Summary

  • Public company disclosures show at least four materially different outcomes for IPv4 portfolios: nil-cost intangibles that create gains on sale, operating intangibles separated from sale inventory, acquired indefinite-lived intangibles, and acquired finite-lived intangibles amortised over 17.5 years.
  • These outcomes do not establish a global error rate or a single correct label. They arise from different acquisition histories, intended uses, accounting frameworks, useful-life judgments and, above all, different evidence about what the holder actually controls.
  • A registry entry is important evidence, but it is not a complete accounting conclusion. Registration does not by itself prove legal title, unrestricted transferability, routability, a reliable useful life, an active market or the cash flows attributable to a particular block.
  • The accounting question should be stated as a bundle of rights and dependencies: who is recognised, which prefixes are covered, under what agreement, with what transfer conditions, for what use, and with which operational services and third-party constraints.
  • Number Resource Society can make that bundle easier to audit through portable, time-stamped and independently verifiable registration facts. It should not prescribe whether an address portfolio is inventory, an intangible asset, indefinite-lived, impaired or absent from the balance sheet.

Scarcity created value before accounting created a vocabulary

The modern IPv4 market began with a transaction that looked simple enough to put in a newspaper headline. In 2011, during Nortel's bankruptcy, Microsoft agreed to pay about $7.5 million for 666,624 legacy IPv4 numbers. The price made scarcity legible. Addresses that had once arrived as an administrative allocation were now being exchanged for cash in a court-supervised sale.

The accounting consequences were less simple. What had Nortel sold? A set of numbers cannot be possessed like a warehouse. The ability to announce routes is not exclusive in the physical sense: another network can originate the same prefix, although the result may be conflict or hijacking rather than lawful use. The registry record matters, but the relevant registry did not describe the numbers as freely held property. The buyer also needed a registration-services relationship and a recognised transfer. The economic asset therefore lay in a combination of registration, contractual standing, practical control, scarcity and market acceptance.

That combination has grown more valuable and more varied since 2011. Companies use addresses in their own networks. They transfer surplus blocks. They lease address capacity to customers. They acquire portfolios in business combinations. They place some portfolios in special-purpose financing structures. Each activity can produce a different accounting unit and a different cash-flow pattern even when the visible identifier is an IPv4 prefix.

The result is a balance-sheet asset with no settled everyday name. Public reports use terms such as IPv4 addresses, IP addresses, rights to use addresses, IPv4 address assets and internet number resources. Registry documents speak of allocations, assignments, registrations, licences, registration rights and holdership. Lawyers may describe a bundle of contractual rights. Network engineers may speak as if the block itself is controlled. Markets quote a price per address, compressing all those distinctions into one denominator.

None of these vocabularies is wholly false. None is sufficient for every purpose. The accounting inconsistency begins where a convenient noun conceals the composition of the right.

Four filings, four defensible-looking answers

The clearest evidence is not a theoretical debate but the decisions companies have already disclosed.

RM plc announced in October 2023 that it had agreed to sell another portion of its IPv4 addresses for $2.2 million in cash. It said the addresses were classified as intangible assets and had been acquired at nil value. That sentence captures one of the defining accounting effects of historical allocation. A resource can have a zero carrying amount and a material sale price. Recognition at nil value does not mean no economic value; it can mean the company did not incur an acquisition cost that became a recognised carrying amount.

Tessellis Group's 2023 annual report drew a different line inside one acquired portfolio. In its purchase-price allocation, 524,000 addresses described as essential for business operations were classified among other intangible assets and valued at EUR18 million at the acquisition date. Another 762,000 addresses not directly necessary for ordinary operations were classified as assets available for sale among inventories and valued at EUR26.2 million. The identifiers belonged to one technical class, but management's purpose divided the accounting classes.

Cogent Communications recorded $458 million for acquired IPv4 addresses after acquiring Sprint's wireline business. Its public filings describe the portfolio as an indefinite-lived intangible asset that is not amortised. The valuation used market information, and the indefinite-life conclusion means management did not identify a foreseeable limit to the period over which the asset was expected to contribute cash flows. It does not mean the addresses are immortal, legally unconditional or immune from impairment.

