Scarcity turned a registry question into a balance-sheet question
IPv4 leasing is often described as a broker market, a temporary workaround for networks that still need old address space while the industry moves, slowly and unevenly, toward IPv6. That description is too small for the AFRINIC case. Leasing is not merely a way to match idle addresses with operators that need them. It is an economic response to scarcity, legal uncertainty and the fact that a registry database entry can carry more operational value than the standard contracts around it appear willing to absorb.
AFRINIC is the Regional Internet Registry for Africa and the Indian Ocean region. Its public materials describe it as a nonprofit, member-based organisation registered and operating under the corporate legal framework of Mauritius. It distributes and manages Internet number resources, including IPv4, IPv6, and Autonomous System Numbers. It also provides services such as WHOIS, reverse DNS, Internet Routing Registry, DNSSEC, and RPKI-related functions. In ordinary language, this sounds like technical administration. In a scarcity market, it becomes an institutional control point.
The reason is simple. IPv4 addresses remain necessary for many production networks, even though IPv6 is the long-term protocol answer. Applications, customers, security systems, allowlists, reputation systems and legacy connectivity still make IPv4 reachability economically relevant. AFRINIC's own exhaustion page records the global background: in February 2011, IANA's final IPv4 pool was distributed to the five RIRs; by 24 September 2015, APNIC, ARIN, LACNIC, and the RIPE NCC had exhausted their free pools; AFRINIC later entered Soft-landing Phase 1 on 31 March 2017 and Phase 2 on 13 January 2020. The policy language is about stewardship. The market reads the same facts as rationing.
That conversion from stewardship to rationing explains why leasing has become central. A direct holder pays registry fees and lives under registry policy. A buyer in the secondary market may tie up substantial capital while still remaining dependent on registry recognition, transfer approval, and future review. A network operator that leases can obtain operational use without putting the registry-facing relationship inside its own operating company. A resource holder that leases can monetise unused or underused addresses while keeping formal registration. Each arrangement solves a real commercial problem. Each also creates a visibility problem for the registry.
"Shadow allocation" is a useful term for that visibility problem, but only if it is used carefully. It does not have to mean theft, hijacking or fraud. It can mean that the economically relevant user of an address block is not the entity the registry sees as the formal resource holder. It can mean that customer assignments, leasing chains, operating geography or beneficial-use arrangements sit downstream of the registry record. In its darker form, it can also mean manipulated or stale records, dormant organisations, and address blocks that have drifted away from the entities for which they were originally recorded. AFRINIC has had to confront both the legitimate commercial version and the alleged abuse version.
The public record shows why this became a governance question. KrebsOnSecurity reported in December 2019 on allegations that AFRINIC address records had been manipulated and that companies tied to Ernest Byaruhanga, then an AFRINIC policy coordinator, had sold address blocks associated with African entities. AFRINIC's then chief executive told Krebs the organisation was investigating. Internet Governance Project then reported in 2021 that AFRINIC's dispute with Cloud Innovation, a large holder of AFRINIC IPv4 space, escalated into litigation and a court-ordered freeze affecting up to USD 50 million in AFRINIC bank accounts. The Number Resource Organization stated in September 2023 that the Supreme Court of Mauritius had appointed an official receiver to preserve AFRINIC's business, oversee elections, and move the registry back toward functional governance. The Register has since followed board elections, annulled votes, proxy allegations, ICANN interventions, renewed litigation, and AFRINIC's own claims that adversarial proceedings were still impeding recovery.
Those events are not all the same. Allegations about record manipulation are not the same as lawful leasing. A contractual dispute is not the same as a criminal finding. A receiver is not proof that the registry function failed. But together they show why IPv4 leasing cannot be treated as a simple broker question in the AFRINIC region. The registry does not merely record a market. It helps define which claims are legitimate, which uses remain recognised, and which operators can depend on continuity.
The administrative price and the market price diverged
AFRINIC's fee schedule makes the institutional mismatch visible. Members pay allocation or assignment fees and annual membership fees. The annual fee category is tied to billable resources held; new members are evaluated before receiving services; later resource requests can trigger further fees; accounts must be in good standing before certain transfers are considered. This is a cost-recovery and membership model, not an auction. It finances registry operations and uses administrative categories to maintain a public coordination function.
