Summary

  • An IPv4 address block is not a machine and does not deteriorate through packet forwarding. When it is recognised as an intangible asset, the accounting question is normally amortisation rather than physical depreciation. If no foreseeable limit to its useful life can be supported, scheduled amortisation is less faithful than annual and event-driven impairment testing.
  • “Indefinite” does not mean “infinite”. IAS 38 makes that distinction explicitly. IPv6 adoption, changes in customer architecture, weakening transfer demand, an adverse registry decision, loss of routing acceptance or a shortened period of control can all turn a previously indefinite assessment into a finite life or an impairment indicator.
  • Technical life, market liquidity and registry risk are different variables. IPv4 can remain technically usable while the observable exit price falls; a liquid-looking market can exist while a particular block is disputed or difficult to transfer; a clean registry record can coexist with weak demand. Combining them into one unsupported useful-life number hides the mechanism of loss.
  • Public filings prove that sophisticated holders do not all reach the same accounting conclusion. DigitalOcean and Cogent have described IPv4 addresses as indefinite-lived intangible assets subject to impairment testing, while Uniti Group reported straight-line amortisation over 17.5 years. These are entity-specific judgments, not a representative sample or a global rule.
  • A credible impairment process starts with the exact unit of account and rights held. It then tests external sale evidence, value in use, registry and contractual status, block size, fragmentation, routing-security state, reputation, deployment cost and dependence on complementary systems. A headline price per address is not enough.
  • Legacy blocks carried at zero or a nominal amount can hold substantial economic value without creating an accounting gain merely because a broker quotes a price. Conversely, a purchased block can become impaired even while it remains routable. Recognition, measurement, market value and operational usefulness answer different questions.
  • The Number Resource Society perspective is valuable when kept within evidence: Internet numbers are durable, governed inputs to connectivity whose economic life depends on institutions as well as code. That perspective supports better stewardship and disclosure; it does not supply a universal valuation multiple or override the reporting framework applicable to a holder.

The asset survives the packets

A server grows old in familiar ways. Fans fail, storage wears, processors become inefficient and replacement parts become scarce. An office building consumes maintenance and loses utility as tenants and regulation change. Accountants can connect a pattern of depreciation to the consumption of those assets' productive capacity, even if the estimate is imperfect.

An IPv4 address block behaves differently. Sending a billion packets to an address does not consume a fraction of the address. Assigning the block to customers does not abrade its bits. A /16 remains a /16 after a busy year. If it is returned by one customer and correctly prepared, it can be used by another. The number itself does not have a mechanical duty cycle.

That physical durability creates an accounting temptation. One response is to declare that the asset never loses value. Another is to impose a conventional life—ten, fifteen or twenty years—because every long-lived asset must supposedly march towards zero. Neither response begins with the actual economic mechanism.

The asset is not merely a sequence of integers. What a holder can exploit is a bundle of capabilities: a recognised registration position, the ability to request changes under an applicable registry relationship, practical control over route authorization and directory data, acceptance by upstreams, and the capacity to deploy the addresses in services or transfer the recognised position under policy. Some of those capabilities can persist. Some can be interrupted. Some can become less valuable while the protocol remains unchanged.

The title's word “depreciation” is therefore a useful provocation but an imprecise accounting label. Under IFRS, a separately recognised non-monetary asset without physical substance is considered under IAS 38, where systematic allocation for a finite useful life is called amortisation. United States reporting likewise places acquired IPv4 addresses in intangible-asset disclosures in the public examples considered here. Everyday business speech may call any fall in book value depreciation. Financial statements should use the classification and terminology required by the applicable framework.

The deeper point survives the vocabulary. Book value should not decline merely because time passed if time did not consume the expected economic benefits. It should decline when the holder's recoverable economic position has declined, or over a supportable finite period if the benefits are expected to be consumed that way.

Begin with the rights and the unit of account

Before selecting a useful life, the holder must state what it has recognised. “IPv4” is too broad. A company may control one contiguous legacy block, many small transferred blocks, contractual rights to use another party's space, a managed lease portfolio, or a business whose customer relationships happen to include addresses. These are not interchangeable units.

