Lu Heng’s elegant theory of registry power misses the harder truth. AFRINIC is under strain not only because the RIR model has weaknesses, but because African public resources have been pulled into a global market and defended through saturation litigation.

On 9 March 2026, AFRINIC’s public case list added another entry to a docket that long ago stopped looking like ordinary corporate litigation. The new matter, Skyconnect v AFRINIC & Anor, SC/COM/PWS/000132/2026, challenges the AFRINIC board’s ratification of the Number Resources Transfer Policy. Taken in isolation, that filing would be only another plaint before the Commercial Division of the Supreme Court of Mauritius. Read against the public record of the last 5 years, it looks like something larger. It looks like one more front in a war of attrition against Africa’s Regional Internet Registry.

That is why Lu Heng’s latest essay deserves a serious answer. It is intelligent, polished and, in its own way, dangerous. Its prose is calm, juridical and soft-spoken. It speaks of liability caps, monopoly gatekeepers and structural contradiction. The softness matters. In governance conflicts, tone can be an instrument. It can launder extraction as theory. It can take a concrete history of scarcity, arbitrage and litigation and repackage it as an elegant meditation on institutional design.

The deepest problem with Lu Heng’s essay is not that it raises structural questions. Some of those questions are real. The deepest problem is that it uses structure to dissolve agency. It asks the reader to look upward, toward abstractions, precisely when the public record demands that we look downward, toward contracts, case files, registry obligations and a continent whose digital public goods have too often been treated as though they exist for someone else’s balance sheet.

This is not the first time Lu Heng has disclosed the worldview beneath his legal theory. In The Associated Press’s 2021 investigation into AFRINIC, he stated, “AFRINIC is supposed to serve the internet, it’s not supposed to serve Africa.” That sentence is not incidental. It is the argument. It treats AFRINIC as a neutral conveyor belt in a borderless market rather than as the regional steward of scarce number resources delegated to Africa under a public-trust logic. Once that premise is accepted, every attempt by AFRINIC to defend regional stewardship can be reframed as parochialism or abuse. Once it is rejected, Lu Heng’s latest text begins to look less like institutional diagnosis and more like a defense of extraction in theoretical language.

The seduction of abstraction

Lu Heng’s central claim is that the Regional Internet Registry system has become unstable because the registries exercise high-consequence power over a now-valuable resource while bearing only symbolic liability when things go wrong. He notes, correctly, that AFRINIC’s Registration Service Agreement, ARIN’s Registration Services Agreement, and the RIPE NCC Standard Service Agreement each contain liability limits. If one stopped reading there, one might conclude that the registries behave like sovereigns while contracting like clerks.

But the same contracts tell a very different story about the nature of the resource itself. AFRINIC’s RSA states that the holder’s right of use exists only within the ambit of the “need” justified in its application, that the number resources “are not property,” that transfers are prohibited except in narrow circumstances or under adopted policy, and that if the holder breaches the permitted use it must remedy the breach and or return the affected number resources to AFRINIC. RIPE’s standard agreement likewise says registration of Internet number resources “does not constitute property” and confers no ownership rights. ARIN’s policy manual says number resources are not sold under ARIN administration and are assigned for the stated purpose, subject to the ongoing terms of the RSA. In other words, Lu Heng highlights the liability clauses and brackets the stewardship clauses, even though the latter are what give the former their legal and normative meaning.

This is not a minor omission. It is the hinge of the argument. The registries are not capitalized as title insurers because the system was never built to guarantee asset values in a commodity market. It was built to maintain globally unique allocation and authoritative registration of a scarce coordination resource through community-developed rules. The liability caps Lu Heng finds so scandalous are real, but they do not prove that the registries were secretly exercising market-sovereign power while disclaiming responsibility. They prove that the registries were designed as stewards, not as warehouse banks for speculative scarcity.

