An island operator preparing for a new submarine-cable landing does not start with a theory of internet governance. It starts with a cabinet, a power contract, a wet segment, a backhaul quote and a board paper. The new system promises a different commercial life. Wholesale bandwidth should be cheaper. Latency to Europe, the Gulf, South Africa or West Africa should fall. A content cache that once looked uneconomic may become plausible. A bank may move more of its processing onshore. A public agency may ask whether disaster-recovery traffic can remain inside the region. A data-centre developer may discover that the landing station is not only a telecom site but a bargaining chip in a future hosting market.
Then the address question arrives. The operator can buy capacity, but can it move customers without renumbering them? It can reach a cable station, but can it announce portable IPv4 space from more than one upstream? It can sell local breakout, but can it support reverse DNS, abuse contacts, RDAP visibility and route-origin evidence that counterparties will trust? It can lease or acquire scarce IPv4 addresses, but can it show a clean chain of recognition through AFRINIC's records? It can promise resilience, but will a registry dispute, account hold or uncertain transfer process make the addresses less portable than the cable?
That is the economics of submarine-cable and address risk. A cable creates capacity. Registry recognition converts capacity into bargaining power. The two are often discussed separately: cable policy as physical infrastructure, address policy as number administration. In edge and island markets they meet. A landing point changes where traffic can localise, where caches and data centres appear, who can sell enterprise services, who can back up public systems, who can escape a single upstream and who needs portable address resources to do any of that without asking an incumbent for permission.
AFRINIC is the decisive institutional setting for this problem because it is the Regional Internet Registry for Africa and parts of the Indian Ocean. Its own public description covers IPv4, IPv6 and autonomous system numbers, with services around WHOIS, RDAP, reverse DNS, Internet Routing Registry functions and RPKI. Those services are not decorative beside a cable. They are the record layer that lets new physical capacity become usable market capacity. AFRINIC also sits in a region where IPv4 is scarce, island and coastal routes remain concentrated, data-centre demand is uneven, and public materials through recent years have described registry governance as contested through litigation, receivership, election disputes, recovery claims and intervention by global coordination bodies.
The question is not cable engineering as such. It is the institutional economics of the moment after a cable lands. Bandwidth scarcity may fall, but address scarcity remains. Route diversity may improve, but route independence still depends on trusted prefixes. Local hosting may become attractive, but the hosting customer still asks whether its address records will survive disputes, outages and provider changes. The policy conclusion is therefore narrow: the registry should behave as a neutral continuity ledger. It should preserve last verified status during outages and disputes, make transfer and leasing evidence predictable, give non-discriminatory access to registry services, and keep physical-network politics separate from address recognition.
The central distinction is between reach and control. A cable gives a market reach: more bits, more paths, more possible services and more plausible local placement of traffic. Address recognition gives operators control: the ability to decide which upstream to use, where a customer moves, how a public endpoint survives a fault, and whether scarce IPv4 can be redeployed without starting a political argument over permission. Where reach improves faster than control, the gains from the cable are captured by the parties already controlling addresses, backhaul or customer identity. Where reach and control improve together, the cable can reduce dependence rather than merely moving dependence to a cheaper route.
Cable capacity creates a new market before it creates a new balance sheet
The economic story of a submarine cable usually begins with cost per megabit. That is understandable. A market that once depended on satellite links or a single ageing international path changes when a high-capacity fibre system arrives. Wholesale price pressure improves, latency falls and carriers gain another route to sell. Governments announce digital-development ambitions. Investors recalculate the case for data centres, call-centre services, cloud on-ramps, security operations, content caches, research networks and public-sector hosting.
But a cable does not automatically create a balance-sheet asset for every local operator. It creates an option. Turning that option into revenue requires a stack of complementary assets: landing rights, power, ducts, metro fibre, cross-connects, backhaul, exchange points, carrier-neutral facilities, commercial contracts, service-level commitments, public trust and usable internet-number resources. The cable may be the most visible part of the investment, but it is not the only scarce input.
This distinction matters in African and Indian Ocean markets because the physical geography of capacity has historically been uneven. A coastal city with multiple wet systems is not the same market as an island with one or two routes, a landlocked country buying capacity through neighbours, or a secondary coastal town connected by backhaul to a distant landing station. The landing map affects who can bargain. A cable that lands in Lagos, Mombasa, Djibouti, Cape Town, Dakar, Accra, Abidjan, Luanda, Maputo, Dar es Salaam, Port Louis or Victoria changes the local opportunity set. It does not eliminate the need for addresses that customers can use without being tied to one carrier.
The operator's board paper therefore has two columns. The first is physical. It asks whether capacity is available, how many routes exist, which carriers control the landing station, how diverse the backhaul is, what protection paths cost, and whether repair time or regulatory delay could strand traffic. The second is recognition. It asks whether the network can use its own ASN, announce prefixes from multiple upstreams, create or update routing records, keep reverse DNS aligned with customer migrations, maintain abuse contacts, issue or change route-origin evidence, and lease or transfer IPv4 space without a cloud over title-like reliance.
Only the combination turns cable capacity into bargaining power. A carrier with access to a new submarine system but no clean address position may still have to sell through someone else's addresses or accept a bundled upstream arrangement. A data-centre operator with power and racks but no stable address plan may attract local content but struggle with enterprise migration. A bank may like lower latency but resist moving critical workloads if address reputation, reverse DNS or provider exit remains fragile. A public agency may want domestic disaster recovery but need a record that survives institutional friction.
This is why cable economics cannot stop at the landing beach. The cable lowers one bottleneck. It reveals the next.
