India connects more than 950 million people to the internet with roughly 41.6 million IPv4 addresses. That’s about 29 per thousand residents. The United States, with under a quarter of India’s population, holds close to 1.6 billion addresses, near 4,700 per thousand. The global average is around 0.45 per person; India sits far below it, the US roughly ten times above.

That gap is a function of timing, not demand. Countries that connected early claimed address space when it was abundant and free; those that arrived later inherited a near-empty pool and, in two regions, built a national institution to manage what little they had. That institution is the National Internet Registry (NIR). China has CNNIC, India has IRINN, Japan has JPNIC, and Korea, Taiwan, Vietnam, Indonesia, Brazil, and Mexico each run their own. That’s nine in total, all under APNIC or LACNIC. No other RIR region uses the model, and APNIC made its moratorium on new NIRs permanent in 2024.

The NIR was meant to put a country in charge of its own number resources. Sometimes it has, and done it well. Elsewhere it has become the reason address space can’t move to where growth needs it. The question worth asking honestly: does the model help countries grow into the address space they need, or does it get in the way?

What an NIR actually is

The delegation chain normally runs IANA → RIR → LIR. APNIC and LACNIC add a rung: IANA → RIR → NIR → member. NIRs are independent nonprofits under national law, not subsidiaries of their parent RIR, and answer to a domestic constituency while implementing regional policy. Membership isn’t always optional: APNIC-region organizations choose their national NIR or APNIC directly, not both; Brazil and Mexico require the NIR route. That distinction shapes how much friction the national layer can impose.

The legitimate case for NIRs

Dismissing NIRs as pure overhead would be inaccurate. JPNIC grew out of a real need in 1990s Japan. Local-language support a distant regional registry couldn’t offer at scale. NIC.br supplied the staff that let LACNIC begin operating as a regional registry in 2000. The sovereignty argument has merit too: address allocation determines who can build networks and on what terms, and a government’s interest in that isn’t inherently illegitimate. JPNIC and TWNIC run transparent registries with published fees and clear transfer rules, and no one credibly argues they hold Japan or Taiwan back. The record is genuinely mixed. It weakens specifically on the functions that matter most to a liquid address market.

Where the friction shows up

Transfer paths are inconsistent or absent APNIC’s NIR transfer table shows Korea’s KISA/KRNIC offering no IPv4 transfer service in either direction, and Vietnam’s VNNIC accepting inbound transfers only. Korea holds about 2,174 addresses per thousand people—among Asia’s highest densities—with no national channel to move surplus to markets that need it. LACNIC adds a three-year hold after initial allocation before a block can transfer, plus a one-year hold after any transfer, combined with mandatory routing through the NIR in Brazil and Mexico.

Registry data reliability has been tested and found wanting in places After an anonymous report in late 2023, APNIC audited its NIRs’ delegation records. Results published in December 2024 flagged 51 delegations at India’s IRINN, of which 44 were non-compliant and recovered; at Indonesia’s IDNIC, roughly 1,200 of 2,974 allocations reviewed were flagged, prompting an independent investigation. APNIC declared zero tolerance for misallocation and committed to auditing every NIR and itself. Registration accuracy underpins routing security; when a national layer can’t guarantee it, the cost is systemic.

NIR members lose their regional vote NIR members aren’t APNIC members individually, so they have no vote in the regional policy process governing their own resources, and several NIRs’ fee schedules (CNNIC, KRNIC, VNNIC) aren’t published in verifiable form. Direct APNIC membership, by contrast, costs a transparent A$1,295 in 2026, halved for least-developed economies.

Leasing remains unsettled APNIC treats leasing as acceptable only within a genuine connectivity service; a 2023 proposal to ban it outright (prop-148) was abandoned after failing to reach consensus. CNNIC describes its allocations as rental rather than ownership and requires its consent for any transfer, favoring incumbents over the smaller networks that most need flexible access. (CNNIC’s separate 2015 history as a distrusted certificate authority is unrelated to its registry role.)

Idle space sits while scarcity bites APNIC exhausted its free pool in April 2011. Of roughly 3.69 billion allocated IPv4 addresses globally, about 3.1 billion are actually announced and routed. The remaining 587 million, about 16%, are allocated but invisible on the internet.

The sovereignty argument, and its paradox

The desire for national control isn’t irrational. The 2016 IANA stewardship transition moved global registry oversight from the US government to the multistakeholder community, but the underlying disagreement over who should govern the internet’s core resources never fully resolved—at WCIT-12 in Dubai, 89 countries signed a treaty asserting sovereign authority over their internet segments, while 55, including the US and UK, declined. An NIR is, functionally, the institutional expression of a country wanting to administer its own number resources rather than depend on a foreign-headquartered body.

The paradox: gatekeeping in service of that instinct produces the opposite of resilience. A locked-in, slow-moving, opaquely recorded address pool doesn’t make a network more independent. It makes it less able to draw on the global pool of idle space when domestic demand requires it. Sovereignty over a resource and liquidity in it aren’t opposed; the best-run registries manage both.

What the comparison actually shows

Testing NIR status against outcomes destroys a tempting but false narrative: NIR countries do not lag on IPv6 adoption as a bloc. In a June 2026 APNIC Labs snapshot, India—an NIR economy—leads the entire comparison group. Singapore and Hong Kong, both without an NIR, sit at the bottom.

Scarcity, not registry structure, is the variable doing the work at the address-poor end. India (29 addresses per thousand), Indonesia (68), and Vietnam (164) are racing to IPv6 for lack of alternative. Among address-rich economies, the pattern breaks down: Germany, at about 1,469 per thousand, leads that group at 74% IPv6 adoption, while better-supplied Singapore sits at 28%. South Korea is the sharpest case—2,174 per thousand, only 29% adoption, and a registry with no transfer path per APNIC’s own table. Abundant supply, no release mechanism, and limited modernization pressure add up to capacity that stays off the global market indefinitely.

Confounders matter here: Singapore and Hong Kong are financial and connectivity hubs whose address wealth and hosting-heavy IPv6 profile reflect their role, not NIR structure. The defensible conclusion is narrower—NIR status alone doesn’t predict outcome. Where the registry runs open and efficiently, the economy performs well; where it adds restriction and opacity, the cost lands on its own members and the wider system.

What would fix it

None of this argues for dismantling the NIR layer. It argues for changing its posture: harmonize with the parent RIR’s policy so an address can leave as easily as under a direct-APNIC economy; guarantee a working two-way transfer path at every NIR (transfers moved about 33 million addresses across the RIR system in 2025—the mechanism works where it’s permitted, national gates just fragment it); give leasing a clear, registry-backed status with defined rules on registration and routing authorization; treat IPv6 acceleration as complementary, since the NIR economies that pushed hardest—India foremost—already show the model can drive modernization; and fix data integrity with RPKI-secured records so routing and market transactions can trust the registry.

A fair objection remains: unrestricted transfer could concentrate address space further, and the US already holds nearly 44% of all allocated IPv4. That’s where leasing matters as a release valve distinct from outright transfer, and it’s worth noting the secondary market has already adjusted for scarcity concerns, with per-address prices roughly halving through 2025 from 2021–22 peaks.

The choice that matters

Whether an NIR helps or hