Asia-Pacific IPv4 scarcity is not one shortage, but many shortages on one ledger.
The same 32-bit address space produces very different economic facts in Tokyo, Seoul, Singapore, Sydney, Hong Kong, Taipei, Mumbai, Jakarta, Dhaka, Manila, Port Moresby and Suva. In mature carrier and cloud markets, scarcity is filtered through procurement teams, brokers, escrow, legal review, audit trails, merger planning and the cost of cleaning up routing records. In high-growth access markets, it is filtered through mobile broadband, low average revenue per user, CGNAT, customer support, abuse logging and the pressure to keep IPv4 reachability available while IPv6 deployment remains uneven. In small island and rural networks, it is filtered through thin upstream choice, expensive transit, cable fragility, satellite fallback, small engineering teams and the fact that a few hundred public addresses may matter more to local resilience than a much larger block matters to a global platform.
That is why APNIC is an unusually revealing case in post-exhaustion Internet governance. The factual setting is broad: APNIC describes its service region as 56 economies across Asia and Oceania, with seven National Internet Registries in China, Japan, Korea, Taiwan, Indonesia, Vietnam and India. Those facts do not themselves tell us how scarcity should be governed. They do show why one uniform policy vocabulary has to operate across radically different balance sheets, administrative capacities and growth profiles.
The old shorthand says IPv4 ran out and IPv6 is the answer. Both halves are true, but the shorthand hides the economic problem. APNIC tells members that the maximum IPv4 space a new or existing member can now receive directly from APNIC is a /23, or 512 addresses, and that anyone needing more should consider transfers. It also says IPv6 is the long-term solution. The scarcity economy lives between those two statements. A /23 is not a growth plan for a carrier, cloud platform, hosting provider, security network, data-centre operator or large enterprise. IPv6 is essential, but it is not yet a complete commercial substitute while customers, counterparties, devices, content, enterprise systems, payment systems, security tooling and compliance processes still require IPv4 reachability.
The market therefore does what markets do when a necessary input is finite, unevenly held and operationally embedded. It prices, leases, transfers, warehouses, rations, arbitrages, documents, litigates and discounts. A scarce IPv4 block is not only a line in a registry database. It can support customers, mail reputation, firewall rules, remote-access services, hosting products, cloud workloads, geolocation settings, contracts, route-origin authorisations, reverse DNS and revenue. If the recognised record is clear, the block is more usable. If the record is delayed, contested or subject to unpredictable conditions, the block is worth less.
APNIC did not create the shortage. It inherited a 32-bit protocol limit and a region whose demand grew faster than the old allocation model could sustain. But APNIC now sits at the institutional point where scarcity becomes recognised market capacity. It decides how transfers are recorded, what documentation is adequate, when old delegations remain valid, how RPKI and reverse DNS follow a change, which contacts are authoritative, and how far conservation policy can go before it starts to behave like control over a capital-like resource.
Official APNIC, RIR, ICANN and NRO materials are useful as exhibits: they tell us about pool limits, transfer conditions, fee formulas, Whois obligations, RPKI mechanics, reverse-DNS delegation and the NIR structure. They should not supply the conclusion. The conclusion has to come from the economics of scarcity. Scarcity turns registry procedure into price and bargaining power. A registry that records legitimate movement cheaply makes the market more legible. A registry that decides too much becomes an institutional premium added to every transfer, lease, network expansion and renumbering decision.
The same registry line, different balance sheets
The first analytical error is to treat the APNIC region as one address economy. It is not. The APNIC service map is a coordination surface, not an economic unit. Japan and Singapore do not experience IPv4 scarcity the way Nepal or Samoa do. An Australian enterprise with historical address space does not experience it like a new ISP in Bangladesh. A hyperscale cloud provider does not experience it like a local wireless access provider trying to keep residential users behind CGNAT without breaking games, payment flows, VPNs, home cameras, business servers or government portals.
Several scarcity regimes now sit on the same registry ledger. Mature-market scarcity is a procurement and balance-sheet problem. Operators with capital can buy through brokers, run diligence, obtain warranties, manage escrow, review blocklists, clean up IRR objects, update ROAs, arrange reverse DNS, and spread the legal and administrative cost across larger projects. Scarcity is expensive but financeable. A data-centre operator can treat addresses as part of a deployment budget. A cloud platform can model address cost per workload. A carrier can build transfer planning into mergers and network consolidation. In this regime, the registry's main economic effect is not whether addresses exist; it is whether recognition, documentation and continuity are predictable enough for capital to move.
Growth-market scarcity is different. The central pressure is subscriber scale. Mobile and broadband providers in fast-growing economies may have millions of customers but limited public IPv4. CGNAT extends the life of existing stock, but it is not a free bridge. It adds equipment cost, logging obligations, port-management complexity, abuse-response burden, lawful-access work, troubleshooting cost and customer-experience risk. It can break or degrade applications that assume end-to-end reachability or stable public addresses. A provider can deploy IPv6 aggressively and still need IPv4 for a long residue of services, devices and counterparties that remain IPv4-only or dual-stack in name but IPv4-dependent in practice.
Island-network scarcity is a third regime. In small Pacific economies, address scarcity is mixed with resilience. If a provider has few upstream choices, limited redundancy, expensive transit and a small engineering team, renumbering is not merely an accounting exercise. It can affect schools, clinics, hotels, banks, port operators, local government services and emergency systems. A reverse-DNS delay, stale contact or transfer uncertainty may be minor for a global buyer with a dedicated network engineering team. For a small local network, it can consume scarce staff time and weaken trust with customers who have no practical alternative.