Uniti Group reported another acquired portfolio as a finite-lived intangible asset amortised on a straight-line basis over 17.5 years. Its 2025 filing also disclosed that certain unused addresses with a net carrying value of $6.7 million were sold for $5.0 million. Here the portfolio had a declining accounting life, a carrying amount and a disposal outcome below that amount.

These examples are not a statistical sample of all holders. They cannot establish how frequently any treatment occurs. Private companies disclose less; many operators never separately identify addresses; reporting rules and transaction facts differ. But the examples are enough to disprove the idea that practice has converged on one self-evident classification.

The balance sheet remembers how the right arrived

Two companies can control similar address blocks on the reporting date and show radically different carrying amounts because accounting remembers acquisition history.

A company that received addresses decades ago may have paid registration fees and network costs but no separately identifiable purchase price. Those expenditures may have been expensed when incurred, embedded in broader operating costs or attached to a service arrangement rather than capitalised as the cost of a distinct asset. The portfolio can consequently remain absent or sit at a nominal amount even after a secondary market demonstrates material exchange value.

A company that buys the same quantity in a standalone transaction has a directly observable cost. Subject to the applicable recognition rules, the purchase makes identifiability and measurement easier. The transaction itself is evidence that the right can be separated and exchanged. Legal, brokerage and transfer costs may also be attributable. The acquired asset enters the books even though an economically similar historical portfolio did not.

A business combination creates another route. Purchase accounting asks the acquirer to identify and measure assets acquired at the acquisition date. A portfolio that was unrecognised in the seller's own statements may become a separately recognised intangible for the buyer. Cogent's acquired addresses are an example of a large fair-value recognition emerging from a broader acquisition rather than from the seller's previous carrying amount.

Inventory introduces intended use. If a company acquires or holds addresses for sale in the ordinary course of business, the portfolio may be analysed differently from addresses used to provide connectivity. Tessellis's split makes the distinction unusually visible: operating need supported an intangible classification for one part, while an available-for-sale purpose supported inventory for another.

The balance sheet therefore does not provide a census of economic IPv4 wealth. It records recognised rights under particular standards, histories and purposes. Adding the reported values of a few companies would not produce a meaningful global market capitalisation. The denominator of all controlled, transferable or economically useful addresses is unavailable, and the unrecognised historical base may be substantial but cannot responsibly be valued by applying one public transaction price to every address.

IAS 38 asks a question about control, not a question about nouns

IAS 38 defines an intangible asset as an identifiable non-monetary asset without physical substance. Identifiability can arise because the asset is separable or because it arises from contractual or other legal rights. Recognition also requires probable future economic benefits and reliable measurement of cost. Control means the entity can obtain the benefits and restrict others' access to them.

IPv4 rights can fit that architecture, but the fit depends on facts.

Separability is supported when a portfolio can be transferred independently or together with a related agreement. A completed market transaction is strong evidence. Yet separability is not identical across all prefixes. Legacy resources may have different contractual histories. Provider-dependent assignments may not be portable. A transfer lock may apply. A registry may require the recipient to meet conditions. More-specific records may complicate division. The fact that one block was sold does not prove every registration can be sold in the same way.

Contractual rights can also support identifiability. ARIN's current Registration Services Agreement expressly refers to registration rights for included number resources and grants a defined set of service-related rights. APNIC describes delegated resources through a licence to use under its documents. RIPE NCC agreements and policies describe registration, allocation, assignment and transfer relationships. Those instruments provide substance, but not one uniform global instrument.

Control is equally fact-sensitive. A holder may be recognised in a registry, maintain route-origin authorisations, operate reverse DNS, control credentials and exclude lawful counterparties from using the registration. That is meaningful control. It remains different from the physical exclusion associated with land. Routing can be contested. Registry services can be suspended. A contract can be terminated. A court or insolvency officeholder can change authority. The accounting conclusion should reflect the enforceable and operational package rather than a metaphysical argument about whether numbers can be owned.