IAS 38 asks whether an intangible resource is identifiable, controlled as a result of past events and expected to produce future economic benefits. Separate acquisition usually supplies an observable cost. Internally accumulated or legacy resources can be harder: the holder may have a valuable position but no qualifying cost at which to recognise a new asset. A balance sheet is not intended to become a live catalogue of every unrecognised source of corporate value.

That distinction explains why an old address block can be economically valuable and yet have a nil carrying amount. The company may have received it decades ago without a purchase price that became an intangible cost. Recognition rules do not permit management to manufacture profit by marking an internally held position to a promotional quotation. When RM plc announced a 2023 sale of addresses, it said the addresses were classified as intangible assets acquired at nil value. The cash sale disclosed value; it did not prove that every similar holder should previously have recorded the same value.

Acquisition changes the starting point. If a buyer pays specifically for a block, cost can include the purchase price and directly attributable preparation costs under the relevant framework. In a business combination, fair-value measurement can identify an IPv4 asset separately from goodwill where the facts and standards support it. Cogent's public reporting on its Sprint transaction is unusually illuminating because it recorded a material IPv4 intangible only after determining the transferred quantity and valuation approach in the context of a bargain purchase.

The unit of account then matters. Is the recognised asset an entire portfolio because blocks are managed and monetised together, or does each block have independently observable risk and cash flow? Can a disputed /16 be hidden inside a clean portfolio of millions of addresses? Can strong small-block sale evidence be applied to a large block without considering the different buyer set? IFRS 13 says the standard requiring measurement determines the unit of account. It does not authorize a valuer to choose whatever grouping produces the smoothest answer.

An impairment review cannot be better than its asset map. The map should reconcile exact prefixes, acquisition cohorts, carrying amounts, contracts, registry records, routing status and the operating services that depend on them. Without that reconciliation, a valuation model may be arithmetically exact and economically fictional.

Exhaustion created scarcity, not a perpetual price guarantee

The modern market story begins in 2011. IANA's central free pool reached its final allocation stage, and Regional Internet Registries progressively moved through their remaining pools under regional policies. The IANA IPv4 address-space registry records the allocation of the /8 blocks; RIPE NCC's 2012 last-block announcement describes the final /8 received by each RIR in February 2011.

Exhaustion changed the way additional capacity could be obtained. It did not freeze an invariant price into the protocol. Scarcity can support value only in combination with demand, transferability and practical use. IPv6 adoption can reduce some demand for additional IPv4. Carrier-grade NAT can stretch an operator's installed pool while adding cost and complexity. Cloud pricing can make dedicated public addresses more visible to customers. Hosting, security, enterprise allowlisting and machine-to-machine dependencies can preserve demand in particular uses. These forces do not move in one direction or on one timetable.

Nor is there one transparent global exchange. RIR transfer logs show completed registration changes, not necessarily the complete consideration, broker fee, financing, lease economics, warranties or rejected transactions. RIPE NCC's public transfer statistics list valid transfers and identify some changes due to business structure. ARIN publishes transfer statistics in its region. These datasets establish activity and exact blocks, but they do not supply a comprehensive denominator for attempted deals or a universal price series.

Private brokers publish observations. They may be useful inputs when the valuer understands the sample, block sizes, region, date and incentives. They are not a substitute for an accessible principal-market analysis. A broker's average may exclude transactions it did not intermediate, abandoned deals, leases, private corporate reorganisations or blocks whose risk made them unmarketable. Available public evidence supports no defensible global impairment rate.

Scarcity is therefore a condition, not a valuation conclusion. Gold is scarce and traded in a deep standardized market. IPv4 blocks are scarce but heterogeneous in size, history, contractual position, registry region, reputation and operational readiness. A valuer must resist turning the memorable date of exhaustion into an assumption that every block appreciates forever.

Indefinite is a reviewable conclusion, not a compliment

IAS 38 allows an intangible asset to have a finite or indefinite useful life. The decisive question is whether analysis of relevant factors reveals a foreseeable limit to the period over which the asset is expected to generate net cash inflows. The standard's language is careful: indefinite does not mean infinite.

