Summary
- Post-exhaustion rules can sound neutral while historical address stock, registry history, customer-assignment archives, routing reputation and financing constraints give incumbents a structural head start over new African network entrants.
- A new African access, hosting or cloud-edge operator can be ready in every visible way and still remain commercially unfinished.
The entrant can be ready for everything except addresses
A new African access, hosting or cloud-edge operator can be ready in every visible way and still remain commercially unfinished. It can have towers leased, fiber ordered, routers installed, staff hired, software tested, peering conversations under way, invoices promised by its first serious customers and a service plan that looks convincing to a lender. It can have exactly the qualities policymakers claim to want from a digital-infrastructure entrant: local knowledge, lower overhead, sharper customer focus, willingness to serve neglected districts and a plan to add resilience rather than merely resell another carrier's product. Yet at the point where the business must become a routed network, one missing input can dominate the spreadsheet: public IPv4.
Across town, the incumbent may be less imaginative. Its customer systems may be older, its product catalog slower to change and its commercial habits less urgent. But it has something the entrant does not. It has address space obtained years earlier, when free-pool allocations were larger and scarcity was less visible. It has registry history. It has assignment files, reverse-DNS patterns, customer numbering habits, upstream relationships, abuse-contact history, routing records, procurement references and a reputation that has been tested in production. It can allocate from inherited stock, clean up old plans, reassign capacity from a shrinking product line or delay a purchase until market conditions improve. It can say to a bank, supplier or enterprise buyer: this network is already real.
The rulebook may treat both firms as applicants, members, resource holders or counterparties. The market does not. The entrant must prove need before scale. The incumbent can show need through scale already built with earlier address endowments. The entrant must buy or lease scarce IPv4 at current prices, or use carrier-grade NAT, cloud NAT and other sharing techniques from the start. The incumbent can grow inside a legacy pool while making more measured decisions about when to buy, lease, renumber or deploy IPv6. The entrant must persuade customers that address sharing will not harm logging, security, allowlists, remote access, hosting, payment systems or compliance. The incumbent can point to continuity.
This is the economics of new-entrant disadvantage. It is not the same subject as the minimum efficient scale of a small ISP, the mechanics of a waiting list, the recovery of unused addresses or the capital value of IPv4 as a balance-sheet line. The issue is narrower and more uncomfortable. Post-exhaustion rules often use neutral language: need, efficient use, complete applications, good standing, conservation, registration accuracy, routing responsibility and small maximum allocations. Each phrase is defensible on its own. Scarce resources cannot be distributed casually. Fraud cannot be ignored. Records must remain accurate. But when these rules meet unequal starting positions, they do not land equally. They interact with history.
AFRINIC is a revealing case because the region combines late institutional creation, a relatively small share of global IPv4, fast growth needs and a registry history marked by scarcity, litigation and governance stress. AFRINIC's published exhaustion material records that the registry entered Soft Landing Phase 2 in January 2020. In that phase, ordinary IPv4 allocations or assignments are constrained to a minimum of /24 and a maximum of /22 per request. Additional requests require high utilization of existing AFRINIC-delegated space. Applications are evaluated through ticketed review; complete files move forward, while incomplete files remain in clarification until they become complete. The policy manual frames the system around familiar registry principles: uniqueness, registration, aggregation, conservation, documentation and fairness.
Those principles are useful as institutional facts. They are not a complete account of economic incidence. A narrow ledger can record equal treatment at the counter while missing unequal conditions outside the counter. If an entrant and an incumbent face the same form, the same words and the same scarcity, the process can look fair. But the incumbent arrives with a stock of usable addresses, a file of past assignments and a history of recognized operation. The entrant arrives with a plan. In a rationed environment, that difference is decisive.
Neutral allocation language can hide inherited advantage
The first mistake in post-exhaustion policy is to confuse identical wording with equal incidence. A rule can apply to everyone and still favor those who entered earlier. A bank that asks every borrower for three years of operating accounts may appear neutral, yet the rule favors firms old enough to have them. A public tender that asks every bidder for a list of comparable projects may appear neutral, yet the rule favors contractors already admitted to the market. A regulator that asks every carrier to prove current customer demand may appear neutral, yet the rule favors carriers with established customer books.
IPv4 scarcity works in the same way. Need-based allocation is often described as a fairness instrument because it asks for operational justification rather than pure purchasing power. That description is only half true. Need is not a natural fact lying on the table. It is an evidenced claim. Evidence requires customers, invoices, network diagrams, subnet plans, upstream assignments, previous utilization, staff knowledge and administrative fluency. The stronger the existing business, the easier it is to document need. The weaker or newer the business, the more its future demand appears speculative even when the commercial opportunity is real.
AFRINIC's policy materials illustrate the problem without resolving it. The policy manual describes conservation as distributing addresses according to actual need and immediate use, while avoiding stockpiling and long-term reservation. It also allows staff to examine materials such as engineering plans, subnetting plans, topology descriptions and routing plans, and expects estimates to be realistic and justifiable. These are sensible safeguards. Yet they create an evidentiary ladder. The operator already serving customers has previous assignments, traffic growth, renumbering experience and operational records. The entrant has forecasts, signed letters, deployment timetables and capital commitments. A process designed to avoid waste will naturally place more weight on what has already happened than on what is about to happen.
That bias is not proof of bad faith. It is a structural feature of the evidence system. Scarcity makes reviewers more cautious. Caution makes documented history more valuable. Documented history tends to belong to incumbents. The result is an allocation regime that can sincerely pursue fairness and still reproduce old advantage.
The phrase "new entrant" should therefore be read broadly. It includes a start-up wireless ISP entering a secondary city. It includes a local data-center firm moving from colocation support into cloud hosting. It includes a regional systems integrator that wants to operate managed connectivity for enterprise clients. It includes a content-edge provider that can reduce latency if it can place services closer to users. It includes a fintech infrastructure provider that needs public endpoints for partners, security controls and audit logs. These firms are not asking for infinite free addresses. They are asking for a way to become credible without first owning the historical proof that only incumbency can supply.