Submarine cables reduce bandwidth scarcity, not address scarcity
Submarine cables attack the scarcity of international transmission. IPv4 scarcity is different. It is not a shortage of fibre pairs, wavelengths or wet-plant design. It is the exhaustion of a 32-bit address space that remains embedded in customer networks, enterprise systems, security controls, hosting contracts, content-delivery arrangements and public procurement. IPv6 may be abundant in arithmetic, but dual-stack operation and IPv4 reachability remain economic facts. A cable can multiply bandwidth while leaving every operator in the same competition for usable IPv4 resources.
The difference is visible in procurement. A carrier can buy a wavelength, indefeasible right of use or wholesale transit contract on a new cable system. It can then order routers, optics and cross-connects. Address resources are not procured in the same way. If the operator needs additional IPv4 space, it may depend on a small remaining allocation, a transfer, a lease, an acquisition, a customer bring-your-own-address arrangement, or more intensive sharing through NAT. Each route has a different risk profile. Each depends on registry records and counterparty recognition.
AFRINIC's exhaustion position sharpens the problem. Its public policy materials record the soft-landing logic for a constrained IPv4 pool, and public reporting described the remaining unallocated pool as limited relative to likely demand. The exact number will move as applications are processed, but the economic point is stable. The region is no longer in an allocation era where new physical capacity can be matched by generous new address supply. Cable investment arrives in a post-exhaustion market.
That changes the value of address mobility. When a new cable lowers the cost of serving customers locally, the operator wants to move traffic closer to the landing point. It may want to host enterprise servers onshore, put security appliances near users, sell cloud-adjacent services, or bring content caches into the country. Each service needs address continuity. If customers must renumber, update allowlists, change reverse DNS, rebuild geolocation, alter mail reputation or rewrite firewall rules, the cable saving is partly consumed by migration friction.
Address scarcity also changes who captures the surplus from the cable. If portable IPv4 space is available and registry-recognised, more operators can compete to use the new route. If addresses remain tied to incumbent carriers, the incumbent can convert the cable into a defensive asset. It may offer cheaper bandwidth only inside a bundle that keeps customers on its prefixes. The customer receives capacity but not independence. The market looks more connected while the exit right remains weak.
This is not a criticism of cable investment. More submarine capacity is a necessary condition for lower costs, better redundancy and local digital growth. It is simply not a sufficient condition. The bandwidth bottleneck and the address bottleneck are different scarcities. Removing the first makes the second more visible, because businesses that could not previously justify local hosting, multi-homing or disaster recovery now can. They arrive at the registry layer precisely when the free IPv4 pool is small and the market value of clean recognition is high.
The institutional risk follows. A registry that treats address recognition as a neutral record enables more actors to exploit the cable. A registry that treats recognition as a discretionary permission over use, geography or commercial model can make the new cable feel less like an open road and more like a toll road with uncertain entry rules.
Landing geography decides where bargaining power can move
A cable landing is a change in geography. Before the landing, a small market may buy international capacity through a distant hub, a single incumbent or a neighbouring state. After the landing, the same market may have a plausible local route to global networks. That can change bargaining power in three directions: between local access networks and upstream carriers, between local hosting firms and foreign data centres, and between customers and the provider whose addresses they currently use.
The change is not automatic. Landing geography creates the possibility of localisation. It does not guarantee it. If the landing station is controlled by one carrier, if backhaul is expensive, if ducts are unavailable, if power is unreliable or if cross-connect access is discriminatory, the cable can reproduce concentration at a lower technical layer. Even when the landing station is open and backhaul is competitive, address recognition determines how many operators can turn physical reach into independent service.
Consider a coastal operator that previously bought transit from one large carrier because that carrier controlled the cheapest international path. A new cable lands nearby and another wholesaler offers capacity. The operator can now multi-home, but only if it can originate its own prefixes or obtain stable authorisation to originate leased space. If its customer base uses provider-assigned addresses from the old carrier, moving traffic means renumbering customers. If its own prefix records are stale, contested or hard to update, the new upstream may demand indemnities or refuse the announcement. The cable has changed the route map, but not yet the market map.
The same applies to landing-station-adjacent data centres. They do not compete only on power, cooling and latency. They compete on customer migration risk. A bank, platform, enterprise group or public agency moving workloads into a local facility wants addresses that can remain stable across providers. If the data centre's connectivity story depends on addresses borrowed from one upstream, the tenant inherits a dependency. If the data centre can support clean BYOIP, portable address use, predictable reverse DNS and route-origin evidence, the tenant gains exit. That is a commercial difference created by registry recognition.
Landing geography also affects secondary markets. A new cable can make previously marginal IPv4 holdings more valuable because local use opportunities expand. A block held by an operator in a market with poor international connectivity may have limited local revenue potential. Once capacity improves, the same block can support hosting, enterprise VPNs, security services, content platforms or wholesale connectivity. Transfer and leasing clarity therefore becomes part of the cable's economic environment. The more predictable the registry evidence, the more the market can reallocate addresses toward new demand. The less predictable it is, the more addresses remain trapped, underused or bundled with incumbents.
This is where AFRINIC's institutional setting matters. Official registry language often describes distribution and management of number resources. In a cable market, those words have distributional consequences. A narrow ledger lets geography alter competition. A broad gatekeeper can freeze geography into a licensing problem. If the registry makes address mobility costly or uncertain, physical capacity may arrive but bargaining power may stay where it was.
The policy lesson is that landing geography should not be allowed to become address-recognition geography. A prefix used from a new coastal or island route should be judged by verifiable authority, uniqueness, contactability and security evidence, not by an informal preference for which carrier, market or political story the registry likes.