Historical-holder scarcity is another regime. Some organisations received address space in an earlier Internet era, when address administration looked technical rather than financial. Their current need may be smaller than their holdings, or their blocks may be embedded in legacy systems and customer arrangements. They hold optionality: use, reserve, sell, lease, contribute to a corporate transaction, or retain for future uncertainty. That optionality has become bargaining power. The holder may not be hoarding in a moral sense. It is holding an operational asset whose replacement cost has risen.
Buyer scarcity has its own structure. A buyer does not simply buy addresses. It buys recognised transferability, clean authority, routing history, usable reputation, service continuity and confidence that the registry will not introduce surprise conditions after commercial terms are agreed. The price paid to the seller is only the visible cost. The real price includes documentation, time, warranties, broker fees, escrow, legal review, engineering cleanup, blocklist remediation, geolocation correction, fee-tier effects and the risk that some issue in the recognised record delays use.
Lessee scarcity reveals yet another market. Leasing exists because many users need IPv4 reachability now but cannot or do not want to buy. A new hoster may need a /24 faster than a permanent transfer can close. A seasonal service may need temporary capacity. A smaller access provider may prefer monthly operating expense to a large upfront purchase. A platform may want segregated address space for customers or product lines. Leasing is therefore not a curiosity at the edge of policy. It is a price signal from users whose demand does not fit the old allocation model or the capital intensity of permanent transfer.
APNIC's institutional problem is that all these regimes touch the same official record. The region's heterogeneity increases the value of a neutral registry ledger, because market participants need a common record across languages, legal systems, corporate forms and routing environments. It also increases the harm caused by procedural friction. A documentation request that is routine for a multinational can be difficult for a small operator with old corporate records or local-language contracts. A fee denominated in Australian dollars can be stable for one member and volatile for another. A delay that a large buyer can absorb may be the difference between winning and losing customers for a smaller provider.
Scarcity in Asia-Pacific is therefore not merely the absence of unused addresses. It is the uneven distribution of capacity to absorb the institutional costs around scarce addresses. The serious economic question is not whether everyone faces the same text. It is whether the same text imposes very different hidden prices.
The final pool is not a growth policy
APNIC's exhaustion history marks the turn from allocation to recognition. The final /8 policy was implemented in 2009 and allowed account holders to receive up to a /22 from the 103/8 pool. A 2014 policy created a separate recovered pool that could also provide an additional /22. In 2019, APNIC reduced the maximum delegation from the final /8 to a /23. In July 2019, it abolished the waiting list for unmet IPv4 requests from the recovered pool, with future recovered space added back into the remaining pool. APNIC now directs members needing more than a /23 toward transfers.
That sequence changed APNIC's economic function. It is no longer primarily a distributor of new IPv4 supply. It is the institution whose recognition makes scarce IPv4 useful. It records holders, recognises transfers, maintains registration data, supports reverse DNS, provides RPKI services, applies policy restrictions, coordinates across RIR boundaries and keeps the public logs that buyers, sellers, lessors, lessees and network counterparties use for diligence.
Recognition is not a neutral administrative word in a scarce market. A buyer can sign a contract and pay a seller, but if the registry does not update the recognised record, the transaction is incomplete in the market sense. A lessor can promise use, but if the registry record, route objects, abuse contacts, ROAs and reverse-DNS delegations do not align, the customer is taking hidden risk. A holder can believe it controls a valuable block, but if corporate authority, historical status or contact control is unclear, the market discounts the block. A network can route a prefix, but if authoritative records and routing-security artefacts do not support the use, counterparties may hesitate.
This is why old public-resource vocabulary has become more fragile. APNIC policy has historically described address space as a scarce public resource and account holders as custodians rather than owners. It says delegation and registration do not confer ownership, and that globally unique unicast address space is licensed for use. Those statements were coherent in the allocation era. They still matter for uniqueness and stewardship. But the market does not need APNIC to call IPv4 property in order to treat it as capital. A /20 or /16 can be sold through a recognised transfer, leased under contract, pledged in planning, used in corporate negotiations, priced by brokers, valued by holders and impaired by uncertainty.
The point is not to import real-estate law into Internet numbers. The point is simpler: procedure now has asset-market consequences. A 24-month use plan required for a transfer is not merely paperwork. A five-year restriction on 103/8 space is not merely an anti-speculation rule. A membership fee tied to holdings is not merely association finance. A requirement that a source not be in dispute is not merely clerical hygiene. Each condition changes liquidity, bargaining power and price.
That does not make every condition illegitimate. A market for globally unique identifiers cannot rely on seller representations alone. Someone must verify authority. Someone must prevent duplicate claims. Someone must stop forged documents from moving valuable resources. Someone must ensure that records are not changed by a hijacked account. The registry's verification function is valuable precisely because the resource is valuable.
The distinction that matters is verification versus economic discretion. Verification asks whether a fact is true: is the source the recognised holder, is the corporate document authentic, is the recipient eligible, is the prefix subject to a specific policy restriction, are contacts authorised, are associated records updated correctly. Economic discretion asks a broader question: does the institution approve of the buyer's motive, timing, inventory level, business model or commercial strategy. The more APNIC stays in the first category, the more it lowers risk. The more it enters the second, the more it becomes an unseen counterparty in every negotiation.
Need-based allocation language is especially awkward after exhaustion. It is best at seeing past and immediate demand. Scarcity makes the future valuable. A cloud buyer may need addresses before launching a region. A small ISP may need reserve capacity because customers will not sign without IPv4 reachability. A Pacific operator may need contingency space precisely because upstream fragility is a real business risk. A buyer may need inventory before revenue arrives. If procedure recognises only the forms of need that large organisations can document cleanly, the market tilts toward those that already have scale and administrative capacity.