The incumbent's advantage is not merely that it has more addresses. It has the ability to make mistakes privately. A firm with surplus space can test a new product, overestimate a customer cohort, reserve capacity for a migration or renumber slowly without immediately returning to the registry or the market. The entrant has less room for error. If it leases too few addresses, it disappoints customers. If it leases too many, it burns cash. If it buys addresses, it locks scarce capital into an input before revenue has matured. If it waits for a registry process, it may miss the sales window. If it relies on NAT, it inherits support and reputation burdens from day one.
Equal language does not remove these asymmetries. Sometimes it makes them harder to discuss. The applicant who complains about disadvantage can be told that everyone faces the same rule. Formally, that may be true. Economically, it is not enough.
AFRINIC's Phase 2 scarcity turns entry into a sequencing problem
AFRINIC's exhaustion regime is designed for a world in which IPv4 is scarce and IPv6 transition remains unfinished. AFRINIC's published exhaustion page records Phase 2 as the current phase and describes the distribution limits that now frame ordinary requests: /24 at the minimum and /22 at the maximum. It also records that requests are handled through tickets on a first-come, first-served basis, with complete applications proceeding to evaluation while incomplete ones require further interaction until the file is considered complete.
This sounds procedural. For an entrant, it is strategic. A /22 contains 1,024 IPv4 addresses. A /24 contains 256. Those blocks can be meaningful for a small deployment, a hosting start, a management network, a set of public services or a carefully designed access product. They cannot, by themselves, create a large broadband footprint, a scaled cloud platform, a mass-market enterprise access provider or a multi-city service without heavy address sharing or further supply. That does not mean the limits are irrational; AFRINIC's remaining pool is finite. It does mean the entrant must design the business around scarcity from the first customer.
The sequencing burden is severe. The firm needs addresses to acquire customers, but it needs customers to justify addresses. It needs a routable public-address plan to convince enterprise buyers, but it may need enterprise contracts to finance the address plan. It needs lender confidence to buy or lease space, but the lender will ask whether the address supply is stable. It needs upstreams and peers to treat it as serious, but upstreams and peers often infer seriousness from operational history. The entrant is asked to show a future that has not yet been allowed to exist.
Incumbents face scarcity too, but their sequencing is different. An established operator can stage growth inside old holdings. It can move residential customers behind shared addressing while preserving public IPv4 for business products. It can reclaim space from obsolete services. It can audit its own estate before buying. It can carry a few underused ranges as insurance against delayed delivery. It can tell new customers that address availability is a capacity-management issue, not a reason to doubt the firm. Scarcity is a cost for the incumbent. For the entrant, scarcity is a gate at the start line.
Phase 2 also elevates the value of completeness. A complete application is not only a document state. It is a time position. The operator able to assemble a file quickly, respond clearly, reconcile corporate records, sign agreements, produce payment and explain network plans has an advantage over the operator still forming its administrative muscle. Again, the rule is neutral. Complete files should move before incomplete files. Yet the capacity to become complete is itself unequally distributed.
For established members, good standing and previous registry interaction are part of ordinary administration. For new firms, every requirement is part of formation. Corporate registration, tax records, bank channels, signatory authority, technical diagrams, upstream contracts, customer commitments, payment timing and registry account setup may all be happening at once. In a thinly financed entrant, the same two or three people may be negotiating tower leases, responding to enterprise requests, hiring engineers, configuring routing and assembling registry evidence. A delay that looks minor from the outside can consume the margin that made the project viable.
This is why new-entrant disadvantage should not be reduced to a complaint about address quantity. Quantity matters, but timing and order matter more. A small allocation delivered late, after the first anchor customer has chosen the incumbent, may be worth less than a smaller but predictable address path before contracts are signed. A formally fair queue can still impose a hidden delay price. A maximum block size can still force early dependence on leases, transfers or address sharing. A conservation rule can still move commercial risk from the shared pool to the entrant's balance sheet.
AFRINIC cannot manufacture abundance. But it can recognize that scarcity turns entry into a sequencing problem. The fairness test is not simply whether the same words apply to all applicants. It is whether the process gives a real entrant enough predictable capacity to become an operating network rather than merely a well-documented applicant.
Historical address stock is a subsidy without an invoice
The largest incumbent advantage is the one that rarely appears as a subsidy: historical address stock. Older networks received space under earlier conditions, sometimes when demand forecasts were easier to satisfy, pool pressure was lower and market prices were not yet the dominant constraint. They may have paid ordinary registry fees rather than current transfer prices. They may have built products, customer habits, routing reputation and balance-sheet credibility around allocations that were not acquired at today's scarcity cost.
This stock behaves like capital. It reduces cash needs. It permits slower procurement decisions. It lowers dependence on external lessors. It gives the firm a buffer against regulatory delay, market spikes and customer surprises. It allows a sales team to promise public IPv4 for higher-margin products while using shared addressing for lower-margin ones. It supports reverse-DNS continuity, firewall allowlists, mail reputation, customer migrations and operational documentation. It lets a company treat addressing as part of its operating estate rather than a daily market exposure.
The entrant sees the same stock from the other side. It has no old pool to rationalize. If it needs public addresses, it must secure them now, under current scarcity and current institutional uncertainty. If it buys, it pays the market price. If it leases, it pays a recurring charge and accepts renewal, routing, reputation and contract risk. If it waits for a small registry allocation, it may still need supplemental supply. If it builds around NAT, it may lower address consumption while raising engineering, logging and customer-support costs. Every path has a cash cost, a timing cost or a credibility cost.
The policy conversation often treats historical holdings as background. That is misleading. In a post-exhaustion market, inherited IPv4 is not merely a technical convenience. It is a private advantage created by timing. It may be legitimate; the incumbent may have built real networks and served real customers. But legitimacy does not make the advantage disappear. A firm that entered when address supply was easier starts the scarcity era with an option that later entrants cannot replicate through effort alone.
