It will also privilege firms whose uses fit the registry's categories over firms whose operations are equally real but less conventional. Efficient markets punish bad use by price and reputation; rationing systems punish category mismatch through delay or denial.

AFRINIC's own policy manual recognises that registry goals can conflict. It says conservation and aggregation often conflict, and that registry decisions may conflict with the interests of individual registries or end users. It requires careful analysis and a fair balance between applicant needs and the needs of the internet community as a whole. That is sound institutional language, but it becomes harder under transfer-market conditions. If the registry both defines community interest and controls transaction approval, resource holders may doubt whether the balance is neutral. The more economically valuable the block, the less persuasive general appeals to community interest become unless the decision criteria are specific.

The regional-use question illustrates the problem. AFRINIC's Soft Landing text says resources are for the AFRINIC service region and out-of-region use should support connectivity back to the region. A strict reading protects regional scarcity. A more permissive reading recognises that internet services, cloud platforms, content delivery, enterprise networks and leasing markets do not map neatly onto continental borders. If a Seychelles-incorporated or African member leases addresses to customers abroad, is the registry dealing with efficient monetisation of a scarce input, export of regional capacity, breach of community bargain, or changed business use that requires new approval? The answer depends less on packet routing than on policy architecture.

If AFRINIC prohibits too much movement, it may keep addresses in lower-value uses, encourage private leasing that avoids transfer scrutiny, and make African-origin addresses less liquid. If it permits any movement without review, it may invite depletion, political backlash and the perception that addresses intended for regional development became global inventory. The efficient architecture is therefore not "free market" versus "no market." It is a disciplined set of transfer rules that separates evidence of operational need from paternalistic control over business strategy.

One practical design principle is to use negative screens rather than broad positive approval. A registry should block transfers involving fraud, unresolved source status, sanctions or legal incapacity, unpaid registry obligations that directly affect membership status, material abuse-contact failures, or recipient inability to maintain accurate records. It should be more cautious about blocking transfers because the recipient's commercial model is unfamiliar or because the registry dislikes leasing. Where recipient need review remains in policy, the standard should be concrete, time-bounded and reviewable, not an invitation for the registry to re-run business planning.

Another principle is to reduce artificial supply discontinuities. AFRINIC's rule that a source cannot receive new IPv4 resources for twelve months after a transfer and cannot have received resources in the preceding twelve months is aimed at preventing arbitrage. That is reasonable. But the registry should publish enough data for members to know whether these restrictions are actually preventing churn or simply reducing liquidity. If the region has only a small number of real transfer candidates and approvals take too long, policy may be conserving addresses on paper while networks solve their needs through less transparent leases.

Allocative efficiency also depends on confidence that a completed transfer will remain stable. If AFRINIC can revisit the recipient's need shortly after approval, or if later policy changes can radically alter transfer rights without transition protections, recipients will price that risk. Transfers will become expensive, slow and legally over-engineered. A registry that wants responsible transfers should offer credible finality: once source status, recipient qualification and record updates are complete, ordinary business evolution should not reopen the transaction absent fraud, material misrepresentation, payment default or clearly defined policy breach.

AFRINIC's challenge is to preserve the social purpose of registry policy while allowing the market to reveal where IPv4 capacity is most needed. That requires humility about what a registry can know. It can verify records, eligibility and compliance better than a court or broker can. It is less well suited to deciding the optimal geographic distribution of every block in a global internet economy. Transfer-market architecture should use the registry where it has informational advantage and constrain it where discretion becomes economic planning.

Rights ambiguity: not property, not nothing

No issue in the IPv4 transfer debate generates more confusion than the status of address rights. The official registry position, reflected in RIR doctrine and repeated in AFRINIC-related reporting, is that IP addresses are not owned as ordinary property. ICANN's 2026 intervention in the Mauritian winding-up matter, as reported by The Register, rested on the same idea: numbering resources allocated through AFRINIC are not assets of AFRINIC available for distribution in a winding-up. That position is essential to registry continuity. If a failing registry's number resources could be treated like corporate assets, the coordination system would be exposed to liquidation logic.