The final pool can preserve a minimum path into the registry system. It cannot be the region's growth policy. Once that is accepted, the core policy question changes from "who deserves newly issued addresses?" to "how cheaply and safely can legitimate address capacity move to productive use?"
When recognition becomes part of the asset
Procedure becomes economic when it affects negotiation. APNIC's transfer rules show the mechanism. A transfer is the movement of number resources from one legal entity to another. The recognised transfer categories include merger or acquisition, historical resources, and unused or excess IPv4 addresses or AS numbers. Participants normally need an APNIC account, supporting information and payment of applicable fees. When the transfer completes, the source no longer has recognised rights to the transferred resources and the recipient becomes the registered holder.
Those are sensible categories. They are also market gates. For transfers of unused or excess IPv4 addresses, including inbound inter-RIR transfers, APNIC asks recipient accounts to provide a detailed plan for use of the transferred resource. Addresses delegated from the 103/8 free pool cannot be transferred for at least five years after original delegation, including in mergers or reorganisations; if the reason for the original request is no longer valid, the resources are expected to return to the registry. For outbound inter-RIR transfers, APNIC says associated records such as sub-assignments, route objects and domain objects will be deleted from the APNIC Whois Database. The recipient may also pay higher annual membership fees after the transfer.
None of this is only administrative detail. It shapes the bid-ask spread. A seller holding a block free of transfer restrictions can command a different price from a seller holding recently delegated 103/8 space that is locked. A buyer whose use plan is easy to document can close faster than one whose use is strategic, contingent or innovative. A cross-border buyer must consider counterpart RIR compatibility. A transfer that requires route-object, reverse-DNS, RPKI and geolocation cleanup carries engineering risk. A block whose source is in dispute is less liquid. A recipient whose fee tier rises after acquisition has a higher total cost than the headline purchase price.
The result is a two-layer price. One price is paid to the holder. The second price is paid in time, documentation, uncertainty, delay risk, operational cleanup and exposure to institutional interpretation. Large buyers can absorb the second price. Smaller buyers often cannot. Liquidity in IPv4 is therefore not only a question of how many unused addresses exist. It is a question of how cheaply recognised control can move from one legitimate holder to another.
This is where registry policy can create or destroy value without ever setting a market price. A narrow verification process reduces uncertainty. It tells buyers and sellers what must be proved, what records will change, what timing to expect and how disputes are handled. A broad discretionary process increases uncertainty. It leaves parties unsure whether a legitimate commercial plan will be treated as acceptable, whether inventory will be suspected, whether leasing-related facts will be viewed as ordinary demand or policy evasion, and whether the closing date is financeable.
Buyers price that uncertainty. Sellers price it too. A seller may prefer a lower bid from a buyer that can close cleanly over a higher bid from a buyer whose documentation or RIR path looks uncertain. Brokers steer deals toward familiar paths. Counsel build contingencies into contracts. Smaller recipients may avoid transfer altogether and instead lease, buy service from an upstream, renumber, overuse CGNAT or defer expansion. The registry did not command those decisions, but its procedure shaped them.
The economic issue is not whether APNIC should ignore fraud or policy restrictions. It should not. It should verify authority, prevent double registration, maintain uniqueness, protect account security and coordinate with other registries. The issue is whether every additional condition has a narrow registry purpose. Does it protect the uniqueness of the address space? Does it protect record accuracy? Does it preserve routing-security continuity? Does it prevent fraud or legal conflict? Does it make the public record more reliable? If the answer is yes, the condition belongs close to the ledger. If the answer is that the institution dislikes the economic use, the condition starts to look like capital allocation.
That line matters because APNIC is not a state, a court, a financial regulator or a telecoms regulator. It is a private membership-based registry performing a coordination function around globally unique identifiers. Its authority is strongest when it is specific. It is weaker when it looks like a licensing power over business strategy.
A scarcity-aware registry should therefore measure the second price it creates. How long do transfer requests take by type? How often are they delayed for documentation? Which documentation problems recur? How often do NIR-related transfers take longer than direct APNIC transfers? How often do inter-RIR transfers fail after commercial terms are agreed? How often do ROA, reverse-DNS, route-object or abuse-contact issues create post-transfer work? How often do small recipients abandon or delay transfers because the process is too costly to navigate? A mature scarcity regime publishes friction, not just rules.
Liquidity is more than address supply
The public APNIC transfer directory is one of the clearest exhibits that IPv4 scarcity is an economic regime rather than a slogan. The APNIC transfer directory contains annual transfer files and a current "latest" file. Its README describes daily summary reports of IPv4 transfers from one organisation to another, required by APNIC transfer, merger, acquisition and takeover policy. It also cautions that the log records information accurate at the time of transfer and does not provide all information related to the transfer.
A July 1, 2026 pull of the current APNIC transfer file showed 13,241 total records, including 10,916 IPv4 records and 2,325 ASN records. The IPv4 records covered about 94.7 million addresses in aggregate. Of those IPv4 records, 2,584 were inter-RIR transfers, with 1,135 inbound to APNIC and 1,449 outbound from APNIC. The earliest transfer date in the file was November 19, 2010; the latest date present in the pulled file was June 30, 2026.
Those figures should not be overread. Transfer logs are not price data. They do not show every lease, every failed negotiation, every private option, every block that could move but does not, every address whose reputation lowers value, or every network that avoided a purchase by buying upstream service. They include different transfer types, including merger and historical-resource movement. A record count is not address volume. One /24 and one /12 are both one record but radically different market events.