This matters because many fairness debates focus on the final allocation rule while ignoring the initial distribution of economic power. If the final rule says every applicant must prove need, the established operator's need is supported by a business built with earlier allocations. If the rule says every applicant must use previous space efficiently, the established operator can show previous space. If the rule says address stockpiling is disfavored, the entrant cannot build a reserve, while the incumbent's old reserve may be reinterpreted as operational continuity. If the rule says assignments must be registered, the incumbent's old assignment archive becomes a credibility asset.
The problem is not that incumbents should be punished for having history. Network continuity matters. Customers depend on stable numbering. Renumbering is expensive. A policy that casually strips incumbents of address space would damage the connectivity it claims to protect. But the opposite error is to treat inherited stock as if it were the natural result of superior current efficiency. Some incumbents are efficient. Some are not. In either case, their starting point was shaped by time.
There is a distributional irony here. The registry system's conservation language is often aimed at preventing speculative holding. Yet after exhaustion, old holdings function as a reserve for those who already had enough. The entrant is told not to ask beyond immediate need. The incumbent can keep operational slack because it was acquired earlier, embedded in production or hard to separate from customer continuity. This is not necessarily abuse. It is the predictable result of a system that moved from abundance to scarcity without resetting the starting line.
Policy cannot erase history. It can decide whether to see it. Seeing it would change the tone of post-exhaustion fairness. The question would no longer be, "Do all applicants face the same requirement?" It would be, "Does the requirement compound the advantage of those who were present before exhaustion?" That is a better question for AFRINIC, and for any registry that wants fairness to mean more than orderly rationing.
Proof of need rewards the firms already able to scale
Need-based policy is attractive because it seems to distinguish builders from speculators. The builder can show a network plan, customers, equipment and a route to deployment. The speculator cannot. In a world of abundant pool supply, that distinction is useful. In a world of exhaustion, it becomes harder to administer without turning need into a reward for firms that already have market access.
The entrant's problem is circular. To win customers, it may need to prove that it can provide public addressing, stable reachability and clean routing. To prove need for addresses, it may need to show customers. To finance address acquisition, it may need customer contracts. To win customer contracts, it may need financing and addresses. Incumbents can break this circle with inherited stock. Entrants cannot. They must ask counterparties to trust a plan that depends on a resource the plan itself is meant to justify.
AFRINIC's policy manual reflects the standard logic. It asks for immediate use, discourages long-term reservation and requires realistic, justifiable estimates. For first IPv4 allocations to local internet registries, the policy manual speaks of membership in good standing and existing efficient utilization from an upstream provider, with verification tied to registered assignments in recognized databases. For end-user assignments, it describes a /24 general minimum and asks for membership in good standing, evidence of existing efficient use from an upstream provider or an immediate need based on network infrastructure. For further provider-independent space, it asks requesters to show how previous assignments have been used and to provide detail supporting one-year growth.
These are not absurd requests. A registry that did not ask such questions would invite waste and fraud. But the economic effect is to privilege firms with a paper trail. A new access provider may have real demand in a peri-urban district, yet its potential customers may not sign hard commitments until service is available. A hosting entrant may have developers ready to migrate, yet those developers may want to see routed service first. A regional cloud-edge firm may have an architecture that uses public IPv4 sparingly, yet the first enterprise clients may require dedicated addresses for compliance, allowlists or operational separation. The entrant can explain the need. It cannot always prove it in the format that mature networks can.
The proof burden also shapes business models. A firm that knows it will struggle to justify public addresses may design for address sharing from the beginning. That can be efficient. It can also narrow the product set. Some customers will not accept shared egress. Some security teams require static public endpoints. Some payment, gaming, remote-access and enterprise systems are sensitive to shared reputation. Some hosting workloads require separation that is easier with public IPv4. If the entrant cannot offer those services early, it may be pushed toward lower-margin segments while the incumbent keeps the premium segments that require public address confidence.
There is also a signaling effect. Registry recognition is a market signal. A firm with its own allocation or assignment appears more durable to customers, upstreams and lenders than a firm dependent on a chain of leases or upstream assignments. Need-based rules, by limiting the entrant's early access, can inadvertently restrict the entrant's ability to acquire the signal that would help it grow. The rule asks for maturity before granting one of the inputs that would make maturity possible.
One answer is to tell entrants to use IPv6. That answer is directionally correct and commercially incomplete. IPv6 deployment is necessary for the long run. But many users, suppliers, enterprise systems, public services and global counterparties still require IPv4 reachability. Dual stack is not a slogan; it is a cost structure. A new African operator that deploys IPv6 well still needs an IPv4 plan for customers, partners and legacy destinations. If the IPv4 plan is weak, the IPv6 virtue does not close the sale.
The fairness challenge is therefore not to abandon proof of need. It is to design proof paths that understand formation. Evidence from signed anchor customers, equipment orders, colocation contracts, spectrum or tower arrangements, upstream letters, peering plans and audited deployment milestones can support real entry even before a long assignment archive exists. Small, predictable starter allocations tied to clear milestone review may be more pro-competitive than a process that waits for the entrant to become established through borrowed addresses. The aim is not free stockpiling. It is to avoid mistaking a lack of history for a lack of genuine demand.
Assignment archives become market power
An address assignment archive looks administrative. It records which customer, service, site or internal function received which range and on what basis. In an abundant era it was mainly a hygiene tool: useful for troubleshooting, reverse DNS, abuse handling and future requests. In a scarce era it becomes part of market power. The firm that can produce years of structured assignments can turn its past into proof. The firm that is still trying to enter must convert plans into evidence without the same archive.
This distinction matters because assignment records do more than satisfy a registry. They help a firm explain itself to counterparties. A bank can see continuity. An enterprise buyer can see operational discipline. A regulator can see customer responsibility. An upstream can see that the applicant understands delegation, registration and abuse response. A lender may not read the policy manual, but it can recognize the difference between a company with a documented address estate and a company still dependent on promises.
