Summary

  • AFRINIC’s detailed financial webpage states that sixteen “Other Expenses” categories totalled USD 854,266 in 2025. The sixteen displayed amounts actually add to USD 854,267. A separate document, the 2025 Unaudited Finance Report, lists Members Training at USD 2,213 instead of the webpage’s USD 2,214; substituting that one figure makes the schedule reconcile exactly to the stated total.
  • The difference is a correction and version-control issue, not evidence of a crime or a material misstatement. The much larger question is what the aggregate categories bought, which office approved each commitment during the receiver-to-Board transition, and what measurable registry service resulted.
  • Remote Site Expenses, Computer Expenses and Bank charges together account for USD 433,541, or 50.75% of the stated envelope. The five largest lines account for 68.53%, and the eight largest for 87.97%, making focused cost-to-output disclosure more useful than a generic defence of the whole schedule.
  • AFRINIC’s stated Other Expenses total rose by USD 199,786, or 30.53%, from USD 654,480 in 2024. Changes in activity, allocation or classification may explain some or all of that movement; the public figures alone cannot establish waste, overpricing, fraud, good value or even whether every amount was paid in cash during 2025.
  • Election costs of USD 1,043,425 and legal costs of USD 877,929 are disclosed separately and are not part of the USD 854,266 schedule. They provide scale context only. They must not be folded into Other Expenses or used to import separate election and litigation controversies into this analysis.
  • AFRINIC is a private number-registry bookkeeper and coordinator. It has no sovereign, taxing or punitive authority over operators or number-resource assets. Its practical importance therefore sharpens, rather than relaxes, the obligation to connect compulsory member-funded costs to a narrow technical function, a responsible decision-maker, valid authority and a verifiable result.

Within its official financial material for 2025, AFRINIC published a detailed schedule labelled “Other Expenses 2025” and “NOTE 15: EXPENSES BY NATURE (2025)”. It gave a total of USD 854,266 and displayed sixteen categories. Add those sixteen webpage figures, however, and the result is USD 854,267. The discrepancy can be traced to one line: the detailed webpage says USD 2,214 for Members Training, while the separate seven-page 2025 Unaudited Finance Report says USD 2,213. Use the latter number and the schedule closes precisely at USD 854,266.

That is what happened. It matters because even a tiny public inconsistency shows whether financial information has a clear source of truth, correction path and accountable owner. Yet its scale also disciplines the conclusion. One dollar is not evidence of criminality, corruption, fraud, embezzlement, waste or a material accounting failure. It is a one-dollar reconciliation question. The proportionate remedy is a dated correction that identifies the intended Members Training figure and explains which publication controls.

The larger governance test lies behind the arithmetic. Three operationally plausible categories—Remote Site Expenses, Computer Expenses and Bank charges—absorbed just over half of the stated envelope. AFRINIC operated through an extraordinary year in which a receiver was charged with securing assets and restoring the Board, the Board was announced as reconstituted in September, and a joint statement in October described continuing collaboration while the receiver’s discharge remained pending.

Aggregate disclosure does not show who defined each need, committed the organisation, accepted delivery, approved an invoice, authorised payment, classified the cost or reviewed the result at each point in that transition. For a private technical bookkeeper funded by members, that missing chain is the substantive issue.

Analysis

One label, sixteen different economic questions

“Other Expenses” sounds residual. It should not be read as a verdict. The label is AFRINIC’s published classification, not a declaration that the spending was exceptional, optional, unnecessary, wasteful or suspicious. Several items could be ordinary costs of keeping a technical organisation functioning. Others may support staff or institutional participation. The public schedule does not disclose enough detail to decide whether any one category was core, supplementary, fixed, variable or the result of an accounting allocation.