Future benefits may arise through service revenue, avoided address-acquisition cost, customer retention, lease income or sale proceeds. Each pattern implies a different unit of account and impairment evidence. A company should not invoke scarcity alone when it cannot connect the portfolio to benefits it can obtain and defend.

The registry vocabulary is part of the evidence and part of the problem

RIR language is deliberately cautious because registries coordinate unique identifiers for a shared network. ARIN said in 2022 that internet number resources constitute a bundle of contractual rights created when resources are issued, while maintaining that they are not freely held property. Its current policy says resources are not sold under ARIN administration and that transfers require express written approval under policy. At the same time, its agreement and fee schedule use the practical language of transferring registration rights.

APNIC's current policy materials say it is contrary to the goals of the system to treat internet number resources as freehold property. They describe a licence subject to membership and policy conditions. RIPE NCC transfer policy speaks of a legitimate resource holder and requires transfers to be reflected in the RIPE Database; completion occurs when the registration records are updated.

These are not cosmetic differences. They affect the evidence available for identifiability, control, useful life and disposal.

A right described as a revocable licence may still be an accounting asset. Many licences are. A contractual restriction may still leave a separable right if transfers routinely occur subject to consent. An absence of freehold title does not automatically prevent recognition. Conversely, a market price does not eliminate contractual conditions. Accounting should neither accept the registry's anti-property language as a command to recognise nothing nor convert transfer practice into absolute ownership.

The institutional difficulty is that the vocabulary varies across regions and resource histories. "Legacy" can describe an allocation made before the relevant RIR existed, but it does not itself specify the current contract. "Holder" may identify the entity in a database without describing all legal interests. "Transfer" can refer to a commercial sale, a merger, a legal-name change, a registry update or a movement between service regions. "Address asset" can include the registrations, customer leases, receivables and related operational rights.

An auditor cannot resolve those differences by choosing the strongest-sounding noun. The answer requires the underlying documents and current state.

Intangible asset is a beginning, not a complete classification

Calling IPv4 rights an intangible asset can be directionally sensible while leaving the most consequential questions unanswered.

What is the unit of account? A /16 may be held as one operational block but sold in smaller prefixes. A portfolio may contain blocks under different agreements, with different routing reputations and transfer histories. A company may control a registry account while subsidiaries operate individual networks. Treating every address as a fungible unit can hide legal and operational constraints that attach at prefix or contract level.

What is included in cost? A standalone buyer may pay a seller, a broker, legal advisers and registry fees. It may incur remediation cost to make a block usable because historical routing or abuse reputation limits acceptance. Some of those expenditures may be directly attributable; others may be operating costs. A one-number market quote cannot answer the allocation.

What benefits does the asset support? Addresses used inside a broadband network may support a cash-generating unit rather than independent revenue. A leased portfolio may generate identifiable contractual cash flows. Addresses held for disposal may be evaluated through expected sale proceeds. One classification across all three uses can blur the evidence needed for impairment.

What is the residual value? A finite-lived asset might retain a sale value after operational use. That value depends on transferability, block size, reputation, registry status and future IPv6 substitution. It is not safely inferred from a headline price.

What disclosures are material? A large recognised portfolio may expose investors to valuation assumptions, transfer restrictions, concentration in one registry relationship and changes in market liquidity. An unrecognised historical portfolio may create future gains on sale that make operating performance difficult to compare. The label "other intangibles" can be technically permitted yet economically uninformative when the amount or disposal is material.

The useful accounting debate begins after the label, not before it.

Inventory depends on business purpose, not on the speed of a sale

Tessellis's classification of some acquired addresses among inventories raises a particularly important distinction. Inventory is not merely an intangible asset that management hopes to sell. It generally reflects an asset held for sale in the ordinary course of business, in production for such sale, or as materials or supplies to be consumed. The company's disclosed purchase-price allocation associated the classification with addresses not directly necessary for ordinary operations and available for sale.

That fact pattern should not become a universal rule. A telecommunications operator disposing of a one-time surplus does not automatically become an address dealer. An asset can remain an intangible while being sold. Classification depends on the company's business model and the role of the portfolio, not on management discovering a buyer.

