Summary

  • AS37662 is registered to the West Indian Ocean Cable Company and was reported as announced at a single query window of 16:00 UTC on 17 October 2026. That is a routing fact about a network, not a measurement of any cable's capacity.
  • The public record verified here establishes no design capacity, no lit capacity, no fibre-pair count, no named landing point, no licence reference and no fault or repair interval for any system the company uses. Three system-level documents that would normally carry those figures could not be read as text on 18 October 2026.
  • Deliverable capacity is jointly produced by an owner's upgrade decision, a landing-station operator's access to the shore end, and a national regulator's permission to land and sell. An announcement reports only the first of the three, which is why a route can be simultaneously full in an allocation ledger and empty in a buyer's experience of it.

The chain that has to hold before anything is delivered

Start with the physical sequence, because every institutional claim in this subject sits downstream of it. A submarine system is a length of optical fibre on the seabed with repeaters along it, terminated at each end in a cable landing station. The fibre itself carries nothing usable until terminal equipment in that station lights specific wavelengths across specific fibre pairs.

The station needs continuous electrical power, cooling and security, and it needs a terrestrial backhaul — usually several, ideally over separate paths — to reach a population centre, a data centre or an internet exchange where traffic can be handed to other networks. On top of that sits a commercial instrument: a capacity agreement, an indefeasible right of use, or a lease, which names a route, a quantity, a delivery point and a term. And on top of that sits a legal permission: a licence, authorisation or registration in each jurisdiction where the system lands or where the service is sold.

Each of those is a separate decision, taken by a different party, on a different timetable. The consortium or owner decides when to add terminal equipment and how much. The landing-station operator decides who gets physical access, on what terms, and how much backhaul capacity is available and at what price. The regulator decides whether a given operator may land, operate and sell, and under what conditions. A system can be technically complete and commercially unavailable in the same week.

A buyer who contracts for capacity on a route is therefore not buying a length of fibre; they are buying the conjunction of three decisions that were made by three organisations, only one of which is likely to have issued a press release.

This matters for how announcements should be read. A statement that a system has entered service, or that capacity on it has been increased, is a statement about one link in the chain. It is not a statement that a named buyer can light a specified quantity to a specified landing point on a specified date, because that outcome also required the station operator and the regulator to have moved. When those two are silent, the announcement is true and the delivery is still unproven. Both facts can hold at once.

Five numbers that get used as if they were one

Before any of this can be assessed, the vocabulary has to be pinned down, because the capacity figures that circulate in this industry are not interchangeable and are routinely compared as though they were.

Design capacity is the theoretical capacity of a system's fibre pairs with terminal equipment specified at its most favourable configuration. It is a property of the wet plant and of the specification sheet. It is the number most often quoted in a system's launch material and the number least connected to what anyone is using.

Initial lit capacity is what was actually equipped and commissioned when the system entered service. It is typically a fraction of design capacity, and it is the figure a ready-for-service announcement usually means, even when the announcement does not say so.

Allocated capacity is what the construction and operating agreements assigned to owners, consortium members and indefeasible-rights holders. Allocation is a bookkeeping fact internal to the consortium. A system can be fully allocated and barely lit, or lightly allocated and heavily lit.

Contracted capacity is what has been committed to specific buyers under specific agreements, each with a route, a term and a delivery point. This is the first number that begins to describe a market rather than a machine.

Deliverable capacity is what a named buyer can light to a named landing point on a named date, which requires that the system, the station, the backhaul and the licence all permit it at the same time. This is the only one of the five that answers a buyer's actual question, and it is the one that almost never appears in public.

A figure without a unit, a date, a route, an allocation basis and a contractual scope cannot be compared with a figure that has them. Most of the confusion in public discussion of African submarine capacity comes from exactly this: a design figure from one document being set against a contracted figure from another, with nothing in between to say that they describe different objects.

What the retained documents establish

Three documents in this record carry text, and it is worth being precise about what each one proves, because the discipline of the exercise is that a document proves its own content and nothing adjacent to it.

The first is the routing registry record. The record returned for AS37662 on 18 October 2026 identifies the holder as "West Indian Ocean Cable Company - West Indian Ocean Cable Company", reports the resource as an autonomous system numbered 37662, and returns "announced": true for a query window that opens and closes at 16:00 UTC on 17 October 2026 (RIPEstat AS-overview for AS37662). It also returns the block in which the number sits: 36864–37887, described in the record as "Assigned by AFRINIC", within the IANA 16-bit Autonomous System Numbers Registry.

That establishes three things and no more. The number is registered to this company under that name. The number was being announced at that moment, which means some part of the global routing system was carrying a route originated by it. And the number falls inside a range whose registry entry is administered by AFRINIC. It does not establish how much address space is announced, how many prefixes are originated, where the holder is incorporated, or whether the company owns, leases or resells a single fibre pair anywhere.

The block description is a property of the number range, not a statement about the company that holds a number inside it — an important distinction, because the block is the kind of detail that invites an inference about domicile that the record does not support.

The second is a trade report dated 19 August 2020, which states that the International Finance Corporation announced a USD 20 million loan to the West Indian Ocean Cable Company "to help it expand and improve affordable internet connectivity in Africa" (Telecompaper, 19 August 2020). That establishes the existence of an announcement, its amount, its date and its stated purpose. It does not establish the tenor, the instrument's detailed terms, whether the money was drawn, what it was spent on, or what conditions attached. It is a report of an announcement, and the sentence is truncated at source. The word "affordable" in that purpose clause is worth pausing on, because affordability is a retail price outcome, and a wholesale capacity loan can influence it only through the wholesale price the resulting capacity enables — which brings the analysis straight back to the three-party chain rather than around it.

The third is a page dated 17 September 2026 describing the company as a wholesaler "providing capacity to international telecoms, cloud operators, content providers and internet service providers within and out of" Africa (ijr, 17 September 2026). The page address embeds a Wikidata identifier, which makes it a derived stub of a knowledge-base entry rather than an independent editorial narrator. It counts as one narrator, not two, and it cannot corroborate itself. What it establishes is that a page on that date carried that description. It establishes nothing about which systems, how much capacity, at which landing points or on what terms.

The fourth element is not a document but a directory profile, last updated on 17 June 2026 and published at the BTW directory entry for the West Indian Ocean Cable Company. It records one autonomous system, AS37662; a region label of "Global"; a legal-type label of "private_company"; an entity type of operator; and an interconnection footprint of 3,497 observed relationships, of which 2,751 are peers, 369 are upstreams and 377 are downstreams. The arithmetic closes exactly, and the peer share is 78.7 per cent. The profile rests on a single evidence basis, carries a confidence score of 0.7, and lists its own gaps explicitly: legal entity country, current services, visible assets, parent or subsidiary links, and responsible public leadership contacts.

Three labels now sit side by side: a directory region of "Global", a directory admission that the legal-entity country is unknown, and an autonomous system number inside a range whose registry entry is described as assigned by AFRINIC. None of the three is a domicile finding, and the temptation to combine them into one is exactly the error this record warns against. What can be said is narrower and still useful: the number is administered in an African range, the directory declines to state a country, and the directory's own region facet is not a statement about where anything is incorporated.

The interconnection figure deserves separate interrogation rather than celebration. A peer share of 78.7 per cent describes a network that exchanges traffic with a very large number of other networks directly. For a company that describes itself as a wholesaler selling capacity to carriers, cloud operators and content providers, that profile raises a genuine question: does it reflect broad open peering at shared facilities, an access-facing footprint, a content-handing posture, or some mixture?

The count cannot distinguish between them, because a count of relationships is not a measure of traffic and says nothing whatever about the capacity of any link. It is a shape, not a size. Anyone citing it as evidence of scale is reading a number for something it does not contain.

The evidence boundaries, dated

Three boundaries limit what this assessment can assert, and each is a fact about the public record on a particular day rather than a finding about the company.

On 18 October 2026, three system-level documents could not be read as text: the EASSy consortium's about page (eassy.org/about.html), the company's own EASSy page (wiocc.net/eassy/), and the Wikipedia article on the EASSy cable system (Wikipedia: EASSy). Each returned no extractable text. A failed reading is a limit on verification; it is not evidence that the documents do not exist, that they say something other than expected, or that anyone withheld anything.

On 18 October 2026 no project-disclosure page for the 2020 facility was located in the public record consulted. The committed amount, instrument type, project identifier and contractual purpose beyond the 2020 press report therefore remain unverified. The existence of the facility is established; its terms are not.

On 17 October 2026 three further pages — the company's subsea-network page, the consortium home page, and a second copy of the 2020 trade item — were returned without verbatim text and are therefore not treated as documents read. Summaries of them exist. Summaries are not source text, and nothing in this article rests on them.

These three boundaries should be read together with their prohibition. They do not support any inference about delay, refusal, concealment, disclosure strategy or good faith. A missing figure is a missing figure. The correct response to it is to say what is missing, say when it was missing, and stop.

The control surface: three ledgers that do not reconcile

With the evidence stated, the structural argument can be made.