The same lesson applies to the relationship between cable landings and local exchange. A new wet system can make a nearby meeting point more attractive because content, regional carriers and enterprise networks have more reason to locate traffic in the landing market. But exchange fabric does not remove the need for portable address identity. If members arrive with provider-bound addresses, the local route may lower traffic cost while leaving commercial dependence intact. If members arrive with recognised independent resources, cable, backhaul and exchange together create a real exit option. The physical facility and the registry record therefore act as complements. Weakness in either one limits the economic effect of the other.
Island markets expose the double concentration problem
Island markets show the coupling most clearly because they face concentration twice. The first concentration is physical. A small island may depend on one or a few submarine systems, a limited number of landing stations, a narrow set of backhaul routes, and a repair environment affected by distance, permits, weather and cable-ship availability. The second concentration is institutional. The same operator may also depend on one registry account, one upstream's addresses, one reverse-DNS arrangement, one route-origin story and one relationship with the address-recognition layer.
AFRINIC's service region includes parts of the Indian Ocean such as Mauritius, Seychelles, Madagascar, Comoros and Reunion, as well as coastal economies whose traffic often passes through a small number of landing stations. These markets are not identical. Mauritius has more international connectivity and hosting ambition than some smaller islands. Madagascar, Seychelles and Comoros have different market structures, cable paths and operator bases. But the economic problem is shared: when the number of physical paths is small, the value of address independence rises.
The reason is simple. If a network can change upstreams without changing customer addresses, it has a resilience tool. If it cannot, a physical outage becomes a commercial lock-in. A carrier may be able to reroute traffic over a second cable, but customers tied to that carrier's address space cannot easily move to another provider. A data-centre customer may want a backup transit provider, but its application stack may be bound to addresses that the incumbent controls. A public agency may want emergency connectivity through a satellite or alternate cable route, but its public endpoints, reverse DNS and allowlists may depend on one address arrangement.
In a large continental market, these frictions are still expensive. In an island market, they can be existential. A hospital, payment switch, customs platform, airport system or government portal may not be able to wait for renumbering during a cable fault. A hotel group or offshore service provider may lose revenue quickly if reachability degrades. A small ISP may not have the cash to maintain standby capacity across multiple routes unless it can preserve customers across them. Address portability is therefore not a luxury. It is part of disaster planning.
The March 2024 West and Central African cable disruptions illustrate the wider point even though the worst effects were not confined to islands. Public network reports observed disruptions across multiple African countries and pointed to failures affecting WACS, MainOne, SAT-3/WASC and ACE. Reporting on the incident also noted route-announcement and nameserver effects around MainOne. The episode was a reminder that a cable fault is not merely a wet-plant event. It can surface as routing changes, DNS reachability, cloud capacity pressure, rerouting congestion and customer-visible service risk.
For island and edge markets, registry uncertainty compounds the same dynamic. If the registry record is trusted, a network can prepare alternate origin, backup transit, emergency reverse-DNS changes and customer migration paths before the outage. If the registry record is contested or hard to change, physical concentration is joined by recognition concentration. The operator may have a second route in theory but not a second recognised operating posture in practice.
The double concentration problem should discipline registry policy. A registry serving island and edge markets must treat continuity as the default. During cable faults, litigation, account disputes or institutional transitions, the last verified address status should not become a bargaining chip. The more physically concentrated the route map, the more boring the ledger must be.
There is a second island lesson that applies beyond islands. Small markets experience concentration as fixed cost. They cannot easily justify duplicate international capacity, spare addresses, multiple legal reviews, standby operations in several cities and large engineering teams. Their resilience must therefore come from simple, standardised rights to move and announce resources. If each continuity measure requires bespoke negotiation, the market has designed resilience only for the largest networks. A registry that supplies predictable evidence reduces fixed cost. A registry that expands discretion turns every backup plan into a custom case, which is another form of dependence.
Backhaul turns a landing into a network economy
A submarine cable lands at a station, but an internet economy does not live at the beach. It lives in the backhaul. Capacity must move from the landing station to metro networks, carrier hotels, IXPs, data centres, mobile cores, enterprise districts, universities, government sites and neighbouring countries. Backhaul decides whether the cable produces competition or simply gives the landing-station operator a stronger wholesale position.
Backhaul also changes the address problem. If local access networks can reach the landing station through diverse terrestrial routes, they can consider using their own prefixes more aggressively. If only one carrier controls the backhaul, smaller networks may have to buy transit from that carrier and use its address space. The customer's internet identity remains attached to the path owner. In that setting, a cable landing can lower the incumbent's cost without lowering the customer's switching cost.
The economics is similar to a port. A new harbour can bring larger ships, but inland roads, customs procedures, warehousing and title documents decide whether the hinterland benefits. In the internet version, ducts and metro fibre are the roads; IXPs and carrier-neutral data centres are the warehouses; registry records are the title documents. A container that cannot be proved to belong to the consignee is not freely tradable. A prefix that cannot be proved to be held, authorised or portable is not freely usable.
AFRINIC's record functions are therefore part of the backhaul economy even if they sit outside the physical network. A rural ISP connecting to a coastal landing through a wholesale backhaul provider may need its own ASN and prefixes to avoid being absorbed into a national incumbent's address plan. A regional enterprise carrier may need clean records to announce the same customer block from two cities. A university network may want to preserve research addresses while changing transit providers. A public cloud on-ramp may need customer bring-your-own-address evidence. Each use case depends on more than a fibre pair.
This is not a repetition of general interconnection economics. The starting point is physical geography. A cable changes the possible traffic map; backhaul determines how far that possibility travels inland; registry recognition determines whether the new routes carry independent commercial identities or merely extend old dependencies. The sequence matters. Without the cable, the address issue may remain latent because there is no cheap alternative path. With the cable and backhaul, the address issue becomes the constraint that decides who can use the new path on equal terms.