Still, the file proves the central point. IPv4 addresses move. APNIC recognises movement. Inter-RIR flows exist. The public ledger maintains an institutional memory of movement. Scarcity is not governed only through final-pool rationing. It is governed through a secondary-market plumbing system whose reliability affects investment, customer growth and bargaining power.
Liquidity has at least five dimensions. Depth is the amount of address space available without moving price too far. Speed is how quickly a deal can close and become operational. Certainty is the likelihood that a legitimate transfer will be recognised as expected. Cleanliness is the absence of hidden liabilities such as stale contacts, blocklist history, disputed authority, broken reverse DNS, invalid ROAs, geolocation errors, customer dependencies, sanctions exposure or unresolved corporate succession. Transparency is the ability to distinguish normal verification from avoidable institutional drag.
APNIC directly controls only some of those dimensions. It cannot create idle supply by decree. It cannot make all historical address use clean. It cannot force sellers to accept prices. It cannot make counterpart RIR processes identical. It cannot make IPv6 instantly eliminate IPv4 demand. But it does affect speed, certainty and transparency. It can make documentation standards clearer. It can make transfer conditions narrower. It can keep logs useful. It can coordinate cutovers with other RIRs. It can make RPKI, reverse DNS and Whois transitions predictable. It can avoid making transfer review a vehicle for general suspicion of secondary markets.
Inter-RIR movement is especially important in Asia-Pacific because the region contains both high-growth demand and historical holdings. A block may move into APNIC space because an operator needs capacity for regional expansion. A block may move out because a holder or corporate group finds a higher-value buyer elsewhere. Neither direction is inherently good or bad. The economic question is whether the recognised path lets addresses move to their highest productive use while preserving uniqueness, contactability and security continuity.
The NIR structure complicates liquidity further. National registries can lower language, support and local-market costs. They can also create practical variation in how regional policy is experienced. A Japanese, Korean, Taiwanese, Indonesian, Vietnamese, Chinese or Indian network may encounter number-resource administration through a national institution. A Pacific island provider may deal directly with APNIC. A cross-border transfer involving an NIR economy may require local and regional alignment. If practical paths differ by economy, the market will price the difference even if the policy vocabulary is regional.
Liquidity therefore cannot be measured only by counting transfers. A region can have many transfers and still impose high fixed costs on small participants. It can have clean transfer rules but weak post-transfer technical guidance. It can have strong logs but little data on delays. It can have national support that helps local members but complicates cross-border counterparties. A useful scarcity dashboard would track not only addresses moved, but median processing time, distribution of delay reasons, inter-RIR completion time, NIR-related timing, object-cleanup issues, dispute frequency and small-recipient outcomes.
The point of such measurement would not be to shame the registry. It would be to reveal where institutional cost is entering the market. In a scarce market, hidden friction is a tax. A registry that can show low, stable, well-explained friction strengthens confidence. A registry that cannot show friction leaves the market to guess, and guessed risk becomes price.
Leasing and shadow allocation reveal unmet demand
Leasing is the market's most uncomfortable evidence because it does not fit the old allocation imagination. The old model assumed a registry, a demonstrated operational need, a direct delegation and a holder using the address space. Leasing separates economic use from permanent recognised transfer. One party may retain the delegation while another party uses the addresses under contract. That can make policy insiders uneasy. It can also reveal demand that the official allocation and transfer system does not satisfy at the right speed, size, duration or capital cost.
The reasons are not mysterious. A hoster may need a small block for a product line and prefer monthly cost to permanent acquisition. A content, VPN, security or monitoring provider may need segregated addresses for customer environments. A startup may need IPv4 before it knows whether demand justifies buying. An access provider may lease while building IPv6 capability, renegotiating upstream service or waiting for a transfer. A corporate group may have internal address space in one entity and operational demand in another. A large historical holder may prefer recurring revenue while retaining strategic optionality.
Leasing can also be a symptom of transfer friction. If permanent transfer is slow, legally heavy, capital-intensive or uncertain, temporary use becomes attractive. If registry policy does not clearly distinguish responsible secondary use from disguised transfer or abandonment, parties may prefer contractual chains that are less visible to the registry. If small buyers cannot finance purchase plus fees plus legal and technical cleanup, leasing becomes the only realistic option. The shadow is not created only by market actors. It is partly produced by the cost of the official path.
This does not mean every lease is benign. Abuse risk is real. A leased block can be used for spam, fraud, phishing, command-and-control infrastructure, mass scanning, proxy services, evasion or short-lived reputational burn. Chains of intermediaries can obscure responsibility. A lessee may disappear before complaints arrive. A lessor may fail to maintain accurate contacts. A route may be technically authorised but operationally opaque. Customers downstream from the lease may suffer if authority or reputation collapses.
The answer, however, is not to pretend leasing demand is illegitimate. Markets do not vanish because policy language dislikes them. They become less transparent. The better registry response is to make responsible secondary use easier to identify. That means accurate holder records, clear operational contacts, usable abuse desks, routing data that matches reality, RPKI coherence, reverse-DNS clarity, and holder accountability for the address space it allows others to use. These are registry-adjacent controls that protect the ledger and the network without trying to regulate every commercial term.
Leasing should also be understood as revealed demand for smaller, faster and more flexible address access. The permanent-transfer market is lumpy. Blocks come in sizes that may not fit the buyer. Closing takes time. Documentation has fixed cost. Reputation cleanup may be uncertain. Leasing offers divisibility and speed. Economically, that is useful. The fact that some leases are risky does not erase the signal. It means the official system should reduce the gap between visible responsible use and opaque chains.