But "not property" is not the same as "no economic right." A resource holder may not own an IPv4 block like land or equipment, yet it may have contractual rights, reliance interests, operational dependencies and customer obligations built around the registry's recognition of its use. A bank licence is not ordinary property either, but losing it has economic consequences. A spectrum authorisation may be subject to public-law limits, but businesses still invest around it. A port concession, route permit or energy interconnection right can be non-proprietary while still deserving stable procedure. IPv4 sits in that family of economically significant administrative entitlements.

This middle position matters for transfer architecture. If addresses are ordinary property, registry policy becomes a nuisance around alienable assets. If addresses are merely revocable permissions, holders cannot make credible long-term plans and transfer prices should collapse under registry risk. The functioning market has evolved between those poles. Holders, lessors, brokers and buyers behave as if IPv4 use rights have value, while registries insist that those rights remain subject to policy, contract and database accuracy. The market exists because both sides are partly right and because neither doctrine has fully displaced the other.

AFRINIC's policy manual itself uses language that reflects ambiguity. It speaks of allocations, assignments, custodianship, public resources, demonstrated need, registration and transfer conditions. Section 5.7 says the source must be the current rights holder recognised by AFRINIC. That phrase does real work. It does not say owner, but it does recognise a right holder whose status matters enough to be checked before transfer. The source's recognition by AFRINIC is the registry-layer basis for market transfer. The registry is not selling property, but it is recognising a transferable status under conditions.

The Cloud Innovation dispute made the ambiguity economically concrete. IGP reported that AFRINIC argued over the use of resources and the continuing significance of the need originally justified in the member's application. Cloud Innovation resisted and framed the issue as overreach into business operations. The public record then moved into injunctions, revocation threats, bank freezes and prolonged litigation. The legal merits belong to the courts and specific contracts. The architectural lesson is that a transfer market cannot function well when the nature of the holder's right is intelligible only after crisis litigation.

Rights ambiguity also affects collateral expectations. A company that pays to receive a transferred block, or that leases addresses for customers, wants to know what can interrupt use. Non-payment? Fraud in the source chain? Abuse complaints? Changed customer geography? A new board's interpretation of Soft Landing? A receiver's need to preserve the status quo? A court order in Mauritius? An ICANN compliance intervention? All of these risks are different. A mature architecture would classify them and make clear which risks attach to the block, which attach to the holder, and which attach to the registry institution.

The strongest market-side critique, articulated in public notes and in statements from Lu Heng, NRS and Larus, is that registry discretion has become economic power without matching liability. This critique is made by interested actors and should be read accordingly. It nevertheless captures a real institutional problem. If a registry can cause continuity-scale loss through suspension or revocation while its own contractual exposure is limited, holders will view the registry as an asymmetric gatekeeper. They will then seek courts, lobbying, proxies, alternative bodies or structural separation to rebalance that risk. The more the registry denies the economic significance of address reliance, the more intense the counter-move becomes.

AFRINIC's defenders have an equally serious concern. If holders can monetise resources freely after receiving them under need-based policies, the registry becomes a subsidised acquisition channel. Scarce public identifiers could be transformed into private inventory, while networks with real regional needs face higher market prices. That risk is not theoretical. KrebsOnSecurity's 2019 reporting on allegedly commandeered African address blocks shows how valuable dormant or poorly controlled resources can become. A rights model with no registry enforcement would invite abuse.

The answer is not to settle the property debate with a slogan. AFRINIC needs a rights taxonomy for transfer purposes. Some rights are registry-recognised use rights. Some are legacy claims. Some are membership-based service entitlements. Some are contractual expectations. Some are market contracts between private parties. Some are public registry records. Each layer needs different remedies. Treating them all as property would over-privatise public identifiers. Treating them all as revocable policy grace would over-politicise operational reliance. A credible transfer market requires enough right-like stability to permit transactions and enough public-resource discipline to protect the ledger.

Gatekeeper risk and member incentives

Gatekeeper risk arises when the entity that records legitimate use also decides, with broad discretion, which economic uses remain acceptable. AFRINIC faces this risk because its transfer architecture imports need review, regional-use logic, good-standing requirements and dispute-status controls into the market. Each control can be justified. Together they make the registry's approval a scarce gateway. In a period of governance uncertainty, the gateway becomes a prize.