The distributional stakes are high in landlocked and secondary markets. A landlocked operator may depend on coastal landing countries for international access. If it also lacks portable address resources, it is exposed to both transit geography and address dependence. A secondary city may gain a backhaul route to a cable but still rely on addresses assigned by a national provider. A new IXP may lower local exchange costs but not solve the address plan for members whose customers are trapped in upstream space. Backhaul creates the stage on which address independence becomes economically meaningful.
A credible registry cannot build ducts or repair roads. It can reduce the cost of taking advantage of them. It does that by keeping records accurate, updates predictable, evidence requirements narrow and services available to all recognised holders and authorised users on equal terms.
Local breakout needs portable identifiers
Local breakout is one of the main promises of new cable capacity. Instead of hauling traffic through distant hubs, an operator can keep more traffic close to users, exchange traffic at a local IXP, connect to caches, reach a nearby data centre, or send regional traffic over a cheaper and shorter path. The user sees lower latency and better resilience. The operator sees lower international transit cost. The economy sees a plausible base for local digital services.
But local breakout depends on portable identifiers. A customer can only move traffic closer to home if the network identity attached to that traffic can move as well. Enterprise firewalls, payment processors, fraud systems, mail platforms, remote-access tools, geolocation databases, security feeds and government allowlists often treat IPv4 addresses as continuity anchors. Renumbering is not simply an engineering task. It is a customer-risk event.
This is why provider-assigned address dependence is so costly in markets newly reached by cables. Before the cable, the customer may have accepted the incumbent's addresses because there was no practical alternative. After the cable, the customer may want a second provider, a local hosting site or a disaster-recovery contract. If its addresses belong operationally to the incumbent, the customer must choose between better topology and continuity. The incumbent has not won because its cable path is superior. It has won because the address migration is expensive.
Portable IPv4 space changes the negotiation. A business with recognised addresses can move workloads between upstreams, local hosts and backup sites while preserving external identity. A hosting provider with clean address authority can offer customers a migration path that does not require rebuilding every external dependency. A public-sector platform can design continuity around multiple routes. A content provider can place caches locally without tying them to one carrier's future terms.
Registry recognition is the legal and technical evidence behind that portability. The holder record, RDAP or WHOIS contact, reverse-DNS delegation, route object, ROA and account standing do not only satisfy engineers. They tell customers and counterparties that the move is recognised. Without that evidence, the customer must trust a private representation by the provider. With it, the customer can rely on public facts.
Cable geography increases the value of this evidence because it increases the number of possible moves. A market with no alternative path has little practical use for portability. A market with a new cable, a second landing, an IXP or a carrier-neutral facility suddenly has many possible moves. The value of the address record rises with the value of exit.
AFRINIC's crisis makes this point more than theoretical. Public accounts of litigation, receivership, elections, transfer-policy conflict and institutional recovery have turned registry continuity into a due-diligence question. A customer considering local breakout on AFRINIC-administered space may ask not only whether the route works today but whether the record can be relied on tomorrow. If the answer is uncertain, the customer may stay with an incumbent or foreign provider even when the local cable route is technically better.
The result is a risk premium on localisation. The market pays twice: once for physical redundancy, and again for proof that addresses can move across the redundancy. A neutral registry lowers the second cost. A discretionary registry raises it.
Data centres and caches convert cables into address demand
Cable landings attract data-centre and caching demand because they change the economics of proximity. A cache near users saves upstream bandwidth and improves latency. A data centre near a landing station can sell lower-latency connectivity, regional content distribution, cloud adjacency, public-sector hosting and enterprise continuity. A security company can process logs closer to customers. A fintech platform can reduce round-trip time to local users. A government can argue that critical services have a domestic or regional hosting option.
None of this makes the article a data-centre thesis. The point is narrower: data-centre demand is one way cable capacity becomes address demand. A facility may have power, racks and connectivity, but it still needs an address plan for tenants. Tenants need public IPv4 addresses for legacy applications, VPN concentrators, mail systems, API endpoints, security appliances, load balancers, customer portals and migration bridges. They may also need clean reputation and geolocation. In many cases the facility will support IPv6, but IPv4 remains part of the commercial product.
The address question appears at several layers. The data-centre operator may have its own space. Tenants may bring their own. A cloud partner may require customer-owned or authorised space for a specific product. A CDN may use addresses from a global pool but need local route and geolocation hygiene. An enterprise may want to keep existing public addresses while moving servers from Europe or South Africa into an African landing market. Each arrangement requires reliable evidence of who may use the address, where reverse DNS is controlled, which AS may originate the route, and what happens if the customer leaves.
Cable geography intensifies this because a new landing can change where demand appears faster than the address market can adapt. When Equiano added major west-coast capacity, when 2Africa expanded landing options around the continent, or when older systems such as EASSy, SEACOM, WACS, ACE, MainOne and SAT-3 shaped earlier regional routes, the address effect was not simply more internet users. It was a change in where commercial services could plausibly be hosted. Project narratives speak about capacity, redundancy and digital growth; those claims are facts to be examined, not a conclusion. The address-market question is who has the recognised identifiers to capture any resulting demand.
AFRINIC's position matters because many emerging facilities and edge operators are smaller than the global platforms. A hyperscale platform can bring vast address holdings, compliance teams and private network arrangements. A local data-centre operator, regional ISP or enterprise host depends more heavily on standard registry recognition. If transfer, lease or BYOIP evidence is predictable, smaller firms can compete. If it is opaque, larger players and incumbent carriers have an advantage because they can absorb uncertainty or force customers into their address estate.