There is a policy trap here. If leasing is treated as inherently suspect, legitimate users are pushed toward less visible arrangements. If leasing is treated as equivalent to ownership transfer, the registry may overreach into private contracts and become a commercial regulator. The better middle ground is narrow transparency: who is responsible for the block, where complaints go, whether routing authorisation is coherent, how abuse is handled, and whether the recognised holder remains accountable. The registry does not need to approve the price, duration, margin or customer segment in order to protect the public record.
Leasing also challenges conservation rhetoric. If an old holder leases unused address space to an operator with real demand, the addresses become more productive than they were while idle. That can be efficient reallocation, even without permanent transfer. If the registry's language makes that movement feel illegitimate, the result may be less conservation in the economic sense. The addresses are conserved in the holder's account but not redeployed to where they produce value.
The scarcity economy will therefore contain both formal transfers and secondary-use arrangements. APNIC's interest should be to keep as much of that activity as possible aligned with accurate records, contactability and routing security. A thin, reliable official path attracts activity into the light. A thick, moralised path increases the premium for staying in the shadows.
The working-capital squeeze below the headline price
IPv4 scarcity is often described as a conflict between rich hoarders and poor users. That is too simple. The hardest burden for smaller operators is not only the market price of addresses. It is the fixed cost of dealing with scarcity at all.
APNIC's fee schedule illustrates the issue. An approved applicant pays a one-off AUD 500 sign-up fee plus an annual membership fee calculated on approved address holdings. APNIC examples show that a new member receiving a /23 and a /48 pays an annual membership fee of AUD 1,709 in 2026, rising in later years under the published schedule; a /23 and a /32 costs AUD 2,256 in 2026. The base fee increased from January 2025, increases again in 2026 and 2027, and then rises 4.75% each January from 2028 unless the Executive Council decides otherwise. Organisations in least-developed countries receive a 50% discount on member fees. Fees are charged in Australian dollars, and payments must arrive free of deductions, set-offs, taxes, levies, bank fees and withholdings.
Those numbers are not dramatic for a large operator. They may be material for a small one. More importantly, the fee is only the visible line. A small provider also pays in staff time, documentation, corporate paperwork, bank charges, currency conversion, legal interpretation, transfer diligence, abuse handling, travel or remote-participation cost, and management attention that could have gone to customers, radios, routers, local peering, fibre drops, backup power or support.
This is the poverty-penalty mechanism in registry form. A fixed institutional overhead becomes regressive when members differ radically in scale. It is not enough to say that everyone faces the same rules. Equal procedure can produce unequal burden. A resource plan, transfer file, renewal issue or reverse-DNS problem that is routine for a staffed compliance department can consume the management capacity of a regional ISP.
Scarcity also changes working capital. If a small operator buys addresses, it ties up cash in a non-depreciating but illiquid operational asset. That cash might otherwise fund capacity upgrades, last-mile equipment, resilience, debt reduction or customer acquisition. If it leases, it accepts recurring expense and counterparty risk. If it deploys more CGNAT, it buys equipment and accepts logging and troubleshooting complexity. If it waits, it may lose customers to a larger provider. If it pushes IPv6 harder, it still has to serve customers and counterparties that require IPv4 reachability. Every path has cost.
The full scarcity budget therefore includes purchase or lease cost, registry fees, broker fees, legal fees, currency cost, transfer documentation, RPKI and reverse-DNS cleanup, Whois and RDAP contacts, IRR updates, geolocation correction, blocklist remediation, CGNAT equipment, CGNAT logging, abuse operations, customer support, working capital tied up in inventory, and management time spent on process rather than growth. The market price of a block is only one line.
For a large buyer, this budget can be part of ordinary infrastructure planning. For a small operator, it can be a gating event. A visible IPv4 price can be financed, negotiated or avoided through technical design. An uncertain registry process is harder to finance because it has no clear ceiling. A bank, investor or owner can understand the price of a /24. It is harder to underwrite "maybe the documents will be accepted, maybe the transfer will close, maybe the technical cutover will be smooth, maybe the fee impact will be manageable".
APNIC's policy tradition contains an underused economic principle: minimised overhead. In the allocation era, that principle meant the administrative burden of obtaining address space should not be excessive. In the scarcity era, it should be revived as a transaction-cost standard. It should not mean pretending abundance still exists. It should mean that the official recognition path is cheap to understand, cheap to navigate and cheap to trust, especially for operators with limited administrative capacity.
The hardest cases are island, rural and low-margin access networks. A small provider may need public IPv4 not because it is inefficient, but because its customers require stable services. A hotel network, school, clinic, port authority, local ecommerce company, government office or radio ISP may not express demand in the procurement language of a large carrier. Yet loss of address continuity can have wider local effects. Small networks often carry more local resilience per address than their size suggests.
That argues for plain-language transfer guidance, small-recipient documentation templates, predictable cure periods, local-language support through NIRs where relevant, transparent fee explanations, published friction metrics and transfer-continuity checklists. It does not argue for suspending scarcity. It argues for reducing the institutional premium that falls hardest on operators least able to absorb it.
IPv6 changes the ceiling, not today's constraint
IPv6 is the long-run addressing architecture the Internet needs. APNIC's training, measurement, policy and deployment support around IPv6 are useful. A region with billions of users, mobile devices, sensors, public services, cloud workloads and enterprise systems cannot build indefinitely on IPv4 plus translation. The address space is too small, the workarounds too costly and the operational compromises too persistent.
But IPv6 is not a full economic substitute for IPv4 today. It is a complement during a long, uneven transition. That distinction is the foundation of the scarcity economy.