The market does something different. Internet Governance Project's 2021 analysis reported that IPv4 transfer-market prices rose from about USD 8 per address in 2017 to about USD 30 by 2021. It used a /16 block of roughly 64,000 addresses to illustrate a market value near USD 2 million. IGP also reported that Cloud Innovation had received rights to millions of IPv4 numbers from AFRINIC and leased addresses to customers at a few dollars per address per year while paying AFRINIC registry fees that were tiny relative to the revenue stream IGP described. Those figures should be read as IGP's reporting and interpretation, not as a court finding about every customer, contract, or policy breach. The economic mechanism, however, is hard to ignore.
When administrative fees and market prices diverge, three behaviours follow. Applicants seek scarce addresses through policy channels if they can satisfy the documentary threshold. Holders with large allocations treat them as strategic assets even if the registry does not call them property. Intermediaries and lessors build businesses around converting registration into usable capacity. The divergence does not prove misconduct. It makes optimisation predictable.
AFRINIC's Soft Landing policy was an attempt to manage that reality through rules rather than price. The public exhaustion page explains that the policy was created to guide the membership through scarcity and support a smoother transition to IPv6. It imposed phase triggers, request handling procedures, efficiency requirements, and later, in Phase 2, much smaller allocation and assignment sizes. In Phase 2, the public page describes a minimum size of /24 and a maximum of /22 per allocation or assignment, with additional-resource requests tied to efficient use of resources already delegated. The policy manual frames IPv4 management around uniqueness, registration, conservation, aggregation, and fairness.
The policy logic is defensible. A registry should not simply hand out the remaining pool to whoever asks first or has the best lawyers. It should preserve resources for genuine network use, register assignments accurately, and prevent waste. Yet scarcity policy is never neutral once an external market price exists. A /22 issued under an administrative process is not worth merely the administrative fee to the recipient. It has operational value, scarcity value, and optionality. Even if the holder cannot or should not sell it freely, it can support customers, delay renumbering, reduce market purchases, or become part of a leasing structure.
That is where the legitimacy dispute begins. AFRINIC can say, consistent with conventional RIR doctrine, that number resources are public resources distributed under policy and not owned as ordinary property. Resource holders can reply that, whatever the doctrinal label, they rely on these resources in contracts, routing, customer service, security operations, and revenue plans. A court can accept that addresses are not corporate assets available for distribution in a winding-up while still having to consider the commercial reliance built around registered use. The doctrine solves only part of the problem.
Leasing is a continuity product, not only a capacity product
The public marketing of LARUS is useful because it states openly what many operators think privately. Its site presents first-party IPv4 leasing as a way to obtain production IPv4 capacity without moving registry-layer contract exposure into the customer's own operating company. It emphasises a direct address pool, the absence of a broker chain, continuity controls, renewal certainty, reverse DNS, abuse handling, geolocation support, and routing validity. In its own words and framing, the product is not merely addresses. It is continuity.
That commercial claim should be treated as a claim by an interested market participant. LARUS and Cloud Innovation share leadership with Lu Heng, and The Register has reported ongoing disputes among AFRINIC, Cloud Innovation, LARUS, NRS, ICANN, and others. AFRINIC has challenged some public representations about court recognition and leasing. In May 2026, The Register reported that AFRINIC issued communiques denying that a court order had established or approved the "Court-Ordered Shareholder-Position Continuity Structure" described in LARUS materials, and that the Supreme Court of Mauritius issued an interim order aimed at statements falsely attributing judicial approval or endorsement to the court. LARUS and Cloud Innovation disputed AFRINIC's characterisation and said the order did not decide IPv4 leasing, ownership, or their business model. That dispute remains a live reason for caution.
But the market logic behind leasing is broader than one company's claims. Buying IPv4 can lock capital into an asset whose legal character remains contested and whose usability depends on registry records. Direct holding can expose an operating company to payment obligations, audits, policy review, transfer restrictions, contact-data accuracy duties, and possible termination or revocation mechanisms. Leasing turns a large capital purchase into an operating expense. It also moves some registry-facing risk upstream to the lessor, at least contractually. For an operator that mainly needs continuity of service, this can be rational.
The operator's calculation is not sentimental. Renumbering a live network can mean customer disruption, firewall and access-control changes, allowlist resets, routing changes, abuse-desk work, geolocation errors, engineering time, lost sales and contractual claims. LARUS's public materials make that continuity cost explicit. A network may care less about being the named holder in a registry database than about having a stable, supported, renewable block that does not collapse under a dispute between unknown intermediaries. If a first-party lessor can credibly reduce that failure path, leasing becomes more than a cheap substitute for ownership.