The detailed webpage lists the sixteen 2025 lines as follows:

Rank Category Detailed webpage amount (USD) Share of stated USD 854,266 total
1 Remote Site Expenses 171,606 20.09%
2 Computer Expenses 159,134 18.63%
3 Bank charges 102,801 12.03%
4 Staff Benefits 81,237 9.51%
5 Insurance costs 70,622 8.27%
6 Staff Welfare 58,021 6.79%
7 Contribution to ICANN 54,247 6.35%
8 Telephone & Faxes 53,870 6.31%
9 NRO Shared Expenses 37,979 4.45%
10 Printing, Postage & Stationery 17,417 2.04%
11 Professional Fees 17,073 2.00%
12 Community Engagement 16,064 1.88%
13 Staff Training 6,163 0.72%
14 Recruitment Costs 4,641 0.54%
15 Members Training 2,214 0.26%
16 Marketing & Communication Expenses 1,178 0.14%

The ranking is useful because it points to the questions with the greatest explanatory value. Remote sites, computers and bank charges total USD 433,541, equal to 50.75% of the stated total. Add Staff Benefits and Insurance costs and the top five reach USD 585,400, or 68.53%. Add Staff Welfare, Contribution to ICANN and Telephone & Faxes and the top eight reach USD 751,538, or 87.97%. A member does not need every low-value receipt published to understand most of the envelope. A well-designed disclosure could explain the few concentrated cost centres first, with security- and privacy-protective aggregation.

Concentration is not culpability. Remote infrastructure, computing and payment rails can be expensive, particularly during organisational recovery. Nor does a familiar label establish necessity. “Remote Site Expenses” does not reveal how many sites existed, where they were, which services they hosted, whether costs were rent, transit, hardware, support or energy, or whether the spending improved availability. “Computer Expenses” does not distinguish equipment from software subscriptions, cloud services, security tooling, maintenance or depreciation-related treatment.

“Bank charges” does not show the number of accounts, transaction volumes, currencies, correspondent fees, merchant fees, foreign-exchange spreads or efforts to reduce avoidable cost.

The same evidential gap runs down the table. Staff Benefits and Staff Welfare may reflect legitimate employment obligations or a changed policy, but the schedule does not expose the policy, eligible population or allocation basis. Insurance costs may protect assets, people or liability risks, yet coverage, premiums, claims and broker arrangements remain unseen. Contributions to ICANN and NRO shared expenses identify institutional counterparties in broad terms, but the names alone do not prove the invoiced amount, necessity, approval or benefit.

Communications, printing, professional fees, engagement, training, recruitment and marketing all admit plausible explanations. None can be evaluated merely by reading its account title.

The one-dollar reconciliation should remain visible

There are three distinct official surfaces in the sealed record, and they should not be collapsed into one. First is the detailed HTML schedule, whose sixteen displayed amounts total USD 854,267 while its stated total is USD 854,266. Second is the seven-page document that calls itself the 2025 Unaudited Finance Report, where Members Training is USD 2,213 and the sixteen-item schedule therefore reconciles to USD 854,266. Third is a separately linked audited 2025 annual financial statement. The detailed page, the unaudited summary and the audited statements are related publications, not interchangeable instruments.

The public record does not identify which Members Training value was corrected, which was entered first or which is authoritative at line-item level. It also does not provide a version note or formal reconciliation. That uncertainty must remain intact. Silently changing the detailed page to USD 2,213 in an analysis would make the arithmetic tidy by erasing the evidence of inconsistent publication. Calling USD 2,214 definitive would ignore the summary value that exactly closes the total.

The accurate formulation is simply that the webpage displays USD 2,214, the unaudited summary displays USD 2,213, and only the latter reconciles the stated total.

The distinction between precision and significance matters. Reproducible arithmetic establishes the one-dollar residual. It does not establish why it arose, whether the underlying ledger contains either value, or whether it affected the audited financial statements. At this scale, the difference is plainly not a basis for implying material harm. But immateriality at financial-statement level is not a reason to leave a public field internally inconsistent. Accurate records are part of a registry’s craft. A dated correction would improve trust precisely because it is modest, transparent and resistant to exaggeration.

What changed from 2024—and what a percentage cannot say

AFRINIC’s published comparator gives Other Expenses of USD 654,480 for 2024. Against the stated 2025 total of USD 854,266, the increase is USD 199,786, or 30.53%. That movement is material enough to merit explanation, but it is not self-interpreting. A higher account balance may reflect more activity, a price change, a different allocation, a category reclassification, a catch-up after deferred work, exchange-rate effects, or some combination. Without units, vendors, accounting policies and outputs, a percentage cannot adjudicate efficiency or fault.