A network cannot replace IPv4 with IPv6 in isolation. Its users must reach IPv4-only services. Its customers may run devices, payment systems, security appliances, cameras, industrial systems, VPNs or software that assume IPv4. Enterprise counterparties may still require IPv4 allowlists. Abuse, lawful-access, logging and support processes may be built around IPv4. Inbound services may need public IPv4 because customers expect it or because intermediaries do. Even when access networks are heavily IPv6-capable, IPv4 remains commercially necessary for reachability.
The APNIC region shows the unevenness clearly. India has seen large-scale IPv6 deployment by major access networks. Japan, Korea, Taiwan, Singapore and Australia have significant but varied deployment across access, enterprise, hosting and government environments. China has pursued state-backed IPv6 goals, but user availability, service support and traffic share are not the same thing. Smaller economies may still depend heavily on IPv4-centric upstream services and customer equipment. A mobile network may carry substantial IPv6 traffic internally and still spend heavily to maintain IPv4 through translation. A data-centre provider may advertise IPv6 and still lose sales if it cannot supply enough clean IPv4.
The phrase "IPv6 transition" can therefore mislead. It suggests a bridge with a visible far end. Operators experience something closer to a dual-stack tax: two address families, two routing and filtering surfaces, two monitoring patterns, two security postures, two customer-support narratives and a long period in which the old scarce input remains commercially necessary. Some operators can reduce IPv4 dependence faster than others. Few can ignore it entirely.
CGNAT is the bridge many access providers pay for. It economises on public addresses by sharing them among many users. But it converts address scarcity into other costs. Port exhaustion becomes a customer problem. Abuse attribution becomes harder. Logs become larger and more sensitive. Lawful requests require more care. Troubleshooting becomes less direct. Applications can fail or degrade. In large mobile networks, these costs may be acceptable because scale demands them. In smaller networks, the equipment and support burden can be a material share of the business.
Transfers and leases are therefore not proof that operators have ignored IPv6. They are proof that IPv6 deployment does not eliminate the need for IPv4 continuity during the transition. A rational operator may deploy IPv6, buy IPv4, lease IPv4 and run CGNAT at the same time. Those are not contradictory strategies. They are a portfolio response to incomplete substitutability.
The policy danger is using IPv6 as a rhetorical escape from IPv4 economics. "Deploy IPv6" is correct as strategic advice and inadequate as an answer to a provider that must serve IPv4-dependent customers this quarter. If a registry, policymaker or incumbent treats IPv4 demand as moral failure because IPv6 exists, the burden falls on operators with the least power over customers, devices and counterparties. The better institutional stance is dual: push IPv6 hard, and make the remaining IPv4 market cleaner, safer and less discretionary while demand persists.
IPv6 changes the long-term ceiling. It does not remove today's constraint. Any APNIC scarcity policy that forgets that distinction will misprice the cost borne by operators in the middle of the transition.
Legacy blocks, paperwork and bargaining power
Legacy and historical holdings are where the economics of scarcity meet the archaeology of the early Internet. APNIC policy recognises historical resources and permits transfer of historical IPv4 resources under defined conditions. Such space can be attractive because it may include larger or older blocks. It can also carry documentation problems that did not matter much when addresses were technical necessities rather than valuable market assets.
Corporate names change. Universities restructure. State agencies merge. Telecom assets are spun out. Data-centre units are sold. Contacts become stale. The original technical manager retires. Old letters are missing. A subsidiary that used the space no longer exists in the same form. The block may be routed by one network, listed under another name, controlled through a portal by a third team and valued by a fourth. Scarcity turns those messy facts into economic risk.
A clean historical block can move more easily. A poorly documented block may be discounted, delayed or stuck. Legal and documentation friction therefore become part of scarcity. This is not a small issue in Asia-Pacific, where corporate law, language, scripts, public registries, national-security review, capital controls, sanctions exposure, foreign-investment rules and insolvency processes differ widely across economies. The cost of proving authority is not evenly distributed.
Large holders and repeat buyers can hire specialists. Smaller holders may not know their records are stale until a transfer, dispute, acquisition or renewal event forces the issue. A buyer may want a discount for unclear authority. A seller may be unable to monetise a legitimate holding because old documents cannot be produced in the expected form. A network may keep routing space operationally because customers depend on it, while the formal record lags behind corporate reality. The market calls this a title problem even if registry policy avoids property language.
APNIC's role should be narrow but serious. It should verify authority without demanding impossible historical perfection. It should distinguish fraud from imperfect documentation. It should provide defined paths for corporate succession, name changes, mergers, historical-resource claims and stale-contact repair. It should record uncertainty where necessary without turning every uncertainty into paralysis. It should isolate disputes where possible while preserving operational continuity for networks and customers not responsible for the paperwork problem.
The power to review delegations becomes more sensitive under scarcity. Policy language that sounded like ordinary stewardship in the abundance era can affect valuable operational capital in the scarcity era. That does not mean APNIC should never review records. It means review should be bounded, evidence-based, auditable and proportionate. A registry that can affect market value by questioning recognition owes the market clear evidentiary standards.
Legacy holdings also complicate the politics of conservation. Some old holders genuinely underuse space. Some hold reserve capacity for systems that are expensive to renumber. Some can sell but choose not to. Some lease. Some are unaware of market value. Some have internal governance that makes sale difficult. Treating all of these cases as moral hoarding is analytically weak. Treating them all as untouchable property is equally weak. The useful question is whether the recognised path makes voluntary, clean reallocation easier than quiet inactivity.
If a historical holder can prove legitimate control, recognition and transfer should be predictable. If a claim is disputed, the dispute should be marked or isolated. If a block is abandoned, recovery can occur under policy. What should be avoided is using uncertainty as discretionary leverage. In a scarce market, unclear evidence standards become bargaining power in the hands of the institution that controls recognition.