Incumbents also benefit from the way old records normalize old choices. A legacy allocation may contain quiet ranges, awkward numbering, old customer commitments and historical inefficiencies. Because these have been embedded in production, they are interpreted through continuity. The question becomes how to preserve service while improving utilization. An entrant's unbuilt plan is interpreted through scarcity. The question becomes whether the need is sufficiently immediate to justify scarce supply. The same conservation language therefore treats old slack and new slack differently. One is a continuity problem. The other is a request for trust.
That asymmetry is not easy to remove. Existing customers should not be disrupted merely because a newer firm would like capacity. But the asymmetry should be named. If a registry asks for proof of immediate use, it should recognize that a young provider's most relevant evidence may be commercial formation rather than assignment history: customer letters, signed facilities, equipment invoices, address-sharing design, security operations, support workflow and revenue milestones. If only historical assignment tables count as serious evidence, the system has converted incumbency into a credential.
The entrant's archive also takes time to build. It needs customers to create assignments, assignments to create utilization records, utilization records to justify more resources and more resources to win customers. This is the institutional loop at the heart of new-entrant disadvantage. The requirement is individually rational, but its sequencing favors firms that crossed the threshold before scarcity became binding.
Policy should not pretend that paper can solve all of this. It can, however, make the first archive possible. A staged allocation path, clear expectations for customer-level registration, proportionate confidentiality rules and fast updates to public records can help entrants create the documentary history that mature firms already possess. If an entrant is required to document carefully from the first customer, the registry should make that work useful, portable and legible to the market. Otherwise the archive remains an incumbent inheritance rather than a bridge to competition.
Routing reputation widens the credibility gap
Public IPv4 is not homogeneous. Two /24s can contain the same number of addresses and carry very different commercial value. One may have clean routing history, stable reverse DNS, a known holder, good geolocation, no serious blocklist residue and acceptance by upstreams. Another may arrive with old abuse history, stale records, uncertain routing permissions, poor geolocation, inconsistent contact data or suspicion from mail providers and security vendors. Scarcity makes these differences more important because there are fewer easy substitutes.
Incumbents typically possess routing memory. They have prefixes that have been originated for years, customer systems that already recognize them, reverse-DNS conventions, abuse desks, security contacts and upstream filters built around their network. Their problems are visible and therefore manageable. If a range has poor reputation, they may know why. If a customer misused space, they may have records. If a route changed, they may have ticket history. Even when the records are untidy, the network has continuity.
The entrant lacks that memory. If it receives a small new allocation, it must create a reputation from zero. If it leases, it inherits whatever history came with the lessor's block. If it buys in the market, it must conduct diligence on routing, geolocation, prior abuse and registry recognition. If it uses upstream-assigned addresses, it may not control the reputation story at all. The cost is not merely technical. Reputation affects sales. Enterprise customers ask whether services will be reachable, whether mail will be accepted, whether partners will allowlist the range, whether security teams will trust traffic and whether the provider can maintain clean abuse handling.
This is another way historical stock becomes a subsidy. The incumbent's addresses are embedded in customer trust. The entrant's addresses must be explained. A new operator selling to banks, public agencies, logistics firms, clinics, schools, content providers or small businesses may discover that public IPv4 is less like a commodity and more like a passport. A clean passport is easier to use. A new or uncertain one receives more questions.
Registry policy can affect this credibility gap through record quality. AFRINIC's policy manual emphasizes registration accuracy. It treats public registration as necessary for uniqueness and troubleshooting. It requires assignments and allocations to be registered and says data must be correct. The registry also supports services around reverse DNS, WHOIS/RDAP, routing records and route-security publication. When those services are predictable and trusted, a new entrant can partly compensate for lack of age. Clean records become a substitute for brand history. A small operator can say: the registry record is accurate, the contacts work, the route authorization is valid, the reverse delegation is maintained and the abuse desk responds.
When institutional trust is weak, the substitute works less well. If counterparties are already anxious about the registry's governance, litigation, resource review or continuity, the entrant's reliance on registry evidence may not be enough. Large incumbents can overcome doubt with scale and relationship capital. New firms have fewer alternative signals. They need the registry's ordinary credibility precisely because they lack their own long record.
This is why AFRINIC's governance history matters to entry economics. Public analysis by the Internet Governance Project in 2021 described AFRINIC as the last regional registry created and as having only a small share of global IPv4. It also discussed the Cloud Innovation dispute, a provisional bank-account freeze ordered by the Supreme Court of Mauritius, earlier allegations of address-record manipulation by a staff member and wider controversy over the registry's resource review stance. Later public discussion has described receivership, contested governance repair and board-legitimacy questions. The legal merits of specific disputes are not the issue here. The entry effect is that new networks must ask counterparties to trust a registry layer that has become visible as a source of risk.
Reputation compounds. An incumbent has address reputation, corporate reputation and registry history. An entrant may have only a business plan plus whatever confidence the registry can lend. If the registry is boring, the entrant benefits. If the registry is politically or legally noisy, the entrant pays a credibility premium.
Address sharing is not a free substitute for public IPv4
A common answer to IPv4 scarcity is address sharing. Carrier-grade NAT and related architectures allow many customers to share fewer public IPv4 addresses. Cloud NAT, load balancers, proxies and application-layer gateways can reduce direct public-address demand. For some services, this is efficient and necessary. For others, it changes the economics of entry.
An incumbent can choose where to use sharing. It may place mass-market residential users behind carrier-grade NAT while preserving public IPv4 for business customers, hosted services, static-address products, VPN concentrators, payment partners, remote monitoring or security-sensitive clients. It can tune the mix over time because it has inherited stock. It can reserve public addresses for the customers most willing to pay. Address sharing becomes a margin-management tool.