That is also why it is wrong to reduce accountability to the narrow question of monetary damages. The RIR system’s accountability has never been only contractual liability. It includes member-controlled governance, community-driven policy development, publicly documented procedures, due diligence, registry transparency, operational continuity obligations, external scrutiny from the other RIRs and ICANN, and, in the most serious cases, the possibility of audit and derecognition under the evolving ICP-2 framework. The ongoing ICP-2 review exists precisely because the broader internet community has recognized that anti-capture rules, auditability and continuity need to be strengthened. Lu Heng writes as though the only serious accountability is tort-style liability proportionate to the market value of IPv4 blocks. But that is merely another way of smuggling property logic into a regime that was never built on ownership in the first place.

If Lu Heng wants to talk about responsibility, then the first responsibility to discuss is the one accepted by every resource holder when it signs the relevant contract. That responsibility is not abstract. It is used according to justified need, within the governing policy framework, in a manner consistent with accurate registry records. His essay asks us to imagine a registry detached from reality. Yet it is his reading that detaches the resource from the obligations under which it was issued.

The public record of his essay cannot absorb

The most striking feature of Lu Heng’s essay is not what it says. It is what it cannot quite bring itself to say plainly. AFRINIC’s own 2022 communiqué states that Cloud Innovation Ltd, registered in Seychelles and “owned and controlled by a Chinese national, Mr. Lu Heng,” was allocated 6.2 million IPv4 addresses by AFRINIC following needs expressed for the region and under a signed RSA. The communiqué says AFRINIC wrote to Cloud Innovation in June 2020 regarding breaches of the RSA and initiated contractual procedures in March 2021 after those breaches were not remedied.

On its own, AFRINIC’s account could be dismissed as the position of a litigant defending itself in court. But the wider registry community did not treat the dispute that way. In August 2021, ARIN publicly intervened because it believed the litigation posed a risk to the stability of the global internet number registry system. ARIN said that AFRINIC had determined the resources were not being used for the purposes for which they were issued, that Cloud Innovation had responded with multiple legal actions to avoid having to return the address blocks, and that the overwhelming majority of the approximately 6.2 million addresses were not being used within Africa. ARIN further stated that Cloud Innovation appeared to lease the space to other parties and that more-specific routing announcements from Hong Kong and the United States took precedence over a South African announcement, resulting in the vast majority of traffic usage occurring outside Africa. ARIN also disclosed that it had earlier refused Lu Heng’s own application for more than 1 million addresses in its own region because requested information was not satisfactorily provided, and the intended use lay outside the ARIN service region.

The RIPE NCC’s public explanation followed the same broad line. RIPE stated that AFRINIC had conducted audits, concluded that Cloud Innovation was not using the resources for their original purpose, informed the company that it might reclaim them, and then faced several litigations. That matters because it shows this is not simply AFRINIC’s internal narrative or African regional exceptionalism. The other registries saw enough in the underlying facts, and enough danger in the consequences of the litigation, to speak publicly and unusually plainly.

The court record tells the same story in another register. AFRINIC’s litigation FAQ says Lu Heng and associated entities have brought over 25 cases in Mauritius and 2 in Seychelles, with associated companies including Larus Cloud Service Ltd, Africa on Cloud Ltd and Crystal Web Pty Ltd. The public case list records a dense chain of actions. Cloud Innovation Ltd v AFRINIC, SC/COM/PET/000275/2021 sought amendment of AFRINIC’s members’ register and financial compensation of USD 1.8 billion. Cloud Innovation Ltd v AFRINIC, SC/COM/MOT/000382/2021 sought an injunction in relation to the resource transfer policy and was later withdrawn on 6 January 2022. Cloud Innovation Ltd v AFRINIC, SC/COM/JICA/000579/2021 sought conservatory attachment of AFRINIC’s 6.9 million unused IPv4 addresses. Africa on Cloud Pty Ltd v AFRINIC, SC/COM/PET/000769/2021 sought to wind up the registry. The case list also records applications to attach bank funds, injunctions relating to elections, a receivership application in SC/COM/MOT/000156/2023, a 2025 winding-up petition in SC/COM/PET/000508/2025, and a 2025 injunction application in SC/COM/WRT/000518/2025 aimed at preventing, among other things, new allocations.

Structures do not file 25-plus lawsuits by themselves. Incentives matter, certainly. Institutional design matters, certainly. But the public record here is not the record of a passive market participant innocently caught inside a defective architecture. It is the record of a powerful resource holder and associated entities pressing every available lever against a continental registry already weakened by internal dysfunction and legal vulnerability.