Address reputation adds another constraint. A newly localised service may fail if its IP ranges carry poor mail reputation, abuse history, incorrect geolocation or disputed WHOIS/RDAP data. A data centre can buy bandwidth and servers; it cannot instantly buy trust. Registry records, abuse contacts, reverse DNS and visible routing history help build that trust. When those records are stable, new cable-enabled facilities can enter the market more quickly. When they are unstable, the market discounts their promises.
The policy implication is not that AFRINIC should choose data-centre winners. It is the opposite. The registry should avoid choosing. It should provide a neutral evidence layer so that cable-enabled hosting demand can find addresses through transparent transfer, lease, authorisation and continuity mechanisms rather than through informal favour, incumbent bundling or excessive review.
Disaster recovery is an address-continuity product
Disaster recovery is often sold as facilities, backup links and failover routing. In cable-exposed markets it is also an address-continuity product. The customer wants to know that if the main route fails, services can be reached through another path without rebuilding identity. That identity may include public IP addresses, reverse DNS, TLS dependencies, mail reputation, firewall rules, API allowlists, security monitoring, payment processor approvals and public-sector audit records.
A cable outage reveals the difference between capacity redundancy and continuity redundancy. Capacity redundancy means an alternate path exists. Continuity redundancy means the customer's service identity can survive the move. A business may have access to a second upstream, but if its address range is tied to the failed carrier or cannot be originated elsewhere with trusted records, the second path is less useful. A government portal may have a backup data centre, but if DNS, reverse DNS, address reputation and route-origin evidence are not prepared, the failover is operationally fragile.
The March 2024 cable failures showed how physical cable events can propagate across layers. Reports identified multiple affected West African systems and rerouting pressure across the region. Public network and provider accounts described country-level disruptions, reduced capacity supporting African regions and effects on MainOne's own online presence. The notable point for address economics is not that every outage was an address failure. It is that cable failures quickly become questions of routing, DNS, capacity substitution and service identity. A backup route that lacks a clean address story may not deliver business continuity.
Island and edge markets should treat this as a design requirement. Operators should be able to preserve last verified registry status during cable outages. They should have predictable procedures for emergency contact updates, reverse-DNS continuity, ROA changes or alternate-origin evidence where appropriate. They should not discover during a wet-plant fault that an unrelated account dispute prevents routine registry-service access. They should not have to choose between restoring customers and preserving a contested administrative position.
This is where the ledger-versus-gatekeeper distinction becomes operational. A ledger says: here is the last verified holder, here are the authorised contacts, here is the route-origin evidence, here is the dispute flag if one exists, and here are the limited changes needed to preserve continuity. A gatekeeper says: before we preserve or update the record, we may reconsider the broader legitimacy of the holder's business model, geography or commercial use. In a disaster-recovery context the second posture is dangerous. It turns continuity into permission.
The same logic applies to public-sector resilience. Governments investing in local hosting after a cable landing may do so for sovereignty, latency, cost or emergency planning reasons. But if the registry layer is uncertain, local hosting may not feel resilient to auditors or procurement officers. A foreign facility with a globally recognised address plan may look safer than a local facility whose address evidence depends on a contested registry. That is not a victory for digital self-reliance. It is a registry-risk discount applied to local infrastructure.
Disaster recovery therefore provides a hard test for AFRINIC. The question is not whether the institution can make strong public statements about stability. It is whether customers can design failover without treating registry discretion as an additional hazard. In cable markets, resilience is not complete until the address record can move with the traffic.
IPv4 scarcity makes cable timing valuable
Cable timing and IPv4 scarcity reinforce each other. A new cable arrives at a date; a scarce address block can be deployed at that date or miss the opportunity. If the address holder, lessee or buyer cannot obtain registry clarity while the cable-enabled market is forming, the first wave of customers may go elsewhere. Time is an economic input.
This is especially important in small markets. When a landing is new, the first carriers, data centres, banks, public agencies and content providers set expectations. They sign anchor contracts, pick connectivity partners, establish geolocation patterns, deploy caches, write procurement references and educate customers. An operator with clean portable IPv4 space can use that window. An operator waiting for transfer approval, lease evidence, account restoration or contact updates may watch the window close.
Scarcity means the delay is not easily solved by asking for more addresses. In earlier periods a network might have grown by applying for a new allocation. In a post-exhaustion environment it must reuse, lease, acquire or share. That makes registry process part of market timing. If a data-centre launch is tied to a cable-ready date, and the address plan is tied to an uncertain registry review, the whole project carries a delay premium. Investors notice. Customers notice. Competitors notice.
The premium is not only financial. It is strategic. A local operator that misses the first wave may become a reseller rather than a platform. A public buyer may standardise on a foreign host. A cache may be placed in a neighbouring hub instead of the new landing market. An enterprise may keep workloads in Europe or South Africa. Once those choices are embedded, later address clarity may not recover the lost position.
This is why IPv4 transfer and leasing clarity becomes more valuable when physical capacity arrives. The cable increases the productivity of addresses. A /24 that supports a modest local service before the cable may support a larger hosting or enterprise product after it. A /22 that was useful for access customers may become valuable for multi-homed services. A block that can be cleanly leased into a landing-station ecosystem may generate more revenue than one trapped in a low-capacity market. The registry does not create this value. Cable investment and customer demand do. But the registry can delay, discount or unlock it.