For the entrant, address sharing may be the default rather than a choice. That difference matters. NAT equipment costs money. Logging infrastructure costs money. Support scripts cost money. Abuse investigations become more complex when many users share one public address. Some applications behave poorly. Some customers blame the access provider when a shared address is blocked or when a remote service flags unusual traffic. Lawful-access and incident-response obligations may require careful port logging and retention. Gaming, remote work, video systems, industrial monitoring and some enterprise VPNs may require special handling. None of this is impossible. But it is work placed at the beginning of the entrant's life.
The cost is also commercial. A new operator trying to win its first serious customers often needs a simple value proposition: better service, local support, lower latency, stronger resilience or a tailored product the incumbent ignores. If the sales conversation becomes a technical explanation of shared addressing, port behavior, logging and exceptions, the entrant loses the clarity that new competitors need. The incumbent can sell continuity. The entrant sells explanation.
Address sharing can also push the entrant into dependence on upstreams or cloud platforms. If it cannot obtain enough public IPv4 for its own edge, it may use cloud-based egress, managed NAT services, transit-provider addressing or leased blocks bundled with other services. These paths can be practical. They also move control away from the entrant. A change in price, reputation, geolocation, upstream routing or contract terms can disturb the customer's service. The operator that entered the market to reduce dependence may begin life dependent on address intermediaries.
The policy manual's treatment of private IPv4 captures the technical tension. It notes that hosts using private IPv4 cannot be reached from the Internet unless enabled through NAT, and that some Internet services may not work properly under NAT. That is a technical statement with economic consequences. If the services that do not work well are the ones with higher margins, address sharing becomes not merely an engineering compromise but a revenue constraint.
IPv6 reduces the long-term pressure but does not remove the short-term asymmetry. A sophisticated entrant may design IPv6-first, deploy dual stack where possible and use translation only where needed. That is good engineering. But if customers still require IPv4 reachability, the entrant remains exposed to the scarce input. An incumbent can treat IPv6 transition as a gradual optimization. The entrant may need to explain both why it is modern enough to use IPv6 and why it still cannot avoid IPv4.
This is not an argument against carrier-grade NAT or sharing. It is an argument against treating them as costless answers to new-entrant disadvantage. When address scarcity forces entrants to use sharing earlier and more heavily than incumbents, it changes product design, support cost, customer mix and financing. A neutral scarcity policy has therefore altered the competitive field before the first invoice is paid.
Market access can be fairer than permission, if recognition is predictable
IPv4 commercialization is often criticized as if the only alternative to need-based allocation were a crude auction won by the richest firms. That framing misses the entrant's practical choice. A new operator does not compare the market with an imaginary world of abundant free addresses. It compares market access with rationed access, discretionary delay, small allocations, uncertain transfer recognition, lease risk and the engineering cost of address sharing.
Price is painful, but it is legible. A quoted lease rate or purchase price can be compared with expected revenue, customer commitments, financing terms and alternative designs. It can be negotiated. It can be placed in a budget. It can be shown to investors. It can be hedged by staged growth. Discretion is harder. If the entrant does not know whether a request will be approved, when a transfer will be recognized, whether a use plan will satisfy the reviewer or whether a future policy interpretation will change the status of a deployment, the risk is difficult to price.
Public market commentary from Lu Heng has argued, from a market-participant perspective, that anti-commercialization rhetoric often compares markets with an egalitarian allocation system that never existed. The useful insight is not any one actor's commercial position. It is the institutional point: need-based allocation in an unequal economy tends to read existing scale as legitimate demand. Larger networks have more customers, more staff, more records and more capacity to absorb delay. A transparent market does not remove inequality, but it can reduce arbitrariness if the registry keeps recognition narrow, fast and predictable.
For entrants, the best world is neither unlimited free allocation nor chaotic trading. It is a system in which scarce addresses can move through clean, auditable channels, with accurate records, proof of control, dispute checks, route-security support, abuse-contact clarity and predictable registry recognition. In such a system, a firm that cannot obtain enough from the remaining pool can lease or buy without wondering whether the registry will later reinterpret ordinary commercial use as suspect. Lenders can finance purchases. Customers can understand the continuity plan. Lessors and lessees can price renewal and reputation risk. Buyers can diligence prior history.
The worst world is a hybrid that combines high market prices with permission uncertainty. The entrant then pays scarcity prices while still facing rationing logic. It may buy addresses but remain uncertain about recognition. It may lease addresses but fear policy stigma. It may receive a small allocation but not enough to build the product. It may be told that markets harm poorer regions while discovering that the old allocation system already advantaged larger firms. The rhetoric of stewardship then becomes another cost.
AFRINIC's history shows why this distinction matters. The Internet Governance Project's 2021 analysis described a sharp divergence between the global market value of IPv4 addresses and the low administrative cost at which AFRINIC resources had historically been available to members. It argued that this gap created arbitrage incentives and intensified disputes over regional use. One need not accept every judgment in that analysis to see the economic mechanism. When a scarce input is worth far more in market use than in administrative fees, rules governing recognition and movement become commercially decisive.
For a new entrant, the policy question is not whether someone else once obtained addresses cheaply. That history cannot be replayed. The question is whether today's entrant can reach usable supply on terms that are clear enough to support investment. If the answer is yes, markets can become a bridge across historical disadvantage. If the answer is no, markets become another arena where incumbents with cash, lawyers and address history win.
The registry's role should be limited but serious. It should not certify business wisdom. It should not decide whether a new cloud-edge provider has chosen the correct customer mix. It should not treat every commercial movement as a moral danger. It should ensure uniqueness, accurate records, chain of authority, dispute notation, route-security support and reliable publication. That narrow role can protect the market without turning the registry into a capital allocator.
Registry history becomes collateral history
New entrants do not only need addresses for routers. They need addresses for finance. A bank, equipment vendor, tower company, data-center landlord, upstream provider or enterprise buyer may not understand every detail of regional registry policy, but each asks a practical question: can this operator keep serving customers if growth occurs? The answer depends partly on address continuity.