Even the receivership saga shows why Lu Heng’s structural critique is incomplete. In October 2024, the Court of Civil Appeal, in a judgment delivered by Senior Puisne Judge N. Devat and Judge D. Chan Kan Cheong, set aside AFRINIC’s appeal and restored the order placing the registry under the Official Receiver, while stressing the urgency of reconstituting AFRINIC’s board. One can debate whether receivership was the best available remedy. What cannot be denied is that the registry’s paralysis did not emerge from theory. It emerged from sustained conflict around real resources and from a legal environment in which those conflicts became operational weapons.

Why the transfer policy has been such a target

There is a reason the transfer policy sits at the center of so much pressure. AFRINIC’s Number Resources Transfer Policy, AFPUB-2020-GEN-006-DRAFT03, is not a marginal administrative tweak. It was submitted in November 2021 by Gregoire Olaotan Ehoumi, Noah Maina and Adeola A. P. Aina. It entered last call in December 2021, reached consensus in January 2022, had its appeal set aside in August 2022, and was formally ratified on 4 February 2026 after years of governance interruption delayed formal board action. AFRINIC’s own overview describes it as a community-developed proposal that moved through an open, consensus-based and transparent policy process.

Its substantive logic is straightforward and, from an African perspective, defensible. AFRINIC says it manages only 7.23 /8s of IPv4 space, a modest inheritance for a continent of more than 1 billion people and rapidly expanding networks. The policy allows regulated transfers of IPv4 blocks and ASNs, including inter-RIR transfers where reciprocal arrangements exist. But it also draws a critical line. AFRINIC-issued IPv4 cannot be transferred out of the region. Only certain categories, such as legacy resources or resources that entered the region through prior transfer mechanisms, may move outside the AFRINIC service region. AFRINIC’s ratification note says the policy is meant to enable controlled redistribution, reduce informal or underground transfers, preserve routing integrity, maintain authoritative registry data and reinforce AFRINIC’s governance mandate.

That is precisely why the policy has become a battlefield. It does not abolish transfers. It disciplines them. It acknowledges market reality without surrendering regional stewardship. It says, in effect, that Africa’s already limited inheritance should not become an unrestricted export channel simply because IPv4 scarcity has made address space more valuable elsewhere. AFRINIC’s own FAQ goes out of its way to state that the policy does not confiscate resources, revoke existing allocations or alter valid registration agreements. It further states that AFRINIC does not assign monetary value to number resources and that they are not commodities issued for speculation.

The public record shows that this flank was tested before. Cloud Innovation’s 2021 injunction application concerning the transfer policy, SC/COM/MOT/000382/2021, was eventually withdrawn. And now, 4 years later, Skyconnect arrives from Guinea to challenge the ratification of that same policy. I cannot prove from the public record that Skyconnect is taking instructions from Cloud Innovation. Serious analysis should say that clearly. But public institutions are not damaged only by formal chains of command. They are damaged by patterned pressure. When a relatively obscure operator appears to reopen a policy front already contested by a much more visible litigant, observers are entitled to ask whether they are looking at spontaneous principle or at another proxy front in a longer campaign.

In low-intensity institutional warfare, the smallest plaintiffs often reveal the largest interests. An almost invisible ISP does not usually wander into a guerrilla conflict over continental number policy by historical accident. One follows the pressure points, and the same strategic value keeps reappearing.

Africa is not terra nullius

The deeper reason this fight matters is that it sits inside a familiar African history. The resource changes. The grammar does not. For centuries Africa has been treated as a place where value can be cheaply extracted, revalued elsewhere and returned, if at all, as dependency. Gold, rubber, diamonds, copper, cobalt, coltan, oil, timber, data, labor, routes, spectrum and now digital infrastructure all enter that same broad archive of asymmetrical appropriation.