AFRINIC's governance crisis raises the stakes because markets already know the registry can become a high-consequence institution. Public reporting has described address-record scandals, the Cloud Innovation dispute, litigation affecting operations, receivership, election turmoil, recovery claims and court involvement. In such an environment, a transfer or lease is not judged only by policy text. It is judged by the probability that recognition will remain stable when the asset becomes valuable.
The economic discipline should be proportionality. Fraud control is necessary. Duplicate claims, forged authority and stolen blocks damage the market. But the review should be tied to verifiable record risk. If a lawful commercial use becomes suspect merely because cable timing makes it profitable, the registry becomes a capital-control institution. It starts rationing the upside of physical infrastructure through administrative judgment.
Markets do not need AFRINIC to guarantee profits. They need it not to create avoidable uncertainty over the record that lets capital meet capacity.
Transfer and leasing clarity become landing-station infrastructure
Landing-station infrastructure usually means power, ducts, colocation, security, meet-me rooms, optical equipment and backhaul. In a post-exhaustion address market, transfer and leasing clarity should be treated as a softer but real part of the same infrastructure. Without it, addresses cannot flow toward the new demand created by the cable.
The word "clarity" matters. A registry does not need to endorse every private contract or act as a broker. It needs to define what evidence is required for recognised use, transfer, lease, sub-allocation, contact update, reverse-DNS change and route-origin publication. It needs to make those requirements predictable across holders and countries. It needs to distinguish fraud risk from dislike of commercial models. It needs to preserve records during disputes rather than using continuity services as leverage.
Leasing is particularly important around cable landings because demand can be uncertain and uneven. A data-centre operator may not want to acquire a large block before tenants arrive. An enterprise may need addresses for a migration period. A content service may test local demand. A public-sector project may need a temporary expansion during a transition. Leasing lets scarce IPv4 capacity move to these uses without forcing immediate long-term purchase. But leasing is only useful if customers, upstreams and registry records can establish who is authorised to use the space and what happens when the arrangement ends.
If AFRINIC treats leasing as inherently suspicious, the market will not disappear. It will move into private representations, foreign intermediaries, incumbent bundles or less transparent arrangements. That raises abuse risk and weakens customer continuity. If AFRINIC treats leasing as an evidence problem, it can improve visibility: authorised user, contact, abuse handling, route origin, term, revocation mechanics and dispute flag. The neutral ledger can make commercial use safer without becoming a commercial judge.
Transfers raise a related issue. A cable landing may reveal that addresses are more productive in one market than another. A predictable transfer environment allows underused holdings to move to new uses, subject to verified holder authority and clean records. An unpredictable environment keeps addresses where they are, often with incumbents or entities that can afford to wait. In a region where physical connectivity is improving unevenly, rigid address mobility can blunt the very development effect cable investors claim to support.
The landing-station analogy is useful because it shows the danger of discrimination. A landing station that offers cross-connects only to favoured carriers undermines competition. A registry that offers evidence, updates or transfer certainty only to favoured business models does the same at the recognition layer. Both forms of discrimination can be hidden behind technical language. Both raise the cost of entry for smaller networks.
AFRINIC's public materials present resource transfers, reverse DNS, WHOIS, RDAP, IRR and RPKI as part of its service surface. The economic task is to make that surface reliable enough for cable-enabled markets. If a new coastal facility can plan its address supply with clear evidence standards, it can sign customers. If every arrangement may become a debate over regional loyalty, need, monetisation or institutional preference, customers will price that risk into contracts or avoid the market.
The practical test is whether a transaction can be explained to a cautious counterparty in one page. Who holds the block? Who may use it? Which AS may originate it? Who receives abuse reports? Who controls reverse DNS? What record proves the arrangement? What happens at expiry, breach or dispute? If those answers are standard, the cable market can move quickly. If each answer depends on a discretionary reading by the registry, the market slows and the risk premium moves to whoever has the strongest lawyers, the largest address inventory or the most entrenched carrier relationship.
Transfer and leasing clarity will not build the cable. It will decide how widely the cable's benefits can spread beyond the owner of the first route.
AFRINIC's crisis adds a registry-risk premium to physical-route risk
Every cable market already has physical-route risk. Cables can be cut by anchors, fishing activity, seabed events, conflict, sabotage fears or accidents. Repairs can be delayed by permits, security conditions, weather and ship availability. Landing stations can face power, flood, access or regulatory issues. Backhaul can fail. Insurance and redundancy cost money. Operators know how to price these risks, imperfectly but routinely.
AFRINIC adds a different risk: registry-risk premium. This is the extra cost attached to address resources when counterparties fear that records, services or institutional authority may become uncertain. It appears as longer due diligence, stronger warranties, lower address valuation, hesitation around AFRINIC-administered space, stricter customer assurance, backup documentation, legal review, or dependence on a carrier whose addresses look simpler.
The premium has factual roots. Public reporting and institutional statements have described a history that includes alleged address-record manipulation, high-value IPv4 disputes, bank-account freezes, years without normal board stability, receivership, election disputes after proxy concerns, board-election and recovery efforts, further litigation and ICANN involvement in proceedings linked to possible winding-up. The NRO's 2023 statement on the appointment of a receiver emphasised preservation of status quo, continued registry services and restoration of functional governance. Public reporting through 2026 described signs of recovery alongside continuing conflict.
A cable investor, data-centre tenant or enterprise customer does not need to adjudicate every claim in that history. It only needs to ask whether the record layer is boring enough to rely on. If the answer is no, physical-route diversification may not receive full credit. A customer may ask: what if the address block becomes disputed? What if reverse DNS cannot be updated? What if an account hold affects RPKI or RDAP? What if a transfer is delayed during a cable-driven launch? What if a public conflict makes a lender uncomfortable? What if an upstream rejects an AFRINIC-administered lease because the evidence is not clean enough?