The largest absolute disclosed increases were Computer Expenses, up USD 61,456 from USD 97,678 to USD 159,134; Staff Welfare, up USD 51,678 from USD 6,343 to USD 58,021; and Remote Site Expenses, up USD 49,295 from USD 122,311 to USD 171,606. Staff Benefits rose USD 20,669, from USD 60,568 to USD 81,237. Community Engagement appears at USD 16,064 in 2025 and was not listed in the 2024 table. Its absence there is not proof that no related activity occurred in 2024; the classification or disclosure may have changed.

Several percentage changes look dramatic because the base was small. Staff Welfare rose 814.7%, and Recruitment Costs rose 364.1%, from USD 1,000 to USD 4,641. Marketing & Communication Expenses rose 53.2%, but the dollar movement was only USD 409. A serious reading should therefore keep absolute and percentage changes together. The welfare increase merits a policy and allocation explanation because both its dollar and percentage movements are notable. The recruitment percentage is much less important to the overall envelope. Neither figure establishes impropriety.

Other categories moved modestly or declined. Bank charges rose USD 8,119, or 8.6%. Insurance increased USD 1,857, or 2.7%. Contribution to ICANN increased USD 939, or 1.8%. Professional Fees increased USD 987, or 6.1%. Telephone & Faxes fell USD 2,181, or 3.9%; NRO Shared Expenses fell USD 1,339, or 3.4%; Printing, Postage & Stationery and Staff Training each fell 26.4%, by USD 6,236 and USD 2,212 respectively. Consultancy Fees, listed at USD 3,585 in 2024, disappears from the 2025 schedule. That could mean the activity ended, moved to another category or was absent; the table alone cannot choose among those explanations.

Members Training illustrates why version control matters even in a year-on-year comparison. Using the detailed webpage’s USD 2,214 gives an increase of USD 226, or 11.4%, from USD 1,988 in 2024. Using the unaudited summary’s USD 2,213 changes the dollar movement by one. Neither result has material economic significance. The point is that analysts should not manufacture a single authoritative figure when AFRINIC’s own public surfaces differ.

An expense is not necessarily a cash payment

The schedule covers an accounting period. It does not disclose, line by line, whether the amounts were paid in cash during 2025, accrued for later payment, prepaid in an earlier period, allocated from another account, translated from another currency, stated with or without tax, or reduced by credits. Language such as “AFRINIC paid USD 171,606 for remote sites” would therefore be stronger than the evidence supports. The defensible statement is that AFRINIC recognised or disclosed USD 171,606 in the Remote Site Expenses category for 2025.

This boundary is important because the unaudited summary also reports year-end cash holdings in three currencies: USD 18,821,319, EUR 1,258,151 and MUR 3,485,473. Those balances cannot simply be added without an exchange rate. More importantly, aggregate cash at year end does not trace a particular expense, prove its payment date or establish its authorisation. Liquidity can answer whether an organisation appears able to meet obligations at an aggregate level. It cannot substitute for an invoice, an approval, a bank mandate or a transaction ledger.

The unaudited summary supplies useful scale. It reports 2025 revenue of USD 6,262,765, administrative expenses of USD 4,614,866, distribution expenses of USD 499,280 and total operating expenses of USD 5,114,146. On that basis, the stated Other Expenses envelope is 16.70% of operating expenses and 13.64% of revenue. These are mechanical ratios, not audited transaction-level findings. They show that the schedule is significant enough to deserve a bridge into the statements, not that it represents the whole cost base.

The summary exposes many of the same account names directly. It also reports an HR line of USD 1,849,751. Four personnel-related categories in the Other Expenses schedule—Staff Benefits, Staff Welfare, Staff Training and Recruitment Costs—sum to USD 150,062. The checked public record does not provide a formal bridge showing whether or how those four amounts sit within HR or another financial-statement line. It would be unsafe either to add them automatically to HR or to assume they are already included. That ambiguity is exactly what an allocation note or reconciliation schedule should resolve.