Member incentives respond accordingly. If transfer approvals, bylaw changes and resource reviews can affect the value of IPv4 holdings, members have stronger reasons to influence elections, committees and policy meetings. The Register's 2025 reporting on AFRINIC election preparations described receiver concerns about potential interference, warnings about credential solicitations, and later allegations involving powers of attorney. Those allegations are contested and should not be treated as final findings. Their significance is that voting authority itself became part of the resource-control surface. Control of the board could shape policy, litigation posture, transfer review and the institution's approach to commercialisation.

This is not unique to AFRINIC, but AFRINIC's crisis makes the mechanism visible. In a boring registry, board service is governance work: budgets, services, policy implementation, transparency, staff oversight. In a high-value scarcity institution, board service can also influence the economic environment for address holders. That does not mean candidates are corrupt or that every endorsement is a capture attempt. It means institutional design must assume higher incentive intensity. Proxy rules, membership verification, conflict disclosures and committee selection need more robustness when the ledger controls valuable resources.

Gatekeeper risk also changes how members perceive enforcement. A small access provider that sees AFRINIC pursue a major holder may welcome discipline if it believes resources are being misused. The same provider may worry if the enforcement theory implies that any changed customer mix, out-of-region service, leasing arrangement or documentation gap could later endanger its own block. A registry that wants member support must show that enforcement is rule-bound rather than discretionary. Otherwise even members who dislike a particular market actor may fear the precedent.

AFRINIC's March 2026 posture, reported by The Register, shows the competing narratives. AFRINIC accused Cloud Innovation, Larus and associated advocacy campaigns of driving litigation and procedural roadblocks, and said legal costs and instability obstructed community initiatives. Lu Heng responded that the issue was structural: a registry model with high-consequence power over economically critical resources but insufficient legal and financial accountability. NRS materials frame the matter in terms of member money, votes, records and exposure to registry chokepoint power. These are participant claims, not neutral findings. They nevertheless show how the transfer-market dispute has become a contest over institutional authority.

Market architecture should reduce the payoff from capturing that authority. The more discretion sits with the board or staff, the more every faction has reason to fight for control. Clear transfer standards, independent appeals, published processing metrics, conflict rules and transition protections lower the prize. They do not eliminate politics, but they make politics less decisive for individual transactions. A member should not need to win an election, hire litigators or join a faction to know whether a transfer can be approved.

The same applies to registry revenue. AFRINIC's fee schedule ties annual membership categories to the aggregate of billable resources held. That is normal for RIRs and supports operations. But the revenue model can interact with transfers and disputes. If large holders litigate or refuse cooperation, revenue and legal costs become intertwined. If members believe fees fund litigation strategies rather than services, they may resist budgets or campaigns may mobilise around financial transparency. If the registry lacks a stable board, fee validation and spending oversight become legitimacy questions. Transfer architecture cannot solve all of this, but it can reduce transaction conflict that spills into the budget.

A healthier incentive design would make routine transfers administratively predictable and reserve political conflict for real policy choices. Members could then debate broad questions - regional conservation, inter-RIR transfer posture, IPv6 transition, abuse controls, legacy conversion - without every debate feeling like an immediate threat to a named holder. When policy and case-specific enforcement are tangled, every participant sees existential stakes. Gatekeeper risk is therefore not just a complaint from large resource holders. It is a governance cost borne by the whole registry community.

The practical test is whether AFRINIC can make its approval function narrower than its policy imagination. A registry can believe in regional development without reviewing every customer's geography. It can oppose fraud without treating all leasing as fraud. It can require accurate records without using record changes as leverage over commercial strategy. It can preserve community control without letting board politics decide transaction outcomes. The transfer market needs AFRINIC to be strong enough to protect the ledger and restrained enough not to become the market's central planner.

Courts, receivership and continuity

AFRINIC's transfer architecture cannot be separated from court and receiver continuity. A market participant evaluating an AFRINIC-origin block must ask not only what the policy manual says but also whether the institution approving the transfer has stable authority. Since 2021, the public record has included court injunctions, a bank-account freeze reported by IGP, receivership described by the NRO, disputed elections, winding-up attempts and ICANN interventions. That legal environment affects transaction confidence even when the underlying registry services continue.