The effect is uneven. Large carriers and global platforms can manage registry-risk premium through legal teams, address inventories, multiple RIR relationships and private network architecture. Small coastal ISPs, island networks, local data-centre operators and enterprise hosts have fewer substitutes. They pay the premium directly or lose customers to a party that can absorb it. A registry crisis therefore acts like a regressive tax on cable-enabled competition.
This is not an argument that AFRINIC should ignore fraud, contracts or policy. Weak records were part of the region's historical problem. The 2019 address-heist reporting and subsequent disputes show why verification matters. The argument is that verification must reduce the premium, not increase it. It should give counterparties a clearer answer about current recognition, authorised use, disputes and continuity. It should not expand into open-ended discretion over whether the registry approves the holder's commercial strategy.
The registry-risk premium also interacts with capital-control risk. When a registry can delay or deny address mobility through discretionary assessments of need, geography or use, it can trap value inside a jurisdictional or institutional boundary. In a cable market, that can decide who captures the upside of new capacity. It is one thing to maintain uniqueness. It is another to ration the economic benefit of a landing by making addresses harder to move toward it.
AFRINIC's legitimacy will be measured less by official recovery narratives than by whether the premium falls. If operators can use AFRINIC-administered resources in cable-enabled markets without extra fear, the institution has rebuilt part of its authority. If every cable project must carry a registry-risk annex, the crisis remains embedded in infrastructure economics.
Physical-network politics must not become address-recognition politics
Submarine cables are political infrastructure. They require landing rights, permits, coastal access, security assessments, state-owned carriers, consortium agreements, foreign investors, regional diplomacy and sometimes great-power attention. A cable route through the Red Sea, around the Cape, across the Atlantic or into an Indian Ocean island can be read as commercial strategy, security strategy and industrial policy at the same time. That political environment is unavoidable.
Address recognition should not be pulled into it. The registry's job is to preserve uniqueness, accuracy, contactability and continuity of number records. It should not decide whether a particular cable consortium, cloud partnership, national digital strategy, regional bloc or carrier alliance deserves address advantage. It should not use physical geography as a proxy for moral judgement about resource use. It should not treat a lawful address arrangement as more or less legitimate because the traffic will exit through a politically favoured landing.
This separation is especially important in AFRINIC's region because cable geography already creates winners and losers. A country with multiple landings has a different bargaining position from a landlocked neighbour. A carrier that controls a landing station has a different position from a small ISP buying backhaul. An island with one route has different resilience needs from a continental hub. If the registry adds a second layer of preference, physical inequality becomes institutional inequality.
Official cable narratives can encourage this drift. They speak of national transformation, continental digital growth, open access, data sovereignty, local hosting and resilience. These are legitimate public goals. But they can be used to justify address controls that exceed the registry's narrow function. A registry may be tempted to say that addresses issued in the region should serve the region in a particular way, through particular business models, under particular interpretations of development. Once that temptation becomes policy practice, the registry is no longer only recording. It is allocating the gains from cable geography.
The better approach is to treat the address layer as a common ledger for actors with different physical strategies. A carrier using Equiano capacity, a data centre near a 2Africa landing, a public agency using a domestic backup site, a regional ISP buying backhaul through a neighbouring state, and an enterprise using leased space for a migration should all face the same evidence questions: who is recognised, who is authorised, what is routed, who is contactable, what is delegated, what is disputed, and how is continuity preserved?
This does not remove law. If a court orders a specific preservation measure, if sanctions prohibit a specific dealing, if fraud is proven, or if a contract dispute affects authority, the registry must record and apply that fact within proper bounds. But it should not convert cable politics into address permission. The record should remain legible to counterparties who disagree about the politics of the cable.
The distinction also protects public institutions. A ministry may want local hosting after a cable landing. It should not need the registry to favour national projects. It needs the registry to make address continuity reliable enough that local providers can compete for the ministry's business. A regulator may want open access at landing stations. It should not need address controls to achieve that goal. It needs competition policy and facility access rules at the physical layer. Using address recognition to solve physical-network politics is mandate laundering: a narrow registry mandate is laundered into broader economic control.
AFRINIC's institutional legitimacy depends on refusing that expansion. The more politically salient cable geography becomes, the more neutral the address ledger must be.
A neutral ledger lowers the cost of route diversity
Route diversity is expensive. Operators pay for extra ports, cross-connects, backhaul, standby capacity, monitoring, engineering time and operational complexity. Customers pay through higher service prices or continuity fees. The business case depends on whether the extra route actually reduces risk. Registry uncertainty can erode that business case because it leaves part of the dependency untouched.
A neutral ledger lowers the cost of diversity by making address authority portable across routes. If an operator can originate its recognised prefixes over multiple upstreams, update route-origin evidence predictably, preserve reverse DNS and keep contacts current, then physical diversity has commercial value. If each alternate route requires new address proof, renegotiation or registry anxiety, diversity becomes a bespoke project for large firms rather than a standard product for smaller networks.
This matters for exchange points as a consequence of cable geography, not as a separate thesis. AFRINIC's own policy manual recognises IXPs as critical elements for internet economies and reserves resources for IXP peering and management use. Local exchange can reduce unnecessary international routing and improve resilience. But it does not eliminate the need for global address trust. A route server still depends on route objects, ROAs, prefix limits and member evidence. A cache still needs addresses with clean reputation. A member still needs the ability to announce its prefixes over more than one path.
The cable connection is direct. A new submarine system can make an IXP more valuable by bringing content, cloud on-ramps or cheaper upstreams closer. The IXP can make the cable more valuable by distributing capacity among local networks. The registry can make both more valuable by ensuring that address records are stable enough for networks to use multiple routes. If any layer fails, the others lose part of their effect.