That sentence fits IPv4 unusually well. The protocol can remain supported without promising endless economic benefits to one holder. An address block may have no contractual expiry and no evident consumption pattern. Yet its future depends on the holder's services, customer design, operating expenditure, legal and registry relationships, technical substitutes and market demand. If, after examining those factors, management cannot identify a foreseeable limit, non-amortisation with annual impairment testing can be coherent. If it can identify a limit, amortisation over that finite useful life may be required.

Indefinite life is not the optimistic choice and finite life is not the prudent choice. Prudence does not mean inventing a short life unsupported by evidence. IAS 38 explicitly says uncertainty does not justify choosing an unrealistically short life. Equally, “the Internet still uses IPv4” does not justify an indefinite conclusion without analysing the entity's period of control and expected benefit.

A finite assessment may arise where the addresses are inseparable in practice from a finite service contract, network concession, customer platform or planned transition. A buyer might acquire addresses as part of a business expected to run down over a known period. Management might have an approved and funded plan to migrate the relevant service and dispose of the addresses. A contractual right to use a block may expire even though the underlying addresses continue. In such cases the economic asset recognised by that holder is finite even if IPv4 as a protocol is not.

An indefinite assessment may be supportable where the holder has a stable registration and contractual position, continuing demand, no planned disposal, maintainable operational capability and no foreseeable technical or legal cut-off. That conclusion brings discipline, not relief: IAS 36 requires annual recoverable-amount testing for indefinite-lived intangibles and additional testing when impairment indications arise. IAS 38 also requires the useful-life assessment to be reviewed each period.

The status can change. If evidence later supports only a finite life, the holder accounts for the change as an estimate under the applicable rules and examines impairment. A model that writes “indefinite” at acquisition and never revisits it is not applying the concept. It is using the label to avoid measurement.

Public filings reveal legitimate disagreement

Three recent United States filings demonstrate why no external observer can prescribe one accounting life to every holder.

DigitalOcean's 2025 Form 10-K describes IP addresses used by customers to host servers as indefinite-lived intangible assets. It reported carrying amounts of $46.657 million at the end of 2025 and $44.822 million at the end of 2024. The company said it performs annual impairment assessment and additional assessment when events or circumstances indicate possible impairment. It reported no impairment for 2023-2025.

Cogent's 2026 quarterly disclosure, covering its acquired Sprint Business assets, reports $458 million of acquired IPv4 addresses. It says fair value was based on recent auction prices with a factor for uncertainty about how the market will function. Cogent concluded the asset had an indefinite useful life, did not amortise it, and performed an annual or triggering-event impairment assessment. It reported no impairment from 1 May 2023 through 31 March 2026.

Uniti Group's 2025 Form 10-K presents a different judgment. It lists IPv4 addresses among finite-lived intangible assets and applies straight-line amortisation over 17.5 years. The filing also says that after 1 August 2025 the company sold certain unused addresses with a net carrying value of $6.7 million for $5 million in cash.

These disclosures are not inconsistent simply because the numbers differ. The entities acquired different portfolios in different transactions, use them in different businesses and apply their frameworks to their own evidence. The filings do not reveal every valuation workpaper, legal opinion or portfolio characteristic. They are examples, not a statistically representative study of holders.

They nevertheless establish four important facts. First, recognised IPv4 assets can be financially material. Second, large public companies and their auditors have accepted both finite and indefinite treatments in disclosed circumstances. Third, market-price uncertainty can be an explicit valuation input even when useful life is indefinite. Fourth, an actual sale below net carrying value can provide sharper evidence for a cohort than an abstract belief that scarcity supports appreciation.

The proper lesson is not to copy Cogent, DigitalOcean or Uniti. It is to demand the evidence that would make the chosen treatment equally intelligible. If a holder selects 17.5 years, what expected consumption, transition or control period supports that number? If it selects indefinite life, what analysis shows no foreseeable limit, and how will annual impairment capture adverse change? If it uses recent auctions, how were block size, region, transaction date and sale conditions adjusted?