The same arrangement, from the registry's perspective, can look like an evasion channel. If the resource holder remains the formal member while beneficial use is distributed among customers across jurisdictions, the registry may not know who is using the block, where the services originate, whether the original need remains true, or whether abuse contacts and operational records reflect reality. A broker chain makes this worse, because each layer can disclaim responsibility for the next. Even a first-party lessor model, which reduces intermediary risk, still separates formal registration from downstream use.
This is why the phrase "beneficial use" matters. Registry records traditionally answer who holds or has been assigned a resource. They do not always answer who benefits economically from it, who controls customer assignment, who bears operational consequences if it is withdrawn, or who can change its routing and abuse posture. Leasing turns the resource ledger into a layered contract stack. The registry sees one layer. The market operates through several.
Shadow allocation has lawful and unlawful forms
AFRINIC's experience warns against treating every shadow arrangement as the same kind of problem. The 2019 KrebsOnSecurity report belongs at one end of the spectrum. It described allegations that address blocks set aside for African entities were quietly commandeered, that records involving old or defunct organisations had been altered, and that companies connected to an AFRINIC insider had sold address space. Krebs reported that researcher Ron Guilmette had tracked the issue for years, that the market value of the alleged diverted addresses exceeded USD 50 million, and that AFRINIC said an investigation was underway. Those allegations, if proved, concern record integrity and possible abuse of institutional access.
Lawful leasing sits elsewhere. A registered holder may have addresses, customers may need addresses, and the parties may contract for use. The holder may provide abuse handling, reverse DNS, ROA support, geolocation assistance, and network documentation. The customer may never claim to be the holder. The registry may still have accurate information about the formal member. The shadow lies not in a false record, but in the incomplete visibility of the economic user.
Between those poles are harder cases. A holder may have obtained addresses under a need statement that later becomes outdated. It may lease to customers outside the region. It may treat assignments as customer turnover rather than a material change requiring fresh approval. It may provide incomplete downstream documentation because customer identities are commercially sensitive or because the policy does not clearly require disclosure. The registry may suspect that allocation intent has been bypassed, but the holder may argue that networks evolve and that constant re-justification would turn the registry into a business-plan regulator.
IGP's 2021 account of the Cloud Innovation dispute captured this tension. AFRINIC, according to IGP, questioned discrepancies between registered usage and actual deployment, alleged inconsistency with the original need, and invoked a regional-service concept. Cloud Innovation resisted, arguing that requiring renewed approval for changing use would be intrusive and operationally unrealistic. IGP's own view was sharply critical of AFRINIC's enforcement posture and also critical of Cloud Innovation's litigation tactics. For the leasing question, the important point is not to adopt IGP's conclusion wholesale. It is to recognise the institutional dilemma: a registry cannot ignore post-allocation use, yet it cannot feasibly supervise every customer-level change without becoming a gatekeeper over commerce.
That dilemma is intensified by the scarcity price. If addresses were abundant and valueless, downstream opacity would be a smaller concern. If a holder changed customer mix, the registry might care mainly about contact accuracy and abuse handling. Under scarcity, the same opacity can look like hoarding, regional evasion, arbitrage, or private resale of a public resource. The registry's instinct is to demand visibility. The holder's instinct is to protect commercial flexibility. The operator's instinct is to protect continuity. All three instincts are rational.
Beneficial use is the missing data layer
Good registry governance in a leasing world needs a distinction between formal holding and beneficial use. Formal holding asks who is accountable to the registry. Beneficial use asks who is actually using the addresses, under what operational arrangement, and with what continuity dependency. The distinction is familiar in finance, company law, and sanctions compliance, but less developed in number-resource governance. AFRINIC's case shows the cost of that gap.
The policy manual states that allocations, assignments, sub-allocations, and other resource assignments must be registered in the AFRINIC database, that registration data must be correct, and that this is necessary to support network operations. It also says the goals of the registry system include uniqueness, registration, conservation, and aggregation, and that documentation helps keep decisions transparent and honest. That is a ledger model. It works best when the ledger records the facts that matter.