The distinction matters because measurement and presentation differ. Inventory is generally subject to cost and net-realisable-value analysis under IFRS. A finite-lived intangible is amortised and tested for impairment when indicators arise. An indefinite-lived intangible is not amortised but faces recurring impairment testing. Assets meeting held-for-sale criteria under another standard may have yet another presentation. Revenue and gain presentation can also differ depending on whether address sales are ordinary output or disposals of long-lived assets.

The public evidence does not reveal how many companies hold addresses as inventory, how many are ordinary traders, or how consistently auditors apply the boundary. There is no reliable global denominator. One Italian group's disclosed split demonstrates that purpose can divide a portfolio; it does not establish an industry convention.

For governance, the lesson is that registry data should not encode an accounting category. A registry can record that a block is registered to an entity, subject to an agreement, transferable under stated conditions and currently associated with particular operational services. It usually cannot know whether management holds that block for internal use, customer leasing, strategic reserve or ordinary-course sale. Requiring the registry to label "inventory" would substitute incomplete administrative knowledge for the reporting entity's business facts.

Indefinite does not mean permanent

The contrast between Cogent and Uniti is most striking in useful life. One public company treats acquired IPv4 addresses as indefinite-lived; another amortises them over 17.5 years.

An indefinite useful life is an accounting judgment that no foreseeable limit has been identified for the period over which the asset is expected to generate net cash inflows. It is not an assertion of perpetual legal ownership. The asset remains subject to annual impairment testing and reconsideration of whether indefinite classification still holds. A change in technology, policy, market depth, routing practice or legal enforceability could shorten the expected period.

A finite life likewise does not predict the date on which the last IPv4 packet will travel. Management may estimate the period over which the acquired portfolio contributes to the specific business, considering expected IPv6 migration, customer contracts, network strategy, regulatory conditions and disposal plans. Straight-line amortisation is an allocation of cost; it is not a physical decay curve.

Both judgments can be reasonable on different facts. A lessor with recurring address revenue, active market evidence and no contractual expiry may see no foreseeable limit. An integrated operator may expect a finite transition or may align the asset with a forecast period supported by acquisition valuation. Different US GAAP reporters can therefore reach different conclusions without one necessarily misapplying the standard.

But the difference should be visible and challengeable. If an address portfolio is material, investors should understand why its useful life is finite or indefinite, which factors could change the estimate, how market prices were used and how registry conditions enter the analysis. A generic note about "other intangible assets" does not communicate the sensitivity.

The registry's contribution is again factual. It can verify whether the registration has a stated contractual term, renewal conditions, transfer locks, policy dependencies or known service discontinuities. It should not translate those facts into a 17.5-year life. The life belongs to the cash-flow context of the company, not to the database administrator.

Market price is observable, but the asset is not homogeneous

IPv4 markets tempt analysts to multiply a quoted price by an address count. That arithmetic is useful for a rough scenario and dangerous as an accounting conclusion.

Block size affects marketability. A large aggregate can serve a major network and preserve routing efficiency, but it narrows the buyer set and may need to be divided. Small prefixes may sell to more buyers but carry higher transaction and administration cost per address. Regional eligibility and inter-RIR compatibility matter. Transfer locks matter. Legacy status and contractual coverage matter.

Reputation matters too. Addresses associated with spam, fraud or previous abuse can encounter filtering and remediation cost. Registry accuracy, reverse-DNS readiness and routing-security controls affect deployment. A clean registration with complete corporate evidence is not economically identical to a disputed block whose historical holder no longer exists.

Use constraints matter. An operator can derive value by avoiding carrier-grade network address translation, preserving customer configurations, supporting legacy software or serving customers that require public addresses. A lessor values expected lease cash flows and utilisation. A trader values sale liquidity. A distressed seller may accept a price that does not represent orderly market conditions.

Observable transfers therefore provide important but incomplete evidence. Cogent's valuation materials referred to a market approach using recent transactions. That is a sensible starting point for an asset with market activity. It still requires adjustments for the characteristics and highest-value use of the specific portfolio. Tessellis's separate operating and sale groups show why a single blended price can obscure purpose. Uniti's disclosed disposal below carrying value shows that a recognised unit value does not guarantee recovery on sale.

No public source used here provides a complete global transaction set with consistent terms, block quality, rights, fees and failed deals. Published transfer lists show approved changes, not necessarily price. Broker indices may omit private terms and unsuccessful offerings. The article therefore does not estimate a worldwide average price, total unrecognised value or frequency of impairment.

Impairment begins where registry facts and business facts diverge

An IPv4 portfolio can lose recoverable value without disappearing from the registry. That is why impairment analysis needs more than a current holder name.

An operating portfolio may face technological substitution. Faster IPv6 adoption in the company's customer base can reduce the avoided-cost benefit of IPv4. Network redesign can strand blocks. Customer migration can shrink demand. A change in abuse reputation can make deployment expensive even though registration remains accurate.

A sale portfolio may face liquidity deterioration, transfer restrictions, fragmentation cost or a smaller buyer pool. A lease portfolio may face falling utilisation, customer concentration, defaults, price pressure or enforcement difficulty. A disputed chain of authority can increase time and legal cost. Loss of registry credentials or a policy breach can weaken practical control.

Positive evidence also matters. Continued lease collections, successful transfers of comparable clean blocks, stable registration rights and sustained operational need may support value. A temporary decline in quoted prices does not automatically erase a long-term benefit. Conversely, scarcity rhetoric does not cure a portfolio that cannot lawfully or operationally be used.

The impairment unit must follow how cash flows are generated. Addresses embedded in a network may not generate cash independently and may be tested with a broader asset group or cash-generating unit. A separately leased portfolio may support a narrower test. Blocks held for sale may have directly observable disposal evidence. Management should reconcile the prefix inventory used in valuation to the authoritative registration state so that duplicated, ineligible or operationally committed space is not counted twice.

Auditors need change evidence. Which prefixes moved during the year? Which were subdivided? Which became subject to a lock or dispute? Which were leased? Which generated proceeds? Which remained operationally necessary? A registry snapshot at year-end cannot answer the movement. A signed history can.

The auditor's evidence problem is wider than ownership

External auditors are not being asked merely to verify that an address range exists. For a material balance, they need evidence supporting existence, rights and obligations, valuation, classification, completeness, presentation and disclosure.

Existence means the exact prefixes can be reconciled to current registration and internal records. The list should exclude duplicate counting between parent and subsidiary, operating and sale portfolios, or lessor and customer. More-specific assignments should be understood rather than treated as additional addresses.

Rights and obligations require the applicable registry agreement, chain of transfers, legal entity name, service status, liens or contractual restrictions known to the company, customer leases and any dispute. A WHOIS entry can corroborate current registration but may not establish the entire chain or the reporting entity's economic interest.

Completeness runs in both directions. The company may omit historically allocated blocks because no fixed-asset record exists. It may also overstate control by counting space used by customers, subsidiaries or counterparties under arrangements that limit sale. Network configuration, registry records, RPKI state, contracts and billing systems may each reveal a different population.

Valuation requires transaction data or cash-flow assumptions matched to block characteristics. Management should distinguish observable approved transfers from asking prices. It should account for fees, brokerage, taxes, reputation remediation, legal review and expected sale time. Lease valuation should not treat contracted revenue as risk-free merely because addresses remain scarce.

Classification requires evidence of business purpose. Board decisions, sale programmes, ordinary trading activity, network forecasts and customer obligations may determine whether a portfolio is used, leased, held for sale or part of a disposal group. A registry cannot supply this evidence.

Presentation and disclosure require judgment about materiality. A company with a nil carrying value may still need to explain a material gain on sale. A company with a large indefinite-lived balance may need to disclose significant valuation and useful-life judgments. The audit opinion does not standardise the underlying noun; it tests whether the financial statements appropriately apply the chosen framework to the proven facts.

A minimum rights schedule would make comparisons possible

Companies should describe the asset before they value it. A minimum rights schedule can do that without pretending every legal system has reached the same property conclusion.

The schedule should identify the reporting entity and each registration entity. It should list exact prefixes, address counts without double-counting more-specifics, the relevant registry or registration service, resource history, agreement type and effective dates. It should say whether the resource is legacy, allocated, assigned, transferred or otherwise registered, using the institution's actual terminology.

It should then identify the economically relevant rights: use, registration maintenance, transfer request, subdivision, leasing or customer assignment where permitted, reverse DNS, RPKI and contact administration. Each right should be linked to its condition rather than described as absolute.

Restrictions belong beside the rights. These include transfer locks, membership duties, payment conditions, policy compliance, pending disputes, legal restraints, customer commitments, route-origin dependencies and any contractual consent. The schedule should separate a restriction on transfer from a restriction on use and from a practical obstacle to routing.

Finally, the schedule should map purpose and accounting treatment. Is the prefix used in the network, leased, held in reserve, actively marketed, contractually committed or expected to be returned? Is it recognised separately, included in goodwill, included in inventory, included in another asset group or unrecognised? What useful life, amortisation and impairment approach applies?

This schedule would not be a new accounting standard. It would be a factual bridge among engineering, legal, treasury and financial reporting. It would help an audit committee see why two blocks with the same address count may not have the same rights or value. It would also expose where management's classification depends on an assertion the registry record cannot support.

The schedule should remain private where contracts or security controls require confidentiality. Public disclosure can aggregate material facts without publishing credentials, customer identities or transaction-sensitive terms.

What a registration service can verify

A high-quality registration service can reduce the most expensive ambiguities without becoming an accountant.

It can verify identity: the legal entity currently recognised, the evidence used to establish succession, and the date at which a name or control change became effective.

It can verify scope: the exact prefixes covered, their parent-child relationships, whether the record is current or superseded, and whether the same addresses appear in conflicting current states.

It can verify contractual basis: the agreement under which services are provided, the version and effective date, whether it is active, and which published policy conditions apply.

It can verify transfer state: whether a transfer is permitted in principle, locked, pending, completed, rejected or subject to a recorded restraint; which parties participated; and the authoritative completion time.

It can verify service controls: who is authorised to maintain registration contacts, reverse DNS and routing-security records, without disclosing secret credentials.

It can verify history: prior recognised holders, subdivisions, consolidations, transfers and corrections through an additive record rather than an unexplained overwritten snapshot.

It can verify attestation quality: who signed, what evidence class was checked, when the attestation expires and whether a later event superseded it.

These facts would materially improve audits. They would not prove expected cash flows, sale price, inventory purpose, useful life, impairment, tax basis or legal priority among creditors. A registry should state that boundary clearly. The credibility of its attestation depends on refusing to certify conclusions it is not institutionally equipped to make.

What the registry must not decide

The desire for comparability can easily become a demand that the registry choose the accounting treatment. That would be a category error.

The registry should not declare that every IPv4 registration is property. Property status can depend on jurisdiction, contract, resource history and the precise legal question. Nor should it declare that no registration can be an asset. Accounting standards recognise many conditional contractual and licence rights.

It should not assign useful lives. A service relationship may have no fixed end while the company's economic use is finite. The registry can state contract duration and policy conditions; management and auditors assess the cash-flow horizon.

It should not publish fair values. Registry transfer data generally lacks complete price, condition and failed-deal information. A registry-sponsored unit price would acquire false authority and could distort fees, negotiations, tax analysis and impairment.

It should not label inventory. That classification depends on ordinary business activity and management purpose. The same block can be operating infrastructure in one company and trading stock in another.

It should not certify legal title free of all claims. It can state the current registration and recorded restraints. Unknown claims, insolvency priorities and jurisdictional disputes remain for competent legal processes.

It should not allow an audit request to expose confidential operations. Public verification can confirm a signed state without publishing customer assignments, private contracts or security arrangements.

Most importantly, it should not make accounting treatment a condition of registration. A company that uses a different standards-compliant classification should not lose service or transfer eligibility. Administrative power over the record must not become power over financial reporting.

The registry's proper maxim is simple: make facts reliable enough for others to reach conclusions, and make the boundary of those facts impossible to miss.

NRS can turn portability into audit evidence

Number Resource Society's positive contribution is not a universal chart of accounts. It is a more portable and verifiable evidence layer around registration rights.

Portability matters because control is weaker when the evidence of control depends entirely on one service provider that the holder cannot leave. If a registry can withhold records, delay a transfer or disappear, the company's asserted right contains an institutional dependency that auditors and valuers must price. A qualified receiving service, able to verify and continue the record under common rules, reduces that dependency.

An NRS-compatible attestation could identify the holder, prefixes, governing service agreement, current service, recognised transfer conditions, active cautions and a signed history. It could be checked by an auditor at the reporting date and later reproduced. If the holder changes service, the evidence history would move with the registration rather than reset.

That design would improve existence and rights testing. It would also make subsequent events clearer. A transfer after year-end could be linked to the prior reporting-date state. A subdivision could reconcile exactly. A dispute notation could show whether it existed before management completed a valuation. An expired lock would not remain as an undocumented oral assurance.

NRS should be evidence-bounded. Its attestation should avoid words such as owned, unencumbered, fair value and indefinite unless it is quoting a specified legal or accounting document and identifying the speaker. The core assertions should be administrative and reproducible: recognised, registered, active, superseded, transferable subject to named conditions, restrained by an identified instrument, or pending review.

Competition among registration services can also improve data quality if every provider must export the same minimum facts and preserve one current authoritative state. A company should not receive a stronger accounting asset merely because its recorder uses more proprietary language. It should receive better evidence because the service can prove what happened.

This is a restrained but important role. Better registration evidence will not eliminate accounting judgment. It will ensure that judgment begins from the same verifiable facts.

Disclosure should reconcile value without inventing a global total

Material holders can improve public reporting through a small set of disclosures.

First, state the nature of the recognised interest. "IPv4 addresses" may be concise, but the note should explain whether the balance represents acquired registration and use rights, a portfolio held for sale, lease-generating assets or a combination.

Second, state acquisition history. Readers should know whether the portfolio was separately purchased, acquired in a business combination, historically allocated at little or no recognised cost, or reclassified from another use. This explains why carrying amount may differ from apparent market value.

Third, state accounting life and method. For a finite-lived balance, disclose the life, amortisation method and major assumptions. For an indefinite-lived balance, explain why no foreseeable limit was identified and which events could change that conclusion.

Fourth, state the valuation basis and sensitivity where material. Market approach should identify the relevant transaction characteristics and adjustments without revealing confidential deal terms. Income approach should explain utilisation, price, churn, cost and terminal assumptions. A hybrid should reconcile its components.

Fifth, state significant registry and transfer dependencies. This need not reproduce contracts. It should explain whether continued registration, policy compliance, service status or transfer approval is necessary to realise value.

Sixth, reconcile movements: acquisitions, subdivisions that do not change total count, transfers, sales, impairments, amortisation and reclassifications. Address count can supplement monetary movement if it is defined consistently.

Seventh, distinguish recognised value from unrecognised potential. A company should not mark historical addresses to an unsupported market total. It can explain that certain portfolios have no material carrying amount and that disposals may create gains.

Such disclosure would improve comparability while preserving unavailable denominators. It would not claim how many IPv4 addresses worldwide are controlled by reporting companies, what proportion is recognised, or what share is available for sale. Those totals are not established by the selected public record.

Audit committees should ask where the accounting judgment can fail

The audit committee's task is not to choose a fashionable label. It is to identify the assumptions that could make a material balance or gain misleading.

For existence: Can the company reconcile every counted prefix to a current authoritative registration and eliminate overlap? Have sold, returned or transferred blocks been removed promptly?

For rights: Which legal entity is recognised? What agreement and policy apply? Is management relying on a database contact that no longer has corporate authority? Are legacy resources treated differently from contracted resources?

For control: Can the company maintain the registration, routing-security authority and operational records? Can it prevent unauthorised disposition? Does another party hold credentials or a contractual veto?

For classification: What evidence shows ordinary-course sale, operating use, leasing or strategic reserve? Has purpose changed? Is reclassification timely rather than chosen to improve a performance measure?

For useful life: What event would cause finite-life revision or indefinite-life reconsideration? How does the forecast treat IPv6 adoption, customer need, policy change and market liquidity? Is the horizon consistent with the business plan used elsewhere?