Other is not election, and it is not legal

AFRINIC separately disclosed 2025 election costs of USD 1,043,425 and legal costs of USD 877,929. Neither amount belongs in the USD 854,266 Other Expenses schedule. The categories can be put beside one another for scale: their arithmetic sum is USD 2,775,620, equivalent to 54.27% of the unaudited summary’s total operating expenses and 44.32% of its revenue. But the calculation does not merge their evidence, approvals, suppliers, purposes or editorial subjects.

This separation prevents two common errors. The first is arithmetical inflation, in which a reader describes USD 854,266 as though it includes election or legal expenditure and then adds one of those costs again. The second is narrative contamination, in which allegations or concerns attached to litigation or election administration are allowed to colour routine-looking technical lines without evidence. This article does not evaluate election suppliers, an election rerun, individual legal providers, legal spend per member, or the lawfulness of those separate commitments. Those subjects require their own records and analysis.

Scale context is still useful because it places expectations on governance. A year containing substantial other, election and legal categories deserves disciplined separation of purpose, authority and assurance. One schedule should not be defended by saying another was more expensive, and one controversy should not be used to condemn a different schedule. Each commitment needs its own date, decision-maker, instrument, counterparty, output and remedy.

The service question beneath the accounts

AFRINIC’s technical role gives several larger lines an intelligible purpose. Remote sites and computing can support the continuity of RDAP, WHOIS, RPKI, the Internet Routing Registry, reverse DNS, registration systems and member services. Communications can connect staff and operations. Insurance can transfer defined risks. Banks enable collections and payments. Participation in ICANN and NRO structures can support coordination across the number-resource system. These are plausible pathways, not demonstrated outputs.

The missing evidence is practical. For remote sites, members need a protected inventory, cost by function, uptime, failover, redundancy and disaster-recovery results. For computer expense, they need the split among equipment, software, cloud and security; an approved plan; useful lives where relevant; and a service result. For bank charges, they need transaction volumes, currency structure, categories of fees, concentration and evidence of periodic benchmarking. Staff-related costs need policy, eligible population, approval and an explanation of the change. Training needs attendance and objectives.

Institutional contributions need the assessed basis, invoice and benefit. Professional fees need scope, selection and deliverables.

None of this requires publishing credentials, personal records, full bank details or commercially sensitive contract terms. A public cost-to-output schedule can aggregate securely. It can identify the category, budget, actual cost, variance, counterparty concentration band, responsible service owner, procurement route, deliverable and result. Confidential appendices can carry invoice-level evidence for authorised reviewers. The choice is not between secrecy and dumping every record onto the internet. It is between accountable design and an account title that asks members to infer the rest.

The value of that design becomes clearest when a cost rose. If Computer Expenses increased because AFRINIC restored a delayed security programme, the organisation should be able to show the approved plan and the resilience gained. If Remote Site Expenses increased because more redundancy was needed, availability and failover results should demonstrate the benefit. If Bank charges reflect unavoidable currency and correspondent costs, a volume-and-rate explanation can establish that.

If Staff Welfare changed because a documented policy was applied consistently in an exceptional transition, the policy and approval surface can be recorded without compromising individual privacy.

Positive evidence must be allowed to produce a positive conclusion. Accountability is not a presumption that higher spending is bad. It is a method for distinguishing necessary continuity from agency cost. The public record presently leaves that distinction open.

A control chain divided by office and date

The authority question cannot be answered by attaching one institutional name to the entire year. AFRINIC’s later records describe a governance vacuum beginning in June 2022, followed by the appointment of Gowtamsingh Dabee as receiver on 12 February 2025 in place of the Official Receiver. His published notice stated the aim and mandate of securing company assets and seeing the Board reconstituted under the Bylaws. It did not publish spending thresholds, bank mandates, purchasing rules or a category-by-category delegation.

AFRINIC announced a Board election result on 12 September 2025 and then presented the Board as reconstituted. A joint Board-receiver statement dated 13 October said the receiver had sought termination of receivership on 8 October, that formal discharge remained pending and that the parties were collaborating. It announced a Finance Committee chaired by Carla Sanderson and an Audit Committee chaired by Fiona Asonga. The same statement said the receiver had appointed Forvis Mazars for the 2022–2024 audits under exceptional circumstances without an annual general members’ meeting. These are documented institutional acts and claims.

They do not identify who approved any one of the sixteen 2025 categories.

Timing therefore matters twice: at commitment and at payment. A purchase ordered while the receiver occupied the central continuity position may have been invoiced or paid after the Board was announced. A service approved late in the year may have crossed the other direction in accounting treatment. The valid office and legal instrument at each date cannot be inferred from the fact that the cost appears in a full-year schedule. Later Board review may provide oversight, correction or a separate legal act. It should not be rewritten as contemporaneous approval if the record does not show that.

Operational management also has a distinct role. Finance and service owners should define needs, maintain accounting records, code transactions, explain variances and verify delivery. Yet management’s preparation of a schedule does not create corporate authority for the commitment. Similarly, an authorised signatory’s release of a payment does not prove that procurement was competitive or that the service was received. Control is a chain because each link answers a different question.

Oversight is not management, and audit is not authority

The Audit Committee’s published charter supplies a general architecture. It places statement integrity, risk, controls, unusual transactions, external audit and recommendations to the Board within the committee’s oversight. It also says the committee must not perform management functions. That boundary protects both accountability and accuracy: the committee should challenge, review and recommend, but it should not be described as having ordered every computer purchase or paid every remote-site invoice.

The public record checked does not include agendas, minutes, an adjustment log, challenged categories or recommendations showing what the Audit Committee actually did with the 2025 Other Expenses schedule. A charter proves an expected responsibility, not its performance in a specific case. The Finance Committee was announced, but the checked record supplies no public charter, variance paper or recommendation for this envelope. Its existence cannot fill the gap by assumption.

The Board occupies another surface. Mauritius company law places duties on the Board to ensure compliant financial statements are completed, dated and signed on its behalf by two directors, and to prepare and send the annual report, statements and auditor’s report to shareholders. The audited 2025 statements record approval for issue on 22 May 2026 and a two-director signature surface associated in the sealed research with Adewale Emmanuel Adedokun and Fiona Makokha Asonga. That is statement-level accountability in 2026.

It does not reveal the contemporaneous approval of every commitment made during 2025, particularly before the Board was reconstituted.

Forvis Mazars LLP served as external auditor of the 2025 financial statements. The sealed record identifies an unmodified opinion with an emphasis of matter concerning ongoing legal proceedings. An external financial audit can provide important assurance about the statements within its scope and materiality. It is not automatically a forensic investigation, procurement review, performance audit, legal opinion on receiver delegations or transaction-by-transaction certification of value. The auditor does not manufacture the corporate authority of the person who committed the company.

Members then occupy their own oversight surface. AFRINIC Bylaw 7.6 allows eligible Registered and Resource Members to consider and adopt financial statements by Ordinary Resolution, receive the auditor’s report, consider the annual report and appoint the auditor. Those are genuine corporate rights. A vote does not give a member access to an invoice that was never disclosed, and a general accounts resolution should not be presumed to approve each transaction at its original date.

The proposed accounts vote cannot be treated as an outcome

On 3 June 2026, by order of the announced Board, AFRINIC’s company secretary issued notice of an annual general members’ meeting for 25 June. Ordinary Resolution VI proposed approval of the 2025 audited financial statements together with the directors’ and auditors’ reports. At the evidence cutoff, the checked public meeting page still provided the notice, but no located certified result or signed minutes established whether the resolution passed, failed, was deferred or was otherwise disposed of. The proposal is evidence of a proposed act, not evidence of the act’s completion.

Nor can the legal effect of a hypothetical approval be declared in the abstract. Section 180 of the Mauritius Companies Act addresses ratification where a director or Board purported to exercise a power vested in shareholders or another person, allowing the principal to ratify in the manner in which that power may be exercised. The located general resolution is not shown to be a transaction-specific section 180 instrument. It does not identify the sixteen commitments, the office whose power was purportedly exercised or the alleged defect. Its outcome is also unknown. It would therefore be wrong to say it automatically cured missing authority.