Accounting policy should make disagreement comparable. Boilerplate turns disagreement into fog.

Technical life is the protocol question

Technical life asks whether the addresses can continue to perform their addressing and routing function in the holder's intended environment. It is narrower than market value and broader than a simple ping test.

IPv4 is deeply embedded in network equipment, operating systems, applications and access networks. The exhaustion of unallocated supply did not disable allocated addresses. Dual-stack networks, translation technologies and IPv4-as-a-service can prolong operational use while IPv6 deployment expands. A block may therefore remain technically serviceable long after its acquisition date.

But technical serviceability has layers. The prefix must be accepted at a useful length by upstreams. The holder needs an origin ASN or a provider arrangement. Route Origin Authorizations and filters must align with intended announcements. Reverse DNS may matter. Customer systems may depend on stable source addresses. Abuse handling and reputation can affect whether counterparties accept traffic. A number that can be placed in a router configuration is not necessarily a fully productive network input.

Technical obsolescence should be evidenced through architecture rather than fashion. Useful questions include: Which revenue services still require public IPv4? Which can operate with IPv6-only access plus translation? What is the tested migration cost? What percentage of the relevant customer workload—not global Internet traffic—remains dependent on the addresses? Which complementary systems would need replacement? What approved investment plan changes that dependence?

The denominator must match the decision. A global estimate of IPv6 capability does not tell a hosting company how many of its contracted customers can surrender dedicated IPv4 tomorrow. A mobile network's translation experience does not determine the life of an enterprise security service built around allowlisted source addresses. Technical life is entity-specific because economic benefit is entity-specific.

A holder should document technical indicators each reporting period: utilisation by prefix, service attachment, revenue dependency, dual-stack and translation readiness, cost of maintaining address-management controls, routing-security coverage, and credible retirement plans. The evidence should distinguish a reversible engineering inconvenience from a structural decline in benefit.

Technical life can be long without being the same as economic life. That is precisely why it deserves its own analysis.

Market liquidity is the exit question

Liquidity asks what the holder could realise from an orderly disposal, how long disposal would take and how uncertain the eligible buyer set is. It is not answered by the protocol's continued use.

IFRS 13 defines fair value as an exit price in an orderly transaction between market entities at the measurement date. The measurement uses market-entity assumptions and the principal market accessible to the entity, or the most advantageous market in the absence of a principal market. Management's intention to hold is not the fair-value premise.

For IPv4, access matters. A holder may operate in one RIR service region but be eligible for an inter-RIR transfer only where both policy systems permit it. A block may be too large for the most active buyer segment or too small to justify diligence costs. Fragmenting a large block can broaden demand but create time, transaction cost, deaggregation and operational consequences. Legacy status, contractual terms and registry agreements can affect the path to recognition.

Observed price should therefore be treated as a vector rather than a scalar. Relevant dimensions include prefix size, contiguous versus fragmented composition, registry region, legacy or post-RIR history, transfer route, date, payment terms, warranties, broker services, reputation, current routing and whether the transaction was distressed. A price for a clean /24 cannot be multiplied across a /12 without analysing market depth. A forced insolvency sale cannot automatically define an orderly exit. An advertised asking price is not a completed transaction.

Time is part of liquidity risk even when accounting fair value is measured at a date. A block that can probably be sold in twelve months with extensive remediation is not economically equivalent to one with immediate qualified demand. Costs of disposal include more than a broker invoice: legal review, corporate records, registry fees, technical separation, customer migration, route and reverse-DNS changes, taxes and possible holdbacks can matter.

RIR logs are valuable because they confirm that specified transfers were registered. Their limitation is equally important: they generally do not report full economic terms or failed attempts. Private sales datasets can add price observations but may not reveal the whole market. The valuer should preserve the sample definition and exclusions instead of presenting an average as an official global rate.

Liquidity indicators for impairment include declining credible offers for comparable blocks, widening dispersion by block size, longer completion times, repeated failed diligence, increased disposal costs, reduced inter-RIR reach, and actual sales below carrying value. Positive evidence can include recent comparable orderly transactions, qualified bids and stable demand from users with demonstrated deployment needs. Both sides should be considered.