Capacity on a submarine route is set by three parties who do not share a ledger. The owner or consortium decides when to upgrade, and its announcements report that decision. The landing-station operator controls physical access to the shore end and the terrestrial backhaul behind it, and its agreements are commercial documents between two companies. The national regulator controls the licence to land, operate and sell, and its decisions appear in filings that are rarely indexed against the commercial name a buyer knows.

A wholesaler's deliverable capacity is the product of all three decisions. A capacity announcement reports only the first. That asymmetry is not a conspiracy and does not require anyone to behave badly. Each party is answering its own question: the consortium, whether to invest in more terminal equipment; the station operator, who gets access and at what price; the regulator, whether the licensee meets the conditions. Nobody in that set is tasked with producing the single number a buyer needs, which is why the number is so often absent even when every party is being reasonably forthcoming within its own frame.

The practical consequence is a familiar paradox. A system can be "full" in a consortium's allocation ledger — every owner's share spoken for — and simultaneously "empty" in a buyer's experience of the route, because the capacity allocated to an owner in a landlocked market was never lit to the buyer's landing point, or because the backhaul from that station is committed elsewhere, or because the buyer's own jurisdiction has not authorised the service. Both descriptions are accurate on the same day. They are answers to different questions, and neither party is lying.

This is why the useful unit of analysis is not the system but the route-and-date. A route between two named landing points, on a named date, with a named delivery point and a named licensee, can be checked. A system, in the abstract, cannot.

The conditional market implication

Only now does the market consequence follow, and it follows conditionally.

African internet service providers, mobile operators and data-centre developers commit to capacity on the assumption that announced system capacity converts into deliverable bandwidth at a predictable price and on a predictable date. Where the upgrade schedule, the rights-of-use terms and the backhaul arrangement are not public, the buyer is pricing an option it cannot verify. That does not make the purchase wrong — most of the time the capacity does arrive — but it means the risk is carried silently until an outage, a delay or a price change reveals which of the three decisions was actually binding.

The second-order effect is in restoration and diversity. If a buyer's route depends on a single system and no published restoration capacity exists, then outage risk is not only unpriced but correlated: every buyer on that route carries the same exposure at the same time, and discovers it together. Published repair intervals and published restoration arrangements are the instruments that would convert that correlated risk into something a procurement team could price. Where they are absent, the market is not necessarily worse off in expectation; it is worse off in information, and it will find out which at the least convenient moment.

The third-order effect is on procurement practice rather than on any single purchase. A buyer who asks for a delivery point, a licence basis and a restoration commitment is asking three questions the seller may not be able to answer from a single document, because the answers live in three organisations. The buyers who ask will get better terms. The ones who accept a system name and a capacity figure will get a contract that describes the first decision and leaves the other two to be discovered.

What would change this assessment

The gap here is narrow and closable, which is worth saying plainly. Four kinds of document would materially alter the picture: a capacity price list naming delivery points, quantities and terms; a landing-station agreement or a regulator filing naming the licensee and the conditions in a specific jurisdiction; a consortium statement of ready-for-service or upgrade capacity carrying a figure, a date and a landing point; and a fault or repair log with intervals and any restoration capacity provided. Any one of those would move the analysis from a shape to a size.

A single result would falsify the working thesis: a published, buyer-facing schedule showing announced capacity converting to deliverable capacity at named points, on named dates, at published prices, with the licence basis stated. If that exists and can be read, the three-ledger problem is an inconvenience rather than a constraint, and this article should be revised accordingly.

Uncertainty, stated

What remains unknown is substantial and should be listed rather than smoothed over. Which submarine systems the company holds capacity on, and in what role — owner, consortium member, indefeasible-rights holder or reseller — is not established by any document in this record. Neither are design capacity, fibre-pair count or initial lit capacity for any system, with the date on which each figure was true. No landing point is named, no landing-station operator is identified, no backhaul arrangement is described, and no licence reference appears.

No rights-of-use or capacity-agreement terms are public: no term, no delivery point, no restoration obligation. No cable-fault or repair record with intervals has been found. The legal-entity country, the ownership structure and the current service catalogue are all listed as gaps in the directory record itself.

The relationship between the announced autonomous system and any submarine capacity is likewise unestablished. A network number being announced tells you that routing exists; it does not tell you what that routing carries, and in a wholesale business the two can be quite separate.

What can be said with confidence is smaller and firmer than the questions. The company holds an announced autonomous system in an African-administered number range. It described itself, through a 2020 trade report of a lender's announcement, as a recipient of development finance for African broadband. It is described, through a derived stub, as a wholesaler selling capacity to carriers, cloud operators, content providers and service providers. And the documents that would turn any of that into a statement about deliverable capacity are, as of 18 October 2026, not readable in public.

For a business whose product is a promise about bandwidth at a place and a date, that is the finding: the promise is sold, and it is not published.