The language of “terra nullius” was once a legal fiction used to empty land of prior political claims. Africa was never literally ownerless, of course. But it has repeatedly been treated as though its resources were functionally available to whoever could map, price and remove them. That logic survives in contemporary form. The International Energy Agency’s Global Critical Minerals Outlook 2024 shows why competition for control over material inputs is intensifying. UNCTAD’s Digital Economy Report 2024 reminds us that digital infrastructure is inseparable from physical extraction, energy and supply chains. And Brookings’ recent work on AI sovereignty underscores that technological power now sits across a stack of choke points that includes minerals, networks, data, compute, standards and governance. Internet number resources belong to that same stack. They may look invisible, but they are not immaterial.

IPv4 address space is not cobalt. It is not lithium. It is not copper. But it is strategic in a similar way. It confers routable identity. It affects market entry, network continuity, trust in registry data and the ability of operators to expand. It also becomes more valuable as scarcity deepens. Once that happens, a regional registry can be treated either as a steward of common infrastructure or as the inconvenient clerk sitting between a trader and a monetizable asset. Lu Heng’s writing belongs to the second tradition. It insists that the registry’s authority is the central danger while saying remarkably little about the political economy that turned African-issued address space into a platform for extra-regional arbitrage.

This is not a plea for xenophobia, and Africa must resist the temptation to turn every digital conflict into a lazy morality play about one nationality or another. AFRINIC’s 2022 communiqué identifies Lu Heng as a Chinese national, and the AP investigation described him as Hong Kong-based. Those facts are not arguments by themselves. The problem is not that a Chinese businessman became involved in Africa’s address space. The problem is that African institutions remain vulnerable to external actors, of whatever passport, treating the continent as the weakest node in a global market for scarce digital resources. Africa should not become an arena for imported corporate wars, nor for a simplified China-versus-America script, nor for a new round of geopolitical ventriloquism in which continental infrastructure is merely the stage on which others perform their rivalries.

The right lesson is not anti-Chinese. It is anti-extractive. It is the same lesson African policymakers should draw, whether the pressure comes from Chinese capital, American market ideology, European paternalism, Gulf finance or homegrown oligarchy. Scarce African public resources should not be normalized as disposable inputs into other people’s accumulation strategies. Number resources are part of Africa’s digital political economy. To pretend otherwise is merely to repeat the older mistake in a new vocabulary.

The answer is not political capture either

Lu Heng tries to widen his case by arguing that higher layers of governance, including ICANN and Smart Africa, have responded to AFRINIC’s fragility by reproducing the same asymmetry between power and responsibility. On this point, one has to be careful. There is a legitimate African concern here. The answer to a registry crisis cannot be raw political sponsorship, governmental overreach or a continental branding exercise that ends by weakening neutrality and procedural legitimacy. A registry captured by states is no more sovereign than a registry captured by markets. It is simply captured differently.

But this only makes Lu Heng’s essay more evasive, not less. The ICANN letters of 25 June and 16 July 2025 did not tell AFRINIC to serve one faction. They expressed grave concern about election integrity, improper influence, forged powers of attorney and AFRINIC’s obligation to the entire African Internet community rather than “any single resource member or group of influential resource members.” The Number Resource Organization’s statement of 26 June 2025 supported transparency and accountability in AFRINIC’s election process and reaffirmed neutrality, community-developed policy and member-driven governance as foundational principles. Those interventions were not a defense of political capture. They were an insistence that AFRINIC still had to meet the conditions of trust required of a regional registry.

In fact, the March 2025 ICANN letter to the receiver makes the central principle unmistakable. AFRINIC is the only entity empowered to assign IP addresses in Africa and the southern Indian Ocean. The resources allocated to an RIR are not assets of the RIR in the ordinary commercial sense. The registry function is held in public trust and cannot be used for the benefit or protection of a single creditor, member or group of members. That is the opposite of the world Lu Heng’s essay quietly assumes. He wants us to see liability mismatch. ICANN’s letter reminds us to see entrusted stewardship.

I am not arguing that African institutions should accept every intervention from ICANN or every continental initiative at face value. Far from it. Africa should insist on clean governance, genuine member control, transparent elections, rigorous due process and public-interest accountability, whether the threat comes from a resource holder, a government coalition or a global coordinator. But none of that rescues Lu Heng’s essay. It only exposes how selective his institutional realism really is. He becomes a structuralist when the subject is registry liability and a market liberal when the subject is regional stewardship. He wants the language of public accountability for AFRINIC and the language of private economic seriousness for the address blocks. That asymmetry is not diagnosis. It is advocacy.