Market liquidity is thus a test of realizability, not a referendum on IPv4's technical relevance.

Registry risk is the control question

Registry risk asks whether the holder can continue to exercise and transfer the recognised position through the institutions that maintain authoritative regional records and associated services. It is neither a market discount nor a routing incident, though it can affect both.

ARIN's current Registration Services Agreement describes included number resources as registration rights and identifies services including registry entries, reverse name service, RPKI, record maintenance and address-space administration. It grants specified rights subject to continuing compliance, including the exclusive right to be registrant in ARIN's database, use within that database and transfer of registration under policy. That language illustrates why a holder's accounting asset should not be described as an unconditioned physical entity.

APNIC's transfer policy requires the source to be the current registered holder and not involved in a dispute over resource status. Its transfer conditions explain that on completion the source no longer has rights to the transferred resources and the recipient becomes registered subject to current policy. Other regions use their own policies and agreements. The details cannot be collapsed into a global legal proposition.

Registry risk has ordinary and exceptional forms. Ordinary risks include stale corporate contacts, unpaid fees where applicable, an unsigned or misunderstood agreement, weak account access, poor merger records and inconsistent organisation names. Exceptional risks include competing claims, litigation, sanctions, fraud, unauthorized record modification and a policy incompatibility that prevents an expected transfer. ARIN maintains a fraud-reporting channel for suspected fraudulently obtained resources or unauthorized record changes; the existence of such a channel is evidence that record integrity has an operational security surface, not evidence of a measured incident rate.

An adverse registry event can affect value through several mechanisms. It can delay a sale, narrow the buyer set, prevent a record change, disrupt RPKI management, undermine upstream confidence, redirect notices or force legal expenditure. The block may remain visible in BGP throughout, yet the holder's ability to control and monetise it may be impaired. Conversely, a temporary routing outage does not necessarily mean the registration position lost value permanently.

The impairment analysis should therefore maintain a registry-risk schedule for each material block: registered entity, agreement status, authoritative contacts, fee status where relevant, transfer eligibility, known claims, corporate chain, RPKI authority and evidence of prior transfers. Legal advice may be required, but a legal memorandum cannot substitute for current registry verification.

Registry risk is institutional depreciation without scheduled decay: control can weaken because the surrounding authority fails to match the asset story. That change belongs in recoverability, not in an arbitrary annual percentage.

Impairment should follow observable mechanisms

IAS 36 states the core principle plainly: an asset should not be carried above the amount recoverable through use or sale. Recoverable amount is the higher of fair value less costs of disposal and value in use. Indefinite-lived intangibles are tested annually; other assets are tested when impairment indications exist.

For an IPv4 block, triggers should be designed around the three separate surfaces.

Technical triggers include an approved migration that makes a material cohort surplus, loss of support in a critical service, inability to route at an accepted prefix length, persistent RPKI or filtering obstacles, or a customer architecture change that removes expected cash flows. A brief outage may not imply impairment if service and demand recover. A permanent design decision can.

Liquidity triggers include credible transactions below the modelled value, a failed sale after adequate exposure, reduced qualified demand for the relevant block size, higher transaction costs, or a market structure change that makes prior comparables stale. The Uniti disclosure of sale proceeds below net carrying value is the sort of fact that requires cohort-specific analysis; it does not establish impairment for unrelated portfolios.

Registry triggers include a dispute hold, loss of authenticated account control, an adverse final determination, inability to establish the corporate chain, threatened revocation under an applicable agreement, or a transfer policy change that removes an assumed exit path. A mere change in public contact data may be an administrative issue; a competing claim to the source entity's status is a recoverability issue.

Other indicators can interact. A block associated with sustained abuse may require a long remediation period and sell at a discount. An overbroad ROA or unauthorized route may damage traffic reputation without changing registration. A lender covenant tied to utilisation can force sales into a weak market. A corporate divestiture can separate the addresses from the service that generated value in use.