The cost of route diversity is also shaped by lender and customer perception. A bank financing a data centre or carrier expansion may ask whether the borrower can keep customers if a primary route fails. The answer includes address continuity. A public buyer procuring resilient connectivity may ask whether a provider can fail over without renumbering. A multinational enterprise may ask whether the same service can be reached from two landing paths. If the registry layer is clear, these questions become engineering and contract terms. If it is unclear, they become institutional risk questions.
Neutrality here is practical. It means preserving last verified status unless a specific, evidenced change is required. It means separating account payment issues from unrelated publication functions where customer continuity would otherwise be harmed. It means giving holders notice and cure where possible. It means recording disputes without disabling ordinary routing evidence beyond what the dispute requires. It means not using RPKI, reverse DNS or RDAP as leverage in commercial conflicts. It means treating similarly situated holders alike.
Such rules do not weaken the registry. They make it more valuable. A registry that lowers the cost of route diversity becomes part of resilience infrastructure. A registry that raises that cost becomes part of the concentration problem. In cable markets, the difference will be visible in who can afford multi-homing, who can sell disaster recovery, who can host locally, and who remains dependent on a single upstream despite new physical capacity.
AFRINIC's region needs more route diversity, not more reasons to distrust it. The ledger should help operators buy diverse paths with confidence that the address layer will not collapse into discretionary review at the moment diversity is needed.
The most useful registry response is therefore procedural modesty. The registry should publish clear evidence categories, keep service channels available, record disputes narrowly, and let operators build diverse physical paths without asking the registry to bless the business case. That modesty is not indifference. It is discipline. It recognises that a registry can do great harm when it tries to solve every market problem through address recognition. Route diversity is created by operators, carriers, IXPs, cable consortia and customers. The registry's role is to keep their identifiers usable while they compete.
The design answer is continuity, evidence and non-discrimination
The design answer is not to make AFRINIC a cable regulator, data-centre planner or industrial-policy body. It is to narrow and harden the registry function so that cable-enabled markets can rely on it. The three principles are continuity, evidence and non-discrimination.
Continuity means the last verified status of a number resource should be preserved during ordinary disputes, cable outages, governance transitions and institutional stress unless a specific legal or technical reason requires a change. RDAP, WHOIS, reverse DNS, IRR and RPKI services should be treated as continuity infrastructure, not optional privileges to be suspended for unrelated leverage. Running networks and downstream customers should not become collateral damage in disputes that can be recorded, ring-fenced and adjudicated without disabling basic recognition.
Evidence means the registry should ask for facts tied to its function: holder authority, authorised user, contactability, route origin, delegation, transfer history, dispute status, fraud indicators and court orders. It should not ask open-ended questions designed to test whether it approves of a business model, customer geography, leasing practice or cable strategy. Where leasing or temporary use exists, evidence can make it safer: term, authority, abuse contact, origin AS, revocation mechanics and customer continuity. Where transfers exist, evidence can make them cleaner: current recognised holder, absence or status of disputes, corporate authority and clear effective date.
Non-discrimination means equal access to registry services for similarly situated holders and authorised users. A small island ISP should not face a higher uncertainty tax than a large carrier. A local data-centre operator should not need political sponsorship to obtain routine record updates. A customer using leased space should not be treated as invisible if the authorised-use evidence is clear. A holder involved in a dispute should have the dispute recorded without unrelated services being impaired beyond necessity. A registry that serves a multi-state region cannot become credible if members believe record access depends on institutional favour.
These principles also separate physical-network politics from address recognition. Landing-station access, cable security, backhaul competition and public hosting strategy should be handled in their proper forums. The registry should not compensate for weak physical competition by rationing address use, and it should not compensate for address scarcity by deciding which cable-enabled businesses deserve growth. Its contribution is to make the address layer boring, auditable and durable enough that other markets can work.
AFRINIC's crisis gives the lesson force. A region with scarce IPv4, high physical-route concentration in many markets, new cable investment, island dependencies, data-centre ambition and contested registry governance cannot afford a registry that behaves like a discretionary gatekeeper. Nor can it afford a registry that ignores fraud and record accuracy. The viable middle is a narrow continuity ledger with strong evidence rules and limited power.
Submarine cables will keep changing Africa's internet geography. Equiano, 2Africa and earlier systems such as EASSy, SEACOM, WACS, ACE, MainOne and SAT-3 are not just capacity projects; they are changes in where economic activity can happen. But capacity is not the same as independence. Independence requires the ability to move traffic, customers and services without surrendering address identity to an incumbent or a fragile institution.
That is the final test. If a new landing lets an operator buy bandwidth but not bargain, the cable has not delivered its full economic promise. If a registry record lets the same operator use portable, transferable or clearly authorised IPv4 resources across diverse routes, the cable's capacity becomes market power. AFRINIC's role should be to make that conversion predictable, neutral and resilient. The cable brings the light. The ledger decides who can use it without asking permission from the gatekeeper.
The institutional stakes are larger than one landing station or one address block. Africa's cable map is becoming denser, but not uniformly so. Some markets will gain multiple landings and carrier-neutral facilities; others will remain dependent on a neighbour, a single wet route or a distant hub. Scarce IPv4 will not be distributed by geography in the same pattern as cable capacity. A credible registry can help those two maps meet by making recognised address use portable, evidenced and continuous. A politicised registry can keep them apart. The economic difference will show up in prices, customer mobility, local hosting, disaster recovery and the ability of smaller operators to turn new capacity into independent business rather than rented dependence.