An incumbent can show a history of registry recognition. It can show address blocks, assignment records, routing continuity, reverse-DNS delegation, customer migrations, abuse handling and staff familiarity with the registry. Even when the bank does not value IPv4 explicitly, that history supports the broader claim that the network is real and durable. It reduces perceived execution risk.
The entrant has a thinner file. It may have a strong plan, signed letters and capable engineers, but it does not have years of address administration. If it relies on leases, the lender may ask about renewal. If it uses upstream-assigned addresses, the lender may ask about portability and customer lock-in. If it builds around NAT, the lender may ask whether enterprise revenue will be limited. If it plans to buy addresses, the lender may ask whether the registry will recognize the transfer, whether the addresses are clean and whether they can support financing security. In each case, the lack of history raises the cost of capital.
This is not the same as treating IPv4 as a balance-sheet asset for accounting purposes. The narrower point is about credit confidence. Address certainty is part of the operating evidence that turns a network plan into a financeable project. If two firms ask for capital to serve the same market, the one with an established address estate can appear safer even if the entrant has better technology and a better customer strategy. Historical stock becomes collateral history.
The effect is strongest where capital is least patient. Many African entrants operate in markets with expensive debt, volatile currencies, thin supplier credit and high working-capital needs. They may need to pay equipment suppliers in hard currency while collecting local revenue. They may face slow public-sector payments. They may need to prepay for transit, power, tower access, software or customer equipment. Adding an uncertain address path to that mix can shrink the project before deployment.
Policy can worsen or improve this. A registry that publishes clear timelines, evidence expectations, transfer conditions and appeal paths helps the entrant explain risk. A registry that is slow, discretionary or politically contested forces the entrant to carry an institutional risk premium. A small operator may not have the bargaining power to pass that premium to customers. It may lower service quality, delay expansion or accept dependence on a larger carrier.
This is why the public ledger must not mistake record history for merit. Old records are useful. They help troubleshoot and prevent duplication. They support accountability. But old records also reflect past access. A firm that lacks them may be new rather than unserious. A lending market that treats registry history as a proxy for quality will tend to favor incumbents unless policy creates credible, predictable ways for entrants to build equivalent proof.
There are practical steps. Starter allocations should be paired with milestone evidence that lenders can understand. Transfer and leasing recognition should produce clean public records quickly. The registry should provide simple confirmation letters or record extracts that help operators explain their status without relying on informal channels. Reverse DNS, RDAP, WHOIS and routing-security publication should be dependable enough that a small firm can point to them as institutional evidence. None of these steps gives the entrant free capital. They reduce needless uncertainty around a scarce input.
The deeper principle is that entry policy and capital policy meet in the same place. If address recognition is uncertain, capital becomes cautious. If capital becomes cautious, entrants scale more slowly. If entrants scale slowly, incumbents face less pressure. A neutral registry delay can therefore become a market-structure decision.
Governance stress raises the new-entrant premium
Institutional instability is not evenly distributed. Large operators can hire counsel, maintain relationships, diversify address supply, buy insurance-like redundancy, join policy debates and wait out disputes. New entrants usually cannot. They depend on the registry being ordinary.
AFRINIC has not been ordinary in recent years. Public reporting has described a registry under unusual strain: allegations of address-record manipulation and improper transfers in the period before 2021; the Cloud Innovation dispute; a Mauritius court order that provisionally froze bank accounts in 2021; arguments over resource review, regional use and contractual authority; later receivership; contested election and board-legitimacy questions; and continuing concern over how institutional continuity is preserved. Some of these matters are disputed. Some are legal rather than technical. All of them affect the economic environment in which entrants ask customers, lenders and upstreams to trust the registry layer.
The Internet Governance Project's 2021 analysis was especially clear on the structural setting. It described AFRINIC as the last regional registry to be created and as having held only a small share of global IPv4, while the African Internet's long-term growth needs far exceed the leftover IPv4 pool. It also argued that the market value of IPv4 and the attempt to govern regional scarcity created incentives for arbitrage and conflict. Those observations matter for entry because they show that AFRINIC's crisis was not merely a personality dispute. It sat at the intersection of scarce addresses, rising asset value, institutional authority and uneven regional development.
For an entrant, governance stress shows up in practical questions. Will a request be reviewed on time? Will a transfer be recognized without unexpected interpretation? Will reverse DNS and routing-security services remain reliable? Will counterparties treat AFRINIC records as stable? Will a court order, receivership direction, board dispute or policy fight slow a decision? Will the registry staff maintain continuity even if governance is contested? The entrant may not be party to any dispute, yet it pays for the uncertainty.
This is why new-entrant disadvantage is not solved by saying that all members share the same registry. They do not share it in the same way. The incumbent has accumulated redundancy: address stock, staff, counsel, existing routes, enterprise contracts, supplier credit and market reputation. The entrant is trying to convert promises into operations. It needs the registry to amplify credibility. If the registry instead requires explanation, the entrant begins with a discount.
Governance stress also changes risk appetite. A new operator considering whether to buy or lease addresses in the AFRINIC context may worry not only about price and block quality but about future policy enforcement, record recognition, court-driven interruptions and reputational controversy. A larger incumbent can absorb such complexity or arbitrage it. A smaller entrant may retreat to a safer but less competitive model: resell the incumbent, depend on upstream addressing, avoid enterprise hosting or delay expansion. The cost is paid by customers who lose a potential competitor.
Receivership and court supervision can preserve continuity in an emergency. They can also remind the market that continuity required emergency preservation. The distinction matters. If a receiver's mandate is understood as stabilizing assets, maintaining operations and restoring governance, that may reassure members in the short term. But entrants still face the question of whether the ordinary institution will remain predictable after the emergency. A registry that needs rescue can continue to function, yet still raise the risk premium for firms with little margin.