The NRO's September 2023 statement is a key factual anchor. It said the Bankruptcy Division of the Supreme Court of Mauritius had appointed a receiver under the Companies Act, restrained AFRINIC from relocation, takeover, merger, restructuring or management control, and tasked the receiver with preserving the status quo, overseeing elections, facilitating a proper board and appointing a chief executive. The NRO welcomed the appointment because it would help members continue receiving registry services and because functional governance could be restored. That is a continuity statement, not a market-policy statement. Still, continuity is a market-policy condition because transfer approvals depend on institutional capacity.

Receivership can protect the ledger from collapse, but it does not answer the transfer-market questions. A receiver can maintain services, preserve assets and stage elections. The receiver cannot easily settle whether IPv4 leasing should be encouraged, discouraged or tolerated; whether regional-use rules should be interpreted broadly or narrowly; whether recipient need review should survive after full exhaustion; or whether legacy resource conversion should be softened to improve liquidity. Those are governance and policy questions. A receiver keeps the registry alive so they can be decided, but the longer receivership or court supervision becomes part of ordinary registry life, the more every transaction carries legal overhang.

The 2025 election cycle shows how continuity and legitimacy diverge. The Register reported that AFRINIC prepared elections under receiver oversight, with senior legal figures appointed to a nomination committee because of concerns about interference. It then reported ICANN's concerns about election arrangements and confusion over Cloud Innovation's corporate listing. The Supreme Court ordered a communique clarifying that Cloud Innovation's listing as a shareholder was erroneous, but it did not reconstitute the nomination committee. Later reporting described suspension and annulment of the vote after allegations concerning powers of attorney and voter documentation. A later election produced eight directors, but The Register noted continuing critics, possible court challenges and investigations.

For a transfer market, that sequence creates a legitimacy discount. A board may be seated, but if members doubt the election process, major policy decisions will be attacked as factional. If court challenges are pending, market participants may hesitate to rely on new interpretations. If the receiver's mandate is contested, participants may wonder whether approvals made during the transition are secure. The problem is not that every dispute is valid. The problem is that the market must price uncertainty until institutional acts become ordinary enough to be accepted.

ICANN's 2026 intervention in the winding-up matter adds another layer. The Register reported that ICANN sought standing to help the court understand AFRINIC's unique role and to make clear that numbering resources administered through AFRINIC are not AFRINIC assets available for distribution. That is a necessary ledger-continuity argument. It protects the registry function against corporate liquidation logic. Yet it also illustrates how local corporate law, global coordination and member economics now intersect. A court deciding a Mauritian company dispute may affect the confidence of networks across Africa and beyond.

Transfer architecture should anticipate court exposure rather than treat it as abnormal. The registry should be able to provide courts with a precise explanation of what a transfer does and does not do. It does not sell AFRINIC's assets. It records a change in recognised use rights under policy. It should not prejudice unrelated members. It should have clear source and recipient checks. It should preserve historical records. It should identify whether any dispute is about title, compliance, payment, fraud, membership or policy interpretation. Courts are more likely to preserve continuity when the registry's own categories are precise.

The same precision would help market participants. A completed transfer should come with enough registry clarity that the recipient knows whether it can route, request reverse DNS, maintain abuse contacts, create route objects where applicable, and rely on the record absent defined breach. A pending transfer should have status categories visible to the parties. A rejected transfer should explain whether rejection is curable. A court-constrained resource should be flagged in a way that protects confidentiality while warning that ordinary processing is limited. Without such architecture, every legal event becomes rumour, and rumour becomes market risk.

Courts and receivers are therefore backstops, not substitutes for registry design. They can prevent institutional collapse, protect the non-asset character of number resources, and restore governance pathways. They cannot make an ambiguous transfer market efficient. That responsibility sits with AFRINIC's community, board, staff and members. The sooner they turn court-managed continuity into ordinary, rule-bound administration, the lower the market discount on AFRINIC resources will be.