RPKI, reverse DNS and reputation are economic infrastructure
IPv4 scarcity increases the economic value of technical continuity. A prefix is not useful in isolation. It needs a public record, operational contacts, routing-security support, reverse DNS, usable reputation and confidence that the recognised holder can authorise changes. Whois and RDAP data, route objects, abuse contacts, ROAs and reverse delegations are not side details. They are part of what buyers, lessees, networks, mail systems, security teams and customers rely on when they decide whether a block is usable.
APNIC's Whois materials describe a database that stores information about IP address ranges, routing policies, reverse DNS delegations and network contact information. Its RPKI materials describe a framework that binds Internet number resources to custodians through certificates, route-origin authorisations and validators. ROAs specify which AS is authorised to originate a route for a prefix, while validators classify routes as valid, invalid or not found. APNIC offers hosted RPKI through MyAPNIC and self-hosted options. Those are factual services, but scarcity gives them financial weight.
Reverse DNS adds another continuity layer. APNIC describes reverse DNS for RIR-delegated address space as a delegation chain through RIR servers to name servers provided by the network or end party. Reverse delegations for IPv4 are based on /8, /16 and /24 reverse zones, and APNIC generates reverse zones from the Whois Database every two hours before normal DNS propagation follows. That timing and structure matter for mail, logging, monitoring, abuse handling, customer systems and reputation.
In a transfer, all of this has to move, be rebuilt or be cleaned up. A buyer may need new ROAs, updated route objects, new reverse DNS, revised abuse contacts, geolocation corrections, blocklist remediation and customer migration. A seller may need to remove old objects and avoid leaving misleading security artefacts behind. For outbound inter-RIR transfers, APNIC says associated records such as sub-assignments, route objects and domain objects will be deleted from the APNIC Whois Database. That is a serious operational event. If it is not coordinated, it can affect routing filters, mail deliverability, monitoring systems and customer services.
This is why the registry is a continuity surface as well as a record keeper. That does not justify broad economic discretion. It justifies service-boundary discipline. APNIC should be able to correct a forged transfer, lock a hijacked account, prevent misleading ROAs, keep abuse contacts accurate, maintain reliable repositories and delegate reverse DNS correctly. Those are ledger-protecting functions. They make the market safer.
What APNIC should avoid is turning operational services into leverage for unrelated disputes. If a fee, documentation or policy disagreement exists, remedies should be proportionate and should preserve running-network and customer continuity where law and security permit. A valid ROA, reverse delegation or Whois record should not become a casual bargaining chip. Scarce address resources support customers downstream from the member account, and disruption can punish people who are not parties to the dispute.
The market already prices technical continuity. A block with clean RPKI, accurate reverse DNS, responsive abuse contacts, stable routing history, usable geolocation and a predictable transfer path is worth more than a block surrounded by stale records and unclear authority. APNIC can create value by lowering the cost of certainty. It destroys value when the continuity surface is hard to price.
A practical improvement would be a public transfer-continuity checklist that goes beyond legal procedure. It should cover ROA withdrawal and creation, maxLength review, IRR and route-object handling, reverse-DNS timing, abuse-contact transition, geolocation expectations, blocklist review, object deletion in inter-RIR transfers and recommended cutover sequencing. This would not be market interference. It would be the registry doing what a good scarcity ledger should do: making recognised movement safer.
Conservation after exhaustion should mean movement
Conservation was one of the founding virtues of address policy. It still matters. IPv4 is finite. Waste has social cost. Fraud and speculation can damage trust. Routing-table growth matters. The problem is that conservation language changes meaning after exhaustion.
When a registry allocates from a free pool, conservation means not giving out more than justified. When a market reallocates already-delegated resources, conservation should also mean making underused resources move safely to productive use. If the first meaning overwhelms the second, conservation becomes anti-liquidity. A block sitting unused at an old holder may be conserved in the narrow administrative sense, but it is economically idle. A buyer with real demand may be delayed because its future use does not fit a documentation template. A lessor may avoid transparent arrangements because secondary use sounds suspect. The result is not conservation. It is deadweight loss.
There is a moral temptation in IPv4 scarcity debates. It is easy to describe markets as hoarding and registries as stewardship. Sometimes that captures real behaviour. It is also easy to treat price itself as evidence of moral failure. That is wrong. Price is the signal that scarcity has economic value. It tells holders to consider whether idle space could be redeployed. It tells buyers to conserve internally. It tells lessors to make capacity available. It tells operators that IPv4 is no longer free growth fuel.
The better question is not whether price should exist. It already does. The question is whether the official system makes price discovery cleaner or dirtier. Cleaner means accurate records, reliable transfers, transparent logs, narrow anti-fraud checks, usable abuse contacts, routing-security continuity and minimal unnecessary delay. Dirtier means opaque discretion, unclear documentation, hostility to leasing, moralised debate and uncertainty about whether a legitimate buyer will be recognised.
Fairness also needs a post-exhaustion definition. It cannot mean that every operator receives the address space it wants from a pool that no longer exists. It cannot mean that every market outcome is equal. Scarcity will reward some combination of capital, timing, historical accident, operational efficiency and bargaining position. Fairness in a scarce registry should mean that the recognised path is predictable, evidence-based, non-discriminatory and affordable to navigate. It should not mean that the registry tries to simulate equality by making movement harder.
This matters most for operators with limited capital. A visible market price can be harsh, but it can be financed, negotiated, leased around or reduced through technical design. A discretionary process is harder to finance because it creates uncertainty about closing, timing and future control. For a small operator, uncertainty may be more damaging than price because it blocks planning.