The lesson is not that AFRINIC should be written off. The African region needs a functioning registry. Staff continuity, technical services and policy administration matter. The lesson is that governance repair has an entry-policy dimension. Restoring board legitimacy, clarifying authority, constraining review powers, publishing predictable timelines and insulating technical services from political shocks are not merely institutional housekeeping. They lower the cost of entry.
Fairness should mean the chance to become credible
The practical reform principle is simple: fairness after exhaustion should mean a credible chance to become credible. That sounds circular because the market is circular. A new operator needs some recognized resources to prove it can use recognized resources. It needs a small amount of trust to generate the evidence that earns larger trust. If policy refuses that first step, it protects incumbents while claiming neutrality.
This does not require unlimited free addresses. In a Phase 2 world, that would be impossible and unfair. It requires a better allocation of proof burden, timing and certainty. The first step is to distinguish between speculative warehousing and formation slack. A firm asking for a large block without infrastructure, customers or a staged plan should face skepticism. A firm with real equipment, signed facilities, upstream negotiations, anchor-customer letters, staff and a coherent IPv6 and IPv4 plan should not be treated as unserious merely because it lacks a long assignment archive.
Second, the registry should make early evidence paths standard rather than discretionary. New entrants should know in advance which documents matter: corporate registration, beneficial authority, network diagram, autonomous-system plan where applicable, upstream or peering letters, customer commitments, equipment invoices, facility contracts, security and abuse plan, reverse-DNS plan, IPv6 deployment plan, NAT and logging plan, and milestones for utilization. The more predictable the file, the less advantage belongs to insiders who know how to speak registry language.
Third, small initial resources should be paired with milestone review that is fast and proportionate. A /24 may be enough for some starts and too little for others; a /22 may be meaningful but still limited. The key is not only the size but the path after it. If a firm uses the first block well, the route to further supply, lease recognition or transfer recognition should be clear. If it fails to deploy, the consequences should also be clear. Predictability disciplines both the applicant and the registry.
Fourth, transfer and leasing channels should be made less ambiguous. Entrants will need market supply. AFRINIC's remaining pool cannot finance the continent's full IPv4 demand. If transfers or leases are treated as suspicious by default, entrants face a false choice between insufficient pool resources and risky market resources. Clean market recognition helps new firms turn cash and contracts into capacity. The registry can still require accurate records, authority proof, dispute checks and abuse contacts. It need not review the entrant's business model as if it were a venture-capital committee.
Fifth, technical records should be made easy for new firms to maintain. WHOIS/RDAP updates, reverse delegation, routing records, route-security publication and abuse-contact data are the credibility tools available to a firm without long history. If these tools are slow, opaque or difficult, incumbents benefit. If they are simple and reliable, entrants can build trust faster.
Sixth, institutional governance should be evaluated partly by its entry effect. Board legitimacy, receivership exit, court compliance, policy clarity and staff continuity are not abstractions. They determine whether a new operator can show a lender or customer that the registry layer is stable. A registry that wants to support development should measure not only how many resources remain, but how much uncertainty its own processes add to the cost of deploying them.
The moral language of equality is less useful than the economics of incidence. A rule is pro-entry if it lowers the total cost of becoming a real operator without inviting waste. It is anti-entry if it forces new firms to buy at market prices, pass through discretionary review, carry reputation risk, deploy expensive workarounds and still prove maturity before they can obtain the resources that maturity requires. Many rules will sit between those extremes. The task is to see the direction of pressure.
The ledger must not call history neutrality
AFRINIC's registry function is necessary. Public IP addresses require uniqueness. Records must be accurate. Routes, reverse DNS, abuse contacts and authority data must be maintained. A world without a trustworthy registry would be worse for entrants as well as incumbents. The critique is not that the ledger exists. It is that the ledger can be too narrow in its understanding of fairness.
The ledger sees an applicant, a request size, a file, a utilization claim, a payment, a membership status and a set of records. It does not automatically see the years in which one firm acquired addresses before exhaustion and another did not exist. It does not see the customer who refuses to sign until public addresses are confirmed. It does not see the lender adding a risk premium because the entrant relies on leased space. It does not see the support cost of NAT, the sales cost of explaining reputation or the opportunity cost of a delayed review. It does not see the incumbent's quiet ability to allocate from inherited stock.
That narrowness is useful for some purposes. Registries should not become omniscient economic planners. But narrowness becomes dangerous when it is mistaken for neutrality. If the registry says, "We applied the same rule," it may be correct. If it says, "Therefore the outcome is fair," it may be wrong.
The post-exhaustion registry needs a richer fairness vocabulary. It should distinguish formal equality from competitive entry. Formal equality asks whether the same rule applies. Competitive entry asks whether the rule allows a capable newcomer to become a real constraint on incumbents. Formal equality asks whether the file is complete. Competitive entry asks whether completeness requirements are predictable enough for firms without registry history. Formal equality asks whether current need is proved. Competitive entry asks whether some address access is necessary to create the evidence of need. Formal equality asks whether conservation is protected. Competitive entry asks whether conservation has become a way to preserve old endowments.
This is where ledger and gatekeeper roles diverge. A ledger records uniqueness, authority, contactability and routing responsibility. A gatekeeper begins to decide which business models deserve capacity, which commercial arrangements are morally preferred and which operators may turn scarce addresses into growth. The first role is necessary. The second is dangerous. It can turn stewardship into mandate laundering: a narrow technical mandate is used to justify broad economic control. Once that happens, address policy becomes a form of capital-control economics, even if it still speaks in the language of conservation.
The answer is not to replace registry policy with political allocation. That would create new discretion and new unfairness. The answer is to keep the registry's role narrow, documented and predictable while recognizing the economic weight of historical address stock. A narrow registry can still improve entry by publishing clear rules, reducing delay, supporting clean transfers, enabling small staged allocations, maintaining trusted routing records and avoiding retroactive surprises. It can be boring in the best sense.