Leasing complicates the facts that matter. The registry may not need every customer contract. It may not need pricing terms, customer identities for small assignments, or detailed business plans. But it may need enough information to distinguish a holder using addresses in its own infrastructure from a lessor assigning capacity to many downstream operators; to identify the entity responsible for abuse response; to know whether geolocation, reverse DNS, and routing records can be corrected; to determine whether regional-use conditions, if valid, are being observed; and to assess whether a request for additional space reflects real unmet need or merely monetisation of prior allocations.
Without such a data layer, enforcement becomes binary. The registry either accepts the formal holder's record as enough, which invites shadow allocation, or demands broad customer-level proof, which invites litigation and claims of commercial overreach. A better system would define categories: direct operational use, customer assignment by an LIR, first-party leasing from a registered holder, brokered leasing, temporary assignment, transfer pending approval, disputed entitlement, and legacy or stale-record remediation. Each category would carry different documentation and visibility duties.
This would not settle every controversy. Some resource holders would object that even category-level reporting exceeds the policy bargain. Some registries would object that it leaves too much room for evasion. But it would reduce improvisation. It would also protect legitimate operators from being swept into enforcement aimed at bad actors. If a lease customer has stable routing, current abuse contacts, clear reverse DNS arrangements, and a direct contract with a first-party lessor, the risk profile is not the same as a block routed through a chain of unknown intermediaries after a questionable record change.
AFRINIC's shadow-allocation problem is therefore not solved by declaring leasing good or bad. It is solved, if at all, by making the layers visible enough that the registry can enforce narrow rules without claiming a permanent veto over every commercial use.
Capital lock-up explains why operators prefer layered arrangements
IPv4 scarcity imposes capital costs on operators. A company that buys a block must fund the purchase, complete transfer or registration processes, maintain the registry account, handle policy obligations, manage abuse, support reverse DNS and RPKI, and accept that the public record may attract scrutiny. The capital then sits in a specialised asset whose value depends on routability, reputation, transferability, and registry recognition. For some operators, especially large networks with long planning horizons, that may be acceptable. For others, it is an inefficient use of capital.
Leasing changes the balance sheet. The operator buys capacity for a period rather than an address asset. It can scale up or down. It avoids a large upfront purchase. It may obtain support for registry-facing functions from the lessor. It can test markets or support customer demand without waiting for a transfer process. This is why leasing persists even when address purchase is available. It is not merely cheaper in a headline sense; it better matches uncertain demand and reduces capital lock-up.
The resource holder has the mirror incentive. If it holds addresses that are not fully used internally, selling may end optionality while triggering transfer complexity. Keeping them idle wastes economic value. Leasing creates recurring revenue while preserving the formal registration position. IGP's 2021 account of Cloud Innovation framed that model as arbitrage between AFRINIC's administrative cost structure and the global value of IPv4. LARUS frames its own current public model as continuity-focused first-party leasing rather than broker spread. The rhetoric differs; the economic base is the same. Scarcity turns address control into a capital instrument.
Registry policy was not designed primarily as capital-market policy. The AFRINIC fee schedule speaks in terms of membership categories, allocation fees, assignment fees, payment timelines, and good standing. The policy manual speaks in terms of need, uniqueness, registration, conservation, and aggregation. When those documents meet a leasing market, they suddenly do work they were not written to do. They allocate scarcity rents. They determine who can monetise old allocations. They define which network operators must buy, lease, renumber, or wait.
That does not mean a registry should become a price regulator. It does mean that pretending there is no capital market around IPv4 is bad governance. If the registry does not recognise leasing, leasing will continue as a shadow practice. If the registry recognises leasing without visibility, policy goals weaken. If it prohibits leasing without feasible alternatives, operators will seek workarounds and litigate. Institutional stability lies in acknowledging the capital problem while narrowing the registry's discretion.
The broker question is really a registry-risk question
Lu Heng's public note index includes a May 2026 entry titled "On Why i.LEASE Exists - and Why the Broker Question Is Really a Registry-Risk Question." The body snapshot available locally is blocked by a security challenge, but the title is consistent with LARUS's public positioning: the relevant comparison is not simply broker versus lessor, but where registry-layer risk sits. NRS likewise frames number-resource governance around money, records, votes, scarcity, economic value, and decentralisation. These are advocacy positions, but they identify a real structural issue.