Toward a neutral transfer architecture

A neutral transfer architecture for AFRINIC would begin by defining the registry's role narrowly and strongly. The registry should be strong in protecting uniqueness, provenance, data accuracy, abuse-contact availability, member accountability and clear policy compliance. It should be narrow in judging the commercial wisdom of the recipient's business model. The goal is not to remove discretion entirely. It is to locate discretion where the registry has expertise and limit it where the registry is tempted to ration markets by institutional preference.

The first building block is source certainty. AFRINIC should maintain a public or member-visible transfer status framework that distinguishes recognised holder, legacy holder, corporate-successor review, fraud investigation, payment hold, court hold, competing-claim dispute and ordinary transferable status. The present policy's requirement that the source be recognised by AFRINIC and not involved in a dispute is a starting point, but the market needs more granularity. A party should know whether a hold is legal, documentary, financial or policy-related, and what process can resolve it.

The second block is recipient accountability without business-plan micromanagement. Recipient need review can remain where current policy requires it, but it should be translated into objective criteria: documented network plan, capacity requirement, utilisation of existing holdings, ability to maintain registry records, and compliance with service-region requirements as clearly defined. The registry should avoid open-ended review of downstream customer identity unless a specific rule or abuse concern justifies it. A transfer market cannot scale if every recipient must submit to broad inquiry into future commercial arrangements.

The third block is finality with defined reopeners. Once AFRINIC approves a transfer, records it, and the recipient signs the required agreement, the transaction should be stable. Reopeners should be limited to fraud, material misrepresentation, court order, serious payment default, clear and defined policy breach, or other published grounds. Later disagreement with the recipient's market behaviour should not retroactively unsettle a transfer unless the behaviour falls within a rule that existed and was intelligible at the time. Finality is what turns a registry entry into a reliable market object.

The fourth block is proportional remedy. AFRINIC should separate record correction, billing enforcement, application fraud, abuse contact failure, regional-use breach, and transfer-condition breach. Each category should have its own notice, cure period, appeal path and remedy range. Revocation or reclamation should be the last resort for the most serious cases. Suspension of future requests, update requirements, warnings, narrow record locks, or conditional approvals may solve many problems without threatening downstream continuity. A registry that reaches first for the harshest remedy invites the harshest litigation response.

The fifth block is policy transparency around leasing. Leasing is the market's answer to firms that need IPv4 capacity but do not want or cannot complete a transfer. It can support efficient temporary use. It can also hide control chains, weaken abuse accountability and create customer continuity risk. AFRINIC should not pretend leasing does not exist. It should state what registry records must show, who is responsible for abuse and contact accuracy, whether sub-allocation or assignment objects are required, how regional-use policy applies, and which forms of leasing create transfer-like obligations. Clarity would reduce both under-the-table arrangements and overbroad enforcement.

The sixth block is independent review. A transfer denial or resource hold should not depend solely on the same staff or board authority that developed the disputed interpretation. AFRINIC's policy development process already contains appeal concepts in other contexts, but transfer-market disputes need a specialised mechanism with commercial tempo. Months of delay can kill transactions. An independent review panel, tightly scoped to transfer and resource-status questions, would lower the incentive to go directly to court. Its decisions should be reasoned, published in anonymised form where confidentiality requires it, and capable of building precedent.

The seventh block is data publication. Without disclosing sensitive customer information, AFRINIC can publish aggregate transfer statistics: requests received, approved, rejected, pending, average processing time, reasons for rejection by category, resources under court hold, and broad outcomes of appeals. This would let members judge whether policy is conserving resources, choking liquidity or merely processing ordinary scarcity. A transfer market cannot be trusted if no one can see whether the rules are being applied consistently.

None of this requires AFRINIC to become a commodity exchange. It requires AFRINIC to recognise that its ledger supports a market whether or not it wants to celebrate that market. The registry's public-interest role is not to abolish the price of scarcity. It is to ensure that scarcity does not corrupt the ledger, capture governance or produce arbitrary deprivation of operational reliance. A neutral architecture would allow addresses to move when transfer criteria are satisfied, block movement when real registry risks exist, and keep the distinction between those two cases visible.