This matters beyond any single entrant. Entry pressure is a development input. It lowers prices, improves service, increases redundancy, supports local hosting, creates skilled jobs and gives customers alternatives when incumbents underperform. Address policy that quietly favors incumbents is therefore not merely a private inconvenience. It affects market structure. In a continent where many users still need better, cheaper and more resilient connectivity, the difference between formal fairness and entry fairness is material.
AFRINIC's scarcity makes the issue visible, but the lesson is broader. Every post-exhaustion registry faces the temptation to treat old records as proof of neutral merit. The temptation is understandable; records are what registries have. Yet history is not neutrality. It is the accumulated result of earlier timing, earlier policy and earlier market position. A fair ledger must record history without worshipping it.
A pro-entry AFRINIC would be strict, thin and predictable
The reform agenda implied by new-entrant disadvantage is not laxity. It is disciplined restraint. AFRINIC should be strict about uniqueness, fraud, authority, record accuracy, route-security evidence, abuse-contact responsibility and the return of unused resources where policy clearly requires it. It should be thin in the sense of avoiding broad discretionary judgment over business models beyond what is needed to administer scarce resources. It should be predictable because predictability is the entrant's substitute for incumbency.
Strictness without predictability favors insiders. If rules are enforced but the timing, evidence burden and interpretation are uncertain, the firms with counsel, consultants and long registry relationships adapt better. Thinly staffed entrants suffer. Predictability without strictness invites warehousing and fraud, which also hurts entrants by exhausting supply and damaging trust. The balance is strict, thin and predictable.
A pro-entry AFRINIC would publish evidence guides that distinguish first-time local internet registries, end users, adjacent-market entrants and additional-allocation requests. It would provide realistic examples of acceptable deployment evidence without forcing every applicant into a template better suited to incumbents. It would state review timelines and explain what pauses the clock. It would treat file completeness as a clear checklist rather than a moving standard. It would make payment and agreement windows easy to understand, particularly for new members dealing with banking friction. It would maintain a documented route for reconsideration if staff reject a request.
It would also make market supply safer. Public transfer and lease records should show who is responsible, how authority is evidenced, which contacts handle abuse, what routing-security steps are in place and whether any dispute affects the resource. The registry should not have to approve every customer plan. It should ensure that the record layer tells the truth needed for networks to route, contact and trust each other. That is a narrower task and a more valuable one.
A pro-entry AFRINIC would avoid using governance drama as a reason to expand discretion. Institutional crisis often creates pressure for more control: more review, more suspicion, more central approval, more policy rhetoric. Yet for entrants, the antidote to crisis is not thicker discretion. It is bounded authority. After litigation, address-record scandal, receivership and election disputes, members need to know where the registry's power starts and stops. Entrants in particular need to know that their business will not become collateral in a future institutional fight.
The aim is not to tilt the field against incumbents. Incumbents carry real networks and real customers. They should not be destabilized for the sake of symbolism. The aim is to stop treating their inherited address position as if it were merely the reward for current efficiency. A strict, thin and predictable registry protects incumbents from arbitrary disruption while giving entrants a viable path to become credible. That is the right kind of neutrality.
The economics of entry should shape the scarcity settlement
IPv4 exhaustion forces every registry to choose a theory of fairness. One theory says fairness is equal procedure: same forms, same policies, same queue language, same conservation principles. Another says fairness must also account for unequal starting positions: who already has addresses, who has records, who has customers, who can wait, who can finance market supply, who can absorb NAT costs and who can explain registry risk to counterparties. The first theory is administratively tidy. The second is economically serious.
AFRINIC needs the second. The region's growth needs are too large and the remaining IPv4 pool too small for a narrow procedural view to be enough. The registry cannot make every entrant whole. It cannot reverse early Internet history, global address concentration or the late creation of Africa's regional registry. It cannot turn a /24 into a national broadband plan. It cannot make IPv6 instantly universal. But it can decide whether its rules compound inherited advantage or reduce it at the margin.
The core thesis is modest. New entrants should not receive unlimited free addresses. They should not be exempt from accuracy, need, payment, authority checks or conservation. They should not be allowed to warehouse scarce space while real networks wait. But a narrow ledger must not mistake historically accumulated advantage for neutral policy fairness. It must not treat the incumbent's archive as proof that incumbency is more deserving. It must not require the entrant to prove maturity without access to the resource that helps create maturity. It must not hide the cost of discretion behind the language of stewardship.
The entrant in the opening scene is not asking the registry to choose winners. It is asking the registry not to choose the old winner by default. The towers, fiber, staff, software, peering plans and invoices matter. So does the lack of old public-address stock. A fair system sees both. It asks for enough evidence to prevent waste, then gives the entrant a predictable way to build the rest of the evidence through operation.
AFRINIC's recent institutional history makes this harder and more urgent. Scarcity has turned address recognition into an economic control point. Litigation and governance stress have made the control point visible. Market pricing has made old allocations valuable. IPv6 transition has not removed IPv4 dependence. In such a setting, fairness cannot be measured only by whether the same sentence appears in every policy manual. It must be measured by whether capable new networks can enter, route, serve customers, finance growth and survive beyond the first year.
The best registry, for this purpose, is not heroic. It is boring. It records accurately, publishes reliably, reviews proportionately, recognizes clean market movement, supports routing security, limits discretion, explains decisions and keeps working through institutional stress. It treats new entrants neither as supplicants nor as suspects, but as potential operators whose credibility must be built in stages. It understands that conservation is not served by freezing market structure around historical endowments.
AFRINIC's scarcity settlement will shape more than address tables. It will shape who can compete in African connectivity, hosting and edge services during the long period in which IPv4 remains necessary and IPv6 remains incomplete. If neutral rules continue to reward historical stock without acknowledging it, incumbents will keep a silent subsidy. If the registry designs for entry as well as conservation, scarcity will still be painful, but it will not automatically become a moat. That is the real fairness test after exhaustion.