Conservation should therefore become a ledger objective rather than a moral licence. Prevent duplicate registration. Prevent fraud. Prevent fake need claims in direct delegations. Enforce specific final-pool restrictions. Keep contacts accurate. Protect routing security. But once addresses are legitimately held, make efficient reallocation under clear rules easier than opacity. The registry's comparative advantage is recognition accuracy, not economic planning.
This is the practical meaning of remaining a ledger rather than becoming a capital controller. The registry should decide whether records are accurate, whether documents prove authority, whether a block is subject to a specific policy restriction, whether a ROA is authorised, whether a reverse-DNS delegation is valid, whether an abuse contact exists, whether a transfer is compatible with counterpart RIR policy and whether a dispute must be marked. It should not decide that a business model is illegitimate merely because it includes leasing, that price discovery is bad because addresses were once cheaper, or that IPv6 aspirations make current IPv4 demand unreal.
The official path should be cheaper than the workaround. If recognised transfer and responsible secondary use are predictable, parties have reason to keep activity aligned with the ledger. If the official path is slow, moralised or uncertain, parties rely more heavily on private contracts, intermediaries, acquisitions, informal assurances and routing workarounds. Excessive control does not eliminate market activity. It moves risk outside the visible record.
What to watch as scarcity hardens
The next stage of Asia-Pacific IPv4 scarcity will not look like a single crisis. It will look like segmentation. Some networks will reduce IPv4 dependence quickly. Others will keep paying for translation, transfers and leases. Some historical holders will sell. Others will lease or retain strategic reserve. Some economies will experience scarcity mainly through national registries. Others will depend directly on APNIC. Some small operators will treat every /24 as survival capacity. Some large buyers will treat addresses as infrastructure capital.
The first watchpoint is transfer friction. The APNIC transfer log is useful, but the market needs more aggregate process data: median processing time, common delay reasons, approval and withdrawal patterns, inter-RIR timing, NIR-related timing, dispute frequency, object-cleanup issues and post-transfer technical incidents. The goal is not to publish confidential deal information. It is to let members distinguish careful verification from avoidable drag.
The second watchpoint is leasing transparency. Leasing demand will not disappear. APNIC can either help responsible secondary use become more visible through accurate contacts, abuse accountability, route-object hygiene and RPKI coherence, or it can leave more of the market in opaque contractual chains. The better path is narrow operational transparency without trying to regulate every commercial term.
The third watchpoint is small-operator cost. Fee discounts for least-developed countries address only one piece of the burden. The larger question is whether APNIC reduces documentation overhead, supports low-capacity markets, offers plain-language transfer guides, provides usable templates for small recipients and measures participation from economies that rarely appear in policy debates. Scarcity policy that works only for sophisticated repeat buyers is not economically neutral.
The fourth watchpoint is NIR alignment. The seven NIRs localise service, but they also create varied practical paths. Watch how regional transfer policy works through national procedures, local fee schedules, local-language support and cross-border transactions. Liquidity will be shaped not only by APNIC policy text, but by the practical cost of moving resources through the relevant institutional channel.
The fifth watchpoint is IPv6 realism. APNIC should continue to push IPv6 deployment aggressively. It should also avoid using IPv6 as a substitute for clear IPv4 scarcity governance. Dual-stack cost, CGNAT burden and uneven customer reachability will keep IPv4 economically relevant for years. The better IPv6 story is not that IPv4 markets no longer matter. It is that cleaner IPv4 governance reduces the cost of transition while networks reduce dependence over time.
The sixth watchpoint is legacy cleanup. Historical holdings will keep appearing in transfers, leases, mergers and disputes. The market needs clear evidentiary paths for corporate succession, name changes, stale contacts and old resource records. Fraud prevention is essential. Impossible paperwork standards are not. The difference matters because documentation burden becomes price.
The seventh watchpoint is technical continuity. RPKI, reverse DNS, Whois, RDAP, IRR data, abuse contacts, geolocation and blocklist history should be treated as economic infrastructure, not administrative afterthoughts. Transfer guidance should make cutovers predictable. Disputes should be isolated where possible rather than converted into avoidable customer disruption.
The eighth watchpoint is the fee and currency layer. Scarcity is not only address price. It is also membership cost, transfer cost, bank cost, currency exposure and the cost of complying with formal process. In a region with very different incomes and operator sizes, APNIC should be able to explain how its fees support the essential ledger and how it avoids placing disproportionate institutional cost on smaller networks.
The ninth watchpoint is policy language. If APNIC debates rely on words such as public resource, stewardship and community without translating them into measurable costs and boundaries, scepticism will grow. If debates focus instead on evidence standards, continuity, friction metrics, transfer safety, small-operator burden and the distinction between verification and economic discretion, APNIC will sound more like a market-enabling ledger and less like an allocator trying to govern a market that already exists.
The final watchpoint is institutional modesty. Scarcity increases the temptation to govern. It makes the record more valuable, the policy room more consequential and the registry more visible. But the correct response is not to expand authority into every economic choice. It is to constrain authority to what the registry can do uniquely well: protect uniqueness, validate recognition, maintain accurate records, preserve technical continuity, reduce documentation friction and keep the official path more reliable than the workaround.
APNIC's opportunity is to make Asia-Pacific IPv4 scarcity legible without pretending it can be morally administered away. The region does not need a registry that suppresses price signals, treats every secondary use as suspect, or taxes the weakest operators through opaque process. It needs a registry whose procedures are cheaper than uncertainty. In a market where IPv4 is scarce, valuable and operationally embedded, that is not a small role. It is the role that matters.