It would also be wrong to say section 180 could never have legal effect. That conclusion would require a particular act, principal, instrument and legal analysis not available here.

NRS’s 24 June member alert is first-class accountability material on this point. It urged members to demand transaction-specific authorisation records and to avoid allowing an accounts vote to become an unexamined waiver, release, ratification or cure. Its detailed monetary case concerned legal spending, not the USD 854,266 Other Expenses schedule. NRS must not be described as the auditor of these sixteen categories, as a court, as AFRINIC’s Board or as the operator of registry services.

Its proper role here is the member-accountability test it actually articulated: separate acceptance of financial reporting from surrender of questions about authority.

What the public record closes—and what it leaves open

The sealed sources close several factual questions. They establish the stated USD 854,266 total, the sixteen webpage categories, the USD 854,267 line sum, the one-dollar difference, and the lower Members Training figure in the unaudited summary that closes the schedule. They establish the USD 654,480 comparator for 2024 and the bounded ratios used here. They establish that election and legal costs were separately stated. They also establish the receiver notice, the later joint statement, the general corporate reporting framework, the Audit Committee’s charter and the proposed wording of Ordinary Resolution VI.

They do not close the transaction ledger. No public general-ledger extracts or account codes were located for the sixteen categories. There is no vendor schedule, invoice set, purchase-order file, contract register, acceptance record, credit-note account, tax treatment or payment confirmation. There is no formal bridge into the audited statements, no published allocation rule for the personnel subcategories and no correction notice for Members Training.

The checked record does not supply the approved 2025 budget, variance papers, receiver directions, bank mandates, Board resolutions, Finance Committee papers or Audit Committee minutes tied to these lines.

Service outcomes remain similarly open. The record does not show site-by-site uptime, redundancy or disaster-recovery gains; a computer asset, software or cloud inventory; banking volume and fee drivers; insurance coverage and claims; staff-policy bases; training attendance; or institutional-contribution deliverables. No independent value-for-money, procurement, forensic or conflict finding was located. The final certified result and signed minutes for Ordinary Resolution VI were not located, nor was a transaction-specific ratification instrument.

These gaps limit both criticism and defence. The evidence does not establish crime, corruption, fraud, improper payment, self-dealing, breach, waste, overpricing or unlawful expenditure. It also does not establish value for money, correct procurement, absence of conflict, valid authority or service delivery for every line. “Unproven” must work in both directions.

Why the bookkeeper standard is the right standard

AFRINIC’s practical leverage comes from maintaining a regional number-resource registry and related coordination services. That leverage does not make it a sovereign. It has no independent power to tax operators, punish them or convert number-resource records into political dominion. Its legitimate centre is narrower: preserving uniqueness, accuracy, transfer records, security assertions and technical continuity through a dependable ledger and associated services.

The cost question follows from that institutional fact. Members fund a service on which network operations can depend. A mandatory or practically unavoidable charge cannot be justified merely through institutional importance, “community” language or claims of stewardship. The organisation must show that the cost maps to its narrow function, that supplementary activity is separately justified, that the amount is proportionate and that a responsible office can answer for the decision. Official AFRINIC publications prove what AFRINIC disclosed, proposed and said it did.

They do not, by their official character alone, prove legitimacy, assent, authority or value.

That is why the USD 854,266 schedule is a useful governance object even though it is not AFRINIC’s complete operating budget. It is small enough to examine category by category and large enough to reveal the agency problem. Members and operators bear the economic incidence, while insiders possess the contracts, allocations and delivery evidence. An aggregate label leaves the party spending the money with more information than the party funding the institution. A cost-to-service and authority ledger narrows that gap.

The standard is not that every dollar outside a packet-forwarding function is illegitimate. A reliable registry needs people, controls, insurance, banking and coordination. The standard is that institutional expansion cannot borrow necessity from the core ledger. Essential continuity should be demonstrated with hard service evidence. Broader engagement, training or institution-building should be identified and justified on its own terms, and, where practical, separated from compulsory funding. AFRINIC earns confidence by evidence, not by asking members to treat its own status as the answer.