The review should avoid double counting. If an adverse registry position is already reflected in comparable transaction discounts, applying a second arbitrary registry-risk percentage may duplicate the same fact. If value-in-use cash flows already include a funded IPv6 transition, a separate full obsolescence deduction may duplicate it. Risk belongs either in cash flows, probability weighting or discount rate according to the valuation method—not everywhere at once.

An impairment trigger is not automatically an impairment loss. It is a reason to measure. That distinction protects the accounts from both complacency and panic.

Value in use and sale value answer different holder questions

A network operator may earn more by using addresses than a market entity would pay to acquire them. Another holder may have surplus addresses generating no independent cash inflow, making disposal evidence more important. IAS 36's higher-of-two approach reflects that difference.

Value in use requires cash flows from the asset in its current condition. For a block embedded in hosting, broadband or security services, directly attributable revenue may be difficult to isolate. The appropriate level may be the smallest cash-generating unit whose inflows are substantially independent. That does not permit management to shelter a weak address asset inside the entire company merely because all networking eventually supports revenue.

The model should identify the counterfactual. What would happen without this address cohort? Would customers leave, accept translated or shared addresses, pay for alternatives, or migrate to IPv6? What replacement or leasing cost would be incurred? Which gross margins depend on dedicated IPv4? How much capital and operating expense is required to sustain the existing benefits? The forecast should be consistent with approved budgets and should not include speculative future enhancements that are not part of the asset's current condition where the framework excludes them.

Fair value less costs of disposal asks about an exit to market entities, not the holder's idiosyncratic value. The valuer may use a market approach based on adjusted comparables, an income approach based on lease or service cash flows, or another supportable method. A market approach needs transaction quality and adjustment logic. An income approach needs utilisation, churn, pricing, operating cost, default, remediation and terminal assumptions. A hybrid must not count the same economic benefit twice.

The two values can diverge for good reasons. A bank's allowlisted addresses may support high internal continuity value but be ordinary in the transfer market. A clean surplus block may have little value in use and strong sale value. A disputed block may remain useful in an existing network while having little immediate transferability. The carrying amount is impaired only when it exceeds the relevant recoverable amount under the applicable standard, not whenever one of the two measures weakens.

This separation also improves governance. Operations can own technical-use assumptions. Treasury and transaction specialists can own liquidity evidence. Legal and registry teams can own control status. Finance can reconcile the models and prevent incompatible facts. One department should not silently decide all three dimensions.

A valuation bridge is better than a price per address

The most tempting model multiplies the number of addresses by a quoted unit price. It is an acceptable opening observation and a poor final conclusion.

A defensible bridge begins with genuinely comparable transactions or a supportable income base. It then explains adjustments rather than hiding them in a single “risk discount”. Block-size adjustment addresses the fact that buyer depth and unit economics can differ between a /24, /20, /16 and very large aggregate. Date adjustment addresses market movement. Region and transfer-path adjustment addresses eligibility and process. Legacy and contractual adjustment addresses the recognised position. Reputation and routing adjustment addresses remediation. Fragmentation adjustment addresses whether the portfolio can be sold as assumed. Costs of disposal are then deducted where required.

Each adjustment needs evidence and a sign. A large block is not automatically worth less per address; the holder must show the relevant market. A legacy block is not automatically more valuable; history can bring either flexibility or documentation risk. Current routing can prove operational use but can also burden transition. A signed registry agreement can improve service clarity but its economic effect depends on the buyer and region.

For an income approach, the bridge starts with productive addresses, not total addresses. It should preserve the denominator: total addresses, reserved or unusable addresses, internally deployed addresses, leased addresses, addresses in remediation and genuinely available inventory. Utilisation claims without these categories are easy to inflate. Revenue should reconcile to contracts and cash collection. Forecast lease rates should reflect renewals, customer concentration and services included. Operating expenses should include registry administration, abuse response, routing security, legal support and sales cost where material.

The terminal assumption deserves special scrutiny. An indefinite accounting life does not justify a perpetual-growth formula that assumes demand survives unchanged forever. The valuer can use declining cash flows, probability-weighted exit scenarios, a finite explicit period with a supportable terminal value, or market evidence appropriate to the facts. The purpose is to represent uncertainty, not to make it disappear behind a discount rate.