Return the 6.2 million addresses

There is a reason the debate keeps circling back to the same number. Six point two million. That is the scale of the allocation at issue. AFRINIC says those addresses were issued to Cloud Innovation under an RSA following needs expressed for the African region. ARIN says the overwhelming majority were not actually being used within Africa. RIPE says AFRINIC’s audits found a mismatch between original purpose and actual use. AFRINIC’s RSA says the right of use exists within justified need, that number resources are not property, and that breach may require return of the affected resources.

If words still have meaning in the governance of internet infrastructure, then the implication is not obscure. The 6.2 million addresses should be brought back within the African public-trust framework unless full compliance with the governing obligations can be demonstrated to the satisfaction of the lawful registry process. They should not remain normalized as a platform for speculative leasing, shadow transfer or extra-regional arbitrage merely because scarcity made them valuable and because litigation proved an effective delaying mechanism.

Returning those addresses would not cure every defect in the RIR system. It would not erase AFRINIC’s own past governance failures. It would not by itself resolve the tension between scarce resources and institutional accountability. But it would establish one basic principle that Africa urgently needs to defend. Scarce number of resources delegated for regional development are not just another tradable output waiting to be abstracted from the region that was supposed to benefit from them. They are part of Africa’s digital commons, however imperfectly administered, and they should be governed accordingly.

Lu Heng says a registry cannot retain consequence-heavy authority while carrying consequence-light responsibility. On that narrow point, one can agree. But the same sentence has an even sharper application to large resource holders. No actor should be able to accumulate a substantial share of a region’s scarce address space, route or lease it primarily outside that region, attack the registry when challenged, and then present himself as the philosopher of accountability. If Lu Heng wishes to speak in the language of responsibility, he should begin where his own critics have long asked him to begin: with the return of the 6.2 million addresses that, on the public record, no longer belong in a system of extraction detached from African purpose.

AFRINIC as continental infrastructure

The lesson for Africa is larger than Lu Heng, larger than Cloud Innovation and even larger than the present lawsuit from Guinea. AFRINIC should no longer be treated as a niche technical body to be noticed only when it collapses. It is continental infrastructure. It sits in the same category of strategic concern as landing stations, terrestrial backbones, exchange points, data centers, satellite gateways, trusted public registries and the legal frameworks that make cross-border digital cooperation possible.

If digital sovereignty is to mean anything beyond conference rhetoric, then African policymakers, regulators, diplomatic missions, operator groups, civil-society actors and public-interest coalitions must stop behaving as though AFRINIC’s survival is someone else’s administrative detail. The African Union’s Digital Transformation Strategy for Africa 2020-2030 speaks the language of a continental digital single market, harmonized policy and African ownership of modern tools of digital management. That language has to apply to number resources too. A continent that talks seriously about AI sovereignty, data governance and strategic technology autonomy cannot treat its regional internet registry as an orphan left to litigants, receivers and exhausted staff.

Africa’s governments do not need to choose a corporate faction. They need to choose the institution. They need to defend free and fair elections. They need to defend community-made policy. They need to defend rigorous compliance, authoritative registry data and the principle that scarce African number resources should be managed for African connectivity first, even when global demand makes them more lucrative elsewhere. The answer to AFRINIC’s fragility is not to abolish stewardship in favor of pure market entitlement. It is to rebuild stewardship so that it is harder to capture, harder to paralyze and harder to narrate away.

That is why Lu Heng’s essay should not be read as a neutral theory paper. It is part of the struggle over meaning. It asks the reader to believe that the real scandal is not the externalization of African-issued scarcity but the registry’s authority to object. It recasts stewardship as predation, discipline as instability and community policy as discretionary abuse. That is rhetorically sophisticated. It is also politically familiar. Extractive systems have always depended on a language that makes appropriation sound like realism and resistance sound like obstruction.

Africa has heard that language before. It should not accept it now merely because it arrives in the measured cadence of internet governance theory. AFRINIC is not a warehouse clerk for the world’s remaining IPv4 hunger. And Africa is not a digital quarry.