A broker normally matches buyers, sellers, lessors, and lessees. The broker may help with paperwork, pricing, diligence, and transfer mechanics. But once the transaction is complete, the operational risk belongs elsewhere. If the chain includes multiple intermediaries, each layer adds counterparty risk. If the registry challenges the formal holder, customers far downstream may discover that their capacity depends on documents they never saw and policies they never negotiated.
A first-party lessor claims to solve part of that problem by leasing from its own pool. The lessee's counterparty is the holder or a related entity controlling the resource layer, not merely a matchmaker. That can reduce chain risk. It does not eliminate registry risk. The holder still depends on the registry's recognition, policy interpretation, court orders, and database state. LARUS's argument is that a specialised lessor is better placed to absorb that upstream risk than a customer whose core business is hosting, cloud, telecom, advertising technology, or enterprise infrastructure.
AFRINIC's legal history makes that claim plausible as a category, even if one disputes LARUS's specific assertions. The Register reported in March 2026 that AFRINIC accused Cloud Innovation, LARUS, and associated advocacy campaigns of a web of litigation and procedural roadblocks. It also reported Lu Heng's response that the structural issue was a registry model concentrating high-consequence power over economically critical number resources without matching liability. That is the registry-risk thesis in plain form. The argument is not that every holder should be free from policy. It is that registry power can cause damage far beyond the administrative fees and liability language surrounding the relationship.
AFRINIC's rebuttal, also reported by The Register, is grounded in the traditional view that IP addresses are not owned as property in the ordinary proprietary sense. That is an important principle. It prevents a local corporate liquidation from treating number resources as inventory to be distributed to creditors. It supports ICANN's May 2026 intervention in the winding-up case, where ICANN said it wanted the court to understand AFRINIC's unique role and the nature of the resources it administers. It also protects the broader coordination system from conflicting claims.
Yet "not property" is not the same as "no economic reliance." A hospital, bank, cloud provider, access network, or hosting company does not need a philosophical property right to suffer serious harm from address instability. The registry-risk question sits in this middle ground. The resources may be public identifiers administered under policy, while the operational dependence on their stable registration is private, real, and capitalised into contracts. Leasing exists because operators want that dependence managed.
AFRINIC's governance crisis made shadow allocation more consequential
Shadow allocation is easier to tolerate when the registry is trusted, the policy process works, and disputes are rare. AFRINIC has been operating under a heavier burden. The NRO's September 2023 statement welcomed the appointment of an official receiver as a route toward functional governance, continued registry services, board election, and CEO appointment. The statement thanked AFRINIC staff for maintaining operations and services during difficult times. That is a continuity signal, but also an admission that the institution needed extraordinary stewardship.
The Register's later chronology shows how difficult the return to normal governance has been. In April 2025 it reported that AFRINIC, without a board for years, was preparing elections under a receiver, with concerns about potential interference. In June 2025 it reported ICANN concerns about election integrity, Cloud Innovation's erroneous listing in Mauritian corporate records, and the court's decision not to reconstitute the nominations committee. Later that month it reported that the election was suspended and annulled after questions about powers of attorney and voting documentation. In July 2025 it reported calls for more transparency and Cloud Innovation's move toward winding-up proceedings. In September 2025 it reported that a fresh election had produced eight directors but that legal and institutional risks remained.
In February 2026, The Register reported more positive signs: AFRINIC said it was close to delivering a budget and action plan, morale had improved, interim management roles had been appointed, and the board was working on a 2027-2030 strategy. The same report noted a figure of 773,376 unallocated IPv4 addresses remaining. A registry emerging from crisis while still holding a residual pool of scarce IPv4 is not merely recovering from a governance problem. It is recovering while still administering a valuable economic bottleneck.
That matters for leasing. If the registry's board legitimacy, bylaws, member categories, or resource-review powers remain contested, then every lease built on AFRINIC space inherits some governance risk. If AFRINIC asserts stronger regional-use, transfer, or commercialisation controls, lease economics change. If courts limit AFRINIC's discretion, resource holders gain bargaining power. If ICANN and the RIR system develop stronger lifecycle or emergency-recognition rules, the registry may gain external support but also face new compliance oversight. Each path affects the price and structure of IPv4 use.
The market will not wait for perfect institutional clarity. Operators still need addresses. Lessors still seek revenue. Holders still face capital choices. Customers still need continuity. Without clear rules, each party will create private solutions that push complexity into the shadows. That is how a registry crisis becomes a market-structure crisis.