Sensitivity should show the variables that could actually reverse headroom: unit exit price, qualified demand, transfer duration, productive utilisation, lease margin, IPv6 substitution in the relevant customer base, registry outcome and remediation cost. A ten-per-cent generic sensitivity may be less informative than a scenario in which a disputed block cannot transfer for two years.

Finally, actual disposals must feed back into the model. If the holder repeatedly sells below appraised value, it cannot dismiss each sale as anomalous without evidence. If it sells above carrying value, that may support value but not necessarily the whole portfolio. Transaction-specific differences should be documented before extrapolation.

The bridge makes judgment auditable. A unit price makes it look effortless.

Evidence must survive both the audit and the network room

An accountant cannot validate IPv4 value from a fixed-asset register alone. A network engineer cannot validate carrying value from a route collector alone. The evidence package must connect financial rights to operating facts.

At minimum, the holder should reconcile the prefix inventory to RIR and delegated registry records, acquisition agreements, corporate records and the general ledger. It should identify which entity is registered, which entity paid, which entity operates the network and which entity receives revenue. Differences may be legitimate within a group, but they must be explained.

For technical use, preserve BGP origin observations, ROA configurations, upstream authorization, reverse-DNS control, internal assignment records, customer attachment and incident history. None alone proves economic ownership. Together they show whether the asset is in its represented condition and whether remediation assumptions are credible.

For market evidence, preserve full transaction details where available: CIDRs, size, region, date, counterparty qualification, price, fees, holdbacks, warranties, rejected bids and completion status. A screenshot of a broker dashboard without sample definition is weak evidence. So is a public transfer record treated as proof of an undisclosed price.

For registry risk, preserve the current agreement, authenticated account roles, paid invoices where applicable, transfer eligibility, correspondence on disputes, corporate succession and legal opinions tied to exact resources. Public Whois or RDAP is a starting point, not the complete rights file.

The audit trail should record contrary evidence. If management uses an indefinite life while a board paper plans a ten-year IPv4 exit, the conflict must be resolved. If a valuation uses strong small-block prices while the portfolio consists of a very large aggregate, the adjustment must be visible. If a sale process failed, the bids and reasons matter even if no transaction appears in an RIR log.

External specialists can help but do not remove management's responsibility. A valuation firm may understand market comparables but not registry policy. Network staff may understand routability but not fair-value premise. Counsel may understand contractual control but not customer substitution. The holder needs a joined conclusion with clearly owned inputs.

This is not procedural excess. An address portfolio can be material, portable and operationally entangled. A concise evidence pack reduces the chance that a large impairment appears years after its underlying cause became observable.

Governance should treat the annual test as stewardship

The annual review is often presented as a compliance ritual. For number resources it can be a useful capital-allocation meeting.

The board or audit committee should receive a bridge from opening to closing carrying amount, additions, disposals, amortisation if any, impairment, reversals where permitted and changes in useful-life judgment. It should also receive operational quantities: total scheduled addresses, productive addresses, surplus addresses, disputed addresses and addresses under remediation. The quantities should reconcile to exact prefix inventories rather than a rounded marketing total.

Management should approve trigger thresholds that cause an interim review. These can include a material completed sale below carrying value, a qualified offer outside the model range, loss of registry control, a formal dispute, a major customer or architecture change, an adverse routing-security event with lasting impact, or approval of a migration plan. Thresholds should prompt analysis, not automate a write-down.

Controls should address who can change registry contacts, RPKI authorizations, transfer requests and internal inventory status. The people able to impair operational usefulness should not be invisible to finance. A compromised account can become a financial reporting event if it jeopardizes control or recoverability. Likewise, finance should not commit to a disposal assumption that network teams cannot execute without customer harm.

Compensation deserves attention. If managers are rewarded for sale proceeds, they may classify productive addresses as surplus. If earnings targets benefit from indefinite life, they may resist finite-life evidence. If lenders use portfolio value or utilisation covenants, assumptions can affect financing. Independent review and transparent sensitivities are therefore more than technical accounting hygiene.