Such a system would also protect AFRINIC itself. Clear rules reduce litigation exposure. Published categories reduce accusations of selective enforcement. Independent review reduces pressure on the board. Finality makes transfers more valuable and more predictable. Proportional remedies prevent ordinary compliance problems from becoming existential disputes. The registry would still say no when the ledger requires it. It would say no as a registry, not as an economic gatekeeper improvising public policy transaction by transaction.

Uncertainty and watchpoints

Several uncertainties remain material for public readers watching AFRINIC's transfer-market architecture. The first is legal continuity. The winding-up proceedings, the public interventions by ICANN, and continuing litigation involving Cloud Innovation, Larus and AFRINIC are not background noise. Watch whether courts in Mauritius clarify the non-asset status of numbering resources while also explaining what reliance interests resource holders retain. A decision that protects the registry from liquidation without addressing member reliance may stabilise the institution but leave the transfer market uncertain. A decision that treats address rights too much like private assets may protect holders while weakening the coordination model. The useful outcome is a careful middle path.

The second watchpoint is the legitimacy of the board and member register. AFRINIC reportedly elected directors in 2025 and signalled recovery activity in 2026, including budget and strategy work. That is progress only if ordinary governance becomes routine. Public readers should look for board minutes, audited membership processes, clear conflict rules, credible meeting notices, published policy-development activity and transparent handling of voting authority. If member status and voting credentials remain contested, every transfer-policy decision will inherit that distrust.

The third is whether AFRINIC treats the 2026 recovery period as operational repair or as a mandate for broad gatekeeping. The February 2026 report that AFRINIC had hundreds of thousands of unallocated IPv4 addresses left makes the next phase important. As the remaining pool shrinks, pressure to police transfers and leasing will rise. Watch whether the registry publishes precise transfer criteria, processing statistics and appeal mechanisms, or whether it relies on general language about misuse, regional protection and stability. General language is easy to applaud and hard to transact against.

The fourth is the revised ICP-2 process. A lifecycle policy for RIR recognition, assistance and possible derecognition is necessary after AFRINIC's crisis. But a global rescue framework can itself become a gatekeeper if triggers are vague. Readers should watch whether the revised policy is limited to registry continuity, data integrity and service stability, or whether it creates broad authority for the global coordination layer to steer contested regional policy choices. Emergency guardianship is most legitimate when it protects the ledger and least legitimate when it chooses economic winners.

The fifth is the treatment of leasing and out-of-region use. Public materials from NRS and Larus show a market-side push to frame IPv4 as strategic capital and registry risk as a business-continuity problem. AFRINIC and allied voices warn that such models can drain African-issued resources into global inventory and undermine regional development. The evidence to watch is not rhetoric from either side. It is policy text, transfer approvals, court orders, registry record practices, abuse accountability, and whether regional networks can obtain the IPv4 they need without entering opaque or politically exposed arrangements.

The sixth is ledger auditability. The 2019 KrebsOnSecurity reporting remains a warning about the value of dormant, misregistered or historically ambiguous address space. Public readers should watch for signs of transparent remediation: documented categories for historical irregularities, clean treatment of legacy resources, clear merger-and-acquisition procedures, and explanations of how disputed records are frozen, corrected or released. A transfer market built on unclear provenance will invite both fraud and over-enforcement.

The final uncertainty is cultural. AFRINIC can recover as a neutral registry only if its community stops treating every market question as a loyalty test. A member who believes in regional development may still need liquid transfer rules. A holder that defends commercial reliance may still owe accurate records and policy compliance. ICANN can defend continuity without becoming a regional policymaker. Courts can protect the registry function without deciding the optimal IPv4 market. The registry's task is to make these distinctions operational.

AFRINIC's transfer market will not be judged by whether it satisfies every ideological camp. It will be judged by whether addresses can move from legitimate sources to accountable recipients through a process that is accurate, predictable, reviewable and resistant to capture. IPv4 scarcity made economic conflict unavoidable. Good architecture can keep that conflict from consuming the ledger. Bad architecture will do the opposite: turn every transfer into a governance fight, every governance fight into a court case, and every court case into another reason to discount AFRINIC's registry record. The region needs a registry strong enough to protect its records and modest enough to remember that a ledger is not a throne.