Registry visibility is better than moral prohibition
Public debate often treats IPv4 leasing as a morality test. One side sees hoarding, arbitrage and extraction from a resource meant for network development. The other sees efficient reuse, capital discipline and protection against registry overreach. Both accounts contain truth. Neither is a policy.
Moral prohibition has a poor record in scarcity markets. If demand is real and supply is limited, a ban without a workable allocation alternative usually creates opacity. Parties rename the transaction. They call it managed service, hosted infrastructure, customer assignment, continuity support, sponsored routing, or network partnership. Some arrangements will be legitimate. Some will be evasive. The registry will then have to infer economic reality from incomplete records, adversarial correspondence, and routing observations. That is the worst possible setting for fair enforcement.
Visibility is more promising. A registry could require that members disclose whether address space is used internally, assigned to customers as an ordinary LIR function, leased on a first-party basis, brokered through intermediaries, or held for future deployment. It could require current abuse contacts and operational escalation paths. It could require that lessors identify the entity responsible for reverse DNS, RPKI/ROA coordination, geolocation correction, and customer termination. It could require aggregate regional-use reporting where policy makes geography relevant, without demanding unnecessary disclosure of every customer identity. It could distinguish temporary assignments, infrastructure customers, and commercial leases.
The point would not be to bless every lease. It would be to make enforcement evidence-based. A lease with clear accountability, no misleading registry record, current operational contacts, and documented customer management is not the same as a dormant block quietly routed by unknown parties. A first-party leasing structure is not the same as a long broker chain. A post-acquisition customer assignment is not the same as a fraudulent record change. The registry's remedy should follow the risk.
AFRINIC's public policy manual already contains the principle that documentation is necessary for transparent and honest decision-making. That principle can be updated for a leasing world without turning the registry into a supervisor of every customer contract. The alternative is selective enforcement: a registry notices some leasing models because they are public, misses others because they are hidden, and then becomes vulnerable to accusations that policy is a weapon rather than a rule.
Legitimacy depends on proportional remedies
The most dangerous registry remedy is total revocation or forced return when the breach is disputed, the downstream effects are large, and the policy text is ambiguous. That does not mean revocation should never occur. Fraud, fabricated organisations, hijacked records, nonpayment after fair process, or clear violations may justify strong action. But under scarcity, the remedy can become more destructive than the wrong it is meant to correct.
IGP's 2021 account argued that AFRINIC's attempt to reclaim Cloud Innovation's resources was disproportionate and legally risky. AFRINIC and its supporters have advanced a different view, focused on policy compliance, regional purpose, and the integrity of the registry system. Courts and formal proceedings must resolve specific legal questions. The governance lesson is broader: when the registry's remedy can interrupt customers, revenue, routing, and institutional solvency, proportionality is not a soft value. It is risk management.
Proportional remedies would separate categories of failure. Incorrect contact data should trigger correction. Missing abuse escalation should trigger operational remediation. Unclear downstream use should trigger disclosure within defined limits. Suspected misrepresentation at application should trigger a focused review of the original facts. Fraud should trigger recovery and, where appropriate, referral to authorities. Commercial leasing inconsistent with a clear policy should trigger a transition path, not an instant cliff for innocent users. Repeated refusal to comply after due process can escalate. The ladder matters.
This approach would also protect AFRINIC. The registry's 2021 experience shows that aggressive remedies can backfire institutionally. IGP reported a court-ordered bank freeze; NRO later welcomed receivership partly to preserve services and restore governance; The Register has reported years of litigation and public controversy. A registry that wants to enforce scarcity policy must survive the enforcement. Narrow remedies are not weakness. They are institutional self-preservation.
For resource holders, proportionality would reduce the incentive to litigate every inquiry as an existential threat. For operators, it would reduce the fear that a lessor's dispute with AFRINIC could abruptly destroy service. For the public, it would make the difference between abusive shadow allocation and ordinary commercial leasing easier to see.
The regional-use dispute sits under the leasing dispute
AFRINIC's service region gives the leasing question a geographic edge. The registry exists to serve Africa and the Indian Ocean region. Its policy process is meant to reflect the regional internet community. Its remaining IPv4 pool was justified publicly as a tool to support regional needs during transition. If large volumes of AFRINIC-issued space are leased outside the region, critics see extraction from African internet development. If regional use is policed too broadly, holders see a post-allocation restraint on global routing and ordinary network commerce.