For valuation: Are comparable transfers actually comparable? Are failed sales, broker fees, reputation, block size and sale time reflected? Are lease cash flows double-counted in both address value and customer-relationship value?

For impairment: Which operating or market indicators are monitored? Did a lower-than-carrying-value disposal reveal information about the remaining portfolio? Has a registry dispute or transfer lock been treated as a legal footnote when it affects recoverability?

For disclosure: Would an investor understand why a nil-cost portfolio produced a large gain, or why another company amortises an apparently similar asset while this company does not?

These questions turn a naming dispute into governance. They also reveal why better registry evidence is valuable without being decisive.

Standard-setters should resist both silence and premature uniformity

The public evidence supports a case for clearer practice, not a case for a one-line universal rule.

Silence has costs. Material IPv4 balances can hide inside other intangibles. Historical portfolios can produce gains that look like recurring operating strength. Different useful-life judgments can affect profit comparability. Inventory and long-lived asset presentation can change margins. Investors may not know whether value depends on a registry contract, customer leases or an assumed resale market.

Premature uniformity has different costs. Requiring every portfolio to be an indefinite-lived intangible would ignore ordinary-course traders and finite economic-use forecasts. Requiring amortisation over one standard period would invent a global technological transition date. Treating all historical allocations as fair-valued assets would violate recognition and measurement principles and rely on a market denominator that does not exist. Treating no address right as an asset would ignore separability, contracts, cash flows and actual transactions.

A more useful standard-setting response would focus on disclosures and decision factors. When material, entities could be asked to explain the nature of the right, acquisition route, intended use, registry dependencies, useful-life judgment, valuation method and movement in count and carrying amount. Examples could distinguish operating rights, trading inventory, lease portfolios and business-combination recognition without asserting that every fact pattern has one answer.

Auditors and regulators can also challenge boilerplate. If management calls a portfolio indefinite-lived, it should identify the evidence. If it uses a market approach, it should explain comparability. If it reports a gain on a nil-cost sale, it should prevent readers from treating the gain as evidence of recurring margin without context.

The goal is disciplined variation. Different answers should follow different facts, not different levels of disclosure.

The accounting inconsistency is a map of institutional uncertainty

IPv4 accounting looks inconsistent because the underlying institution distributes several things at once: a unique number, a registry record, a contractual relationship, operational control, policy obligations and access to a transfer process. Markets then add price, leasing and financing. Courts and insolvency processes add claims about disposition. No single institution defines the whole bundle.

The public filings show the consequences. RM could sell nil-cost intangibles. Tessellis could divide acquired addresses between operating intangibles and inventory. Cogent could recognise an indefinite-lived asset. Uniti could amortise a finite-lived asset and later sell part below carrying value. Each report exposes a different junction of history, purpose and rights.

It would be comforting to select one treatment and call the others anomalies. The evidence does not support that conclusion. Nor does it support a claim that all companies disclose enough. The proper diagnosis is narrower: economically important IPv4 interests are being accounted for under general standards without a standard factual description of the right.

NRS can help by standardising that description. A portable record can prove who is recognised, what is covered, which agreement applies, what changed, which conditions remain and when a transfer became authoritative. The record can survive a change of service provider and can carry disputes without silently changing history.

The institution should stop there. It should not tell Cogent whether an asset is indefinite-lived, Tessellis whether a portfolio is inventory, Uniti what amortisation period to use or RM how to present a disposal gain. Those are conclusions about purpose, cash flows and applicable accounting standards.

The durable division of labour is therefore clear. Companies describe use and expected benefits. Auditors test recognition, measurement and disclosure. Standard-setters define accounting principles. Courts decide legal disputes. Registration services maintain verifiable facts. NRS makes those facts portable enough that no incumbent can monopolise the evidence.

An asset does not need one universal property label to be accounted for. It needs a precise account of the rights and dependencies that produce value.

When that account is missing, the balance sheet fills the gap with inconsistent names.

When it is available, disagreement can become reasoned judgment rather than institutional guesswork.

Sources