Disclosure should be proportionate but specific. Material holders can explain classification, useful life, testing frequency, valuation technique, important unobservable inputs and impairment outcomes as required by their framework. They should avoid implying that an RIR endorsed their valuation or that a quoted market average is official. Where uncertainty is material, saying what can change the result is more useful than adding adjectives to scarcity.

Good stewardship also recognises unbooked value. A legacy block with nil carrying amount may deserve the same security, registry and utilisation controls as a purchased asset. The absence of a balance-sheet amount is not evidence of operational insignificance. Conversely, a large carrying amount does not make a block permanently strategic.

Accounting is most useful here when it forces the company to ask what it controls, how it earns, how it could exit and what could break.

What a Number Resource Society adds

The Number Resource Society idea is strongest when it changes the quality of those questions rather than offering an ideology of permanent appreciation.

It begins from a sound observation: Internet numbers are not disposable technical trivia. They are durable coordination resources whose usefulness depends on shared protocols, registries, operators, contracts and trust. Their stewardship can affect continuity, competition, security and the ability of new networks to connect. Treating them as managed capital can expose neglected inventory and weak control.

That perspective is positive for holders and for the Internet. A holder that knows its prefixes, protects registry accounts, publishes accurate routing authorization, responds to abuse and plans transfers carefully creates more reliable economic value than one that merely waits for scarcity to raise prices. Better records reduce disputes. Better routing security reduces avoidable interruption. Better disclosure lets lenders and investors distinguish productive portfolios from speculative totals.

The evidence boundary matters. NRS does not prove that every address should be recognised as an asset. It does not decide whether a particular life is finite or indefinite. It does not convert a transfer listing into a sale price, determine a registry dispute or guarantee that IPv4 demand will outlast a forecast. Those questions remain with the applicable accounting framework and facts.

Its practical contribution is a three-ledger discipline. The technical ledger records deployability and dependence. The market ledger records accessible liquidity and transaction evidence. The institutional ledger records registration, contractual and policy control. Finance can then measure the recognised asset without asking one ledger to impersonate the others.

That is a more mature form of advocacy than declaring addresses digital real estate. Real estate has title systems, local markets, physical condition and zoning; IPv4 has different institutions and failure modes. The analogy can attract attention but should not drive measurement.

NRS is therefore compatible with conservative reporting precisely because it asks holders to preserve evidence. It treats value as earned through stewardship, not bestowed by a slogan.

The policy that follows the economics

A defensible accounting policy can be stated without pretending to settle every jurisdiction.

First, identify the recognised intangible and its unit of account. Separate purchased blocks, acquired portfolios, leases, customer relationships and internally held legacy positions. Reconcile exact resources to rights and cost.

Second, determine useful life from expected economic benefit and period of control. Do not use physical wear. Do not equate protocol persistence with an infinite life. If no foreseeable limit is supportable, treat the life as indefinite under the applicable framework and perform the required annual and event-driven tests. If a foreseeable limit exists, amortise over the supportable finite life and test when indicators arise.

Third, measure impairment through the mechanism. Technical life establishes current and expected use. Market liquidity establishes realizable exit evidence. Registry risk establishes whether control and transfer assumptions remain valid. Keep the inputs separate, then reconcile them in fair value less costs of disposal and value in use without double counting.

Fourth, let new evidence change the conclusion. An actual sale, a failed transfer, a dispute, a customer migration or a revised architecture can matter more than last year's policy memo. Useful life is an estimate, not an identity.

Finally, disclose uncertainty honestly. The public record does not provide a global rate of IPv4 impairment, a complete transaction denominator or a universal retirement date. That absence is not permission to guess. It is a reason to show entity-specific evidence and sensitivity.

An address block does not wear out. The economic system around it can. The most faithful accounts recognise that difference: no automatic erosion for imaginary physical consumption, no complacency about indefinite life, and prompt impairment when use, liquidity or governed control can no longer support the carrying amount.

Sources