IGP's 2021 analysis was sceptical of strong regional-use enforcement and argued that Africa's long-term internet growth would not be sustained by the remaining AFRINIC pool alone. AFRINIC's public materials, by contrast, emphasise regional stewardship, community policy, and the need for justified use. The Register reported in March 2026 that AFRINIC had adopted a transfer policy that, in many circumstances, prevents members from transferring assigned IPv4 assets outside the region it administers, and that supporters viewed the policy as frustrating models dependent on treating African-issued number resources as liquid inventory for global leasing or export. That reporting should be read as a snapshot of a contested policy environment, not as the final word on every lease.
The economic problem is that IPv4 is globally routable while registry legitimacy is regional. An address issued in one region can be used to serve customers elsewhere, announced from networks elsewhere, or embedded in multinational infrastructure. A strict regional-use rule may preserve the original allocation rationale but collide with global network design. A permissive rule may reflect operational reality but weaken the regional community's claim that scarce local resources are being stewarded for local development.
Leasing intensifies this conflict because it unbundles formal holding from customer location. A holder in the AFRINIC region may lease to operators abroad. A multinational operator may use African-issued space in a global architecture. A hosting or cloud business may assign addresses dynamically across customers whose end users are everywhere. A policy that treats geography as simple can become administratively unrealistic. A policy that ignores geography can become politically illegitimate.
The best answer is likely not a slogan about borders. It is a clearer test of material regional connection, disclosure, and transition. Which resource categories carry regional-use duties? Do those duties apply only to resources issued after specific policy dates? What counts as originating services in the region? How are multinational architectures handled? What data can the registry request without intruding into customer confidentiality? What remedy applies when a holder's use evolves? These are the questions that determine whether leasing becomes legitimate adaptation or shadow export.
ICANN, courts, and the non-asset principle
AFRINIC's winding-up litigation adds a second layer to the economics of leasing. The Register reported in May 2026 that ICANN successfully applied to become a party to Cloud Innovation's attempt to dissolve AFRINIC. ICANN said its purpose was to help the court understand AFRINIC's unique role and to make clear that numbering resources allocated through AFRINIC are not AFRINIC assets available for distribution in a winding-up. That intervention is important. It protects the numbering system from being treated like a corporate estate.
The non-asset principle, however, should not be stretched beyond its job. It says that AFRINIC does not own the numbers as ordinary assets to be distributed to creditors. It does not automatically decide the rights of a resource holder, the legality of leasing, the meaning of regional use, or the proportionality of a registry enforcement action. Those issues require contracts, policies, member records, court orders, and evidence.
Courts need a precise map. AFRINIC is a Mauritius-registered entity with a regional and global coordination function. Its members and resource holders rely on its records. Its policies are created through a community process but implemented by a corporate organisation subject to local law. Its database entries are not ordinary property deeds, yet they can support substantial operational value. A winding-up court must avoid treating number resources as divisible assets; a contract court must still decide whether a registry or holder breached legal duties; a governance court may need to decide member status, voting rights, and bylaw compliance.
This complexity strengthens the case for registry visibility. If the court record must explain who uses which addresses, under what authority, and with what reliance, the registry should not discover those facts only during litigation. A transparent beneficial-use framework would help courts separate the resource itself, the registry's administrative role, the holder's contractual position, and the downstream operator's continuity interest.
It would also discipline advocacy. NRS, LARUS, Cloud Innovation, AFRINIC, ICANN, ISPA, and other actors have all used strong public language. Some of it reflects genuine institutional concern; some of it is plainly strategic. Public readers should separate facts from claims: a court order is not a marketing slogan; a registry communique is not a final judgment; an advocacy campaign is not neutral evidence; and an official doctrine about property does not erase practical dependence.
What a workable AFRINIC leasing regime would measure
A workable regime would begin by accepting that leasing exists. Denial merely rewards opacity. AFRINIC's public mission is to maintain an effective number-resource registry, not to make economic reality disappear. The policy question is how to bring leasing into a controlled and auditable frame without converting the registry into a price regulator or customer-by-customer licensing authority.
The first measure is accountability. Every leased block should have a clearly responsible registry-facing holder, an operational contact, an abuse contact, and a party responsible for reverse DNS and routing support. If RPKI/ROA coordination is relevant, the responsible party should be known. If a broker is involved, the chain should not obscure who can fix a live operational problem.
The second measure is beneficial-use category. The registry does not need to know every commercial term. It does need to know whether the block is used internally, assigned to ordinary connectivity customers, leased as a capacity product, embedded in managed hosting, used for anycast or infrastructure, or held pending deployment. These categories should be public enough to support policy and private enough to protect customer confidentiality.
The third measure is policy nexus. If AFRINIC policies impose regional-use, demonstrated-need, transfer, or efficiency requirements, the registry should say exactly how those requirements apply to leased space. It should not rely on vague after-the-fact concepts of acceptable business model. Members should know what facts they must update, what changes require approval, and what uses are prohibited.
The fourth measure is remedy ladder. A lease that lacks an updated abuse contact should not face the same remedy as a fraudulent record change. A regional-use dispute should not be treated automatically like address theft. A customer-facing continuity issue should trigger transition duties before revocation where innocent downstream users exist. Strong enforcement remains possible, but it should follow a path that a court and the community can understand.
The fifth measure is institutional separation. Resource review, policy development, board politics, election disputes, and litigation strategy should not be allowed to collapse into one discretionary process. AFRINIC's recent history shows why. If members suspect that resource review is tied to election control or litigation posture, enforcement loses legitimacy even when the underlying concern is valid.
Uncertainty and watchpoints
The first uncertainty is legal. AFRINIC's dispute with Cloud Innovation and related entities has generated years of litigation, and The Register's 2026 reporting shows that winding-up, takedown, bylaw, and other proceedings remain relevant to the public story. Readers should watch court orders rather than claims about court orders. A narrow interim order about public representations is not the same as a final ruling on leasing. An intervention to explain the non-asset status of number resources is not the same as a ruling on every holder's rights.
The second watchpoint is AFRINIC's board and executive recovery. The September 2025 election and February 2026 reports of budget, action plan, interim management, and strategy work are positive signs, but recovery is measured by routine. Does AFRINIC publish clear board records, hold member meetings, maintain policy processes, appoint stable leadership, and reduce litigation-driven paralysis? A functioning board matters because leasing rules, transfer rules, resource reviews, and bylaw reforms require legitimacy.
The third watchpoint is the treatment of the remaining IPv4 pool. The Register's February 2026 snapshot reported 773,376 unallocated IPv4 addresses. As that pool shrinks, the pressure on leasing, transfers, and post-allocation review will increase. If AFRINIC reaches practical exhaustion without a credible leasing and beneficial-use framework, shadow allocation will become more attractive, not less.
The fourth watchpoint is policy language around regional use and transfers. Public reports indicate that AFRINIC has moved toward stronger controls over out-of-region transfer or export in some circumstances. The details matter. A policy that clearly distinguishes new allocations, legacy or pre-policy holdings, intra-region assignments, multinational use, first-party leases, brokered leases, and abusive transactions can reduce conflict. A policy that relies on broad discretion will invite more litigation.
The fifth watchpoint is record integrity. The 2019 KrebsOnSecurity allegations still matter because they concern the registry's core asset: trust in the database. Public readers should look for transparent closure or remediation around historical record problems, not because every allegation should be presumed true, but because unresolved record-integrity questions make later enforcement look selective or reactive.
The sixth watchpoint is the evolution of ICP-2 and emergency RIR lifecycle rules. AFRINIC's crisis pushed the RIR system to consider what happens when a registry becomes dysfunctional. Emergency assistance and derecognition mechanisms may be necessary. They must be narrow enough to protect the ledger rather than create a higher-level gatekeeper whose discretion is even less accountable to the affected region.
The final watchpoint is operator behaviour. If more operators choose first-party leasing over direct holding, that will reveal what the market thinks about registry-layer risk and capital lock-up. If leasing contracts become more transparent about renewal, abuse handling, routing validity, geolocation, and registry dispute contingencies, the market may mature. If instead leasing moves into more obscure chains, AFRINIC and other registries will face a deeper shadow-allocation problem.
AFRINIC's case is not a referendum on whether IPv4 leasing is virtuous. It is a test of whether a registry designed for coordination can govern scarcity without denying the economics scarcity created. Leasing exists because operators need continuity, holders see capital value, and the registry contract is not the whole economic story. Shadow allocation grows when that reality is hidden. The better answer is neither a war on markets nor surrender to arbitrage. It is a registry system that can see enough to govern, restrain itself enough to be trusted, and protect continuity without pretending that scarce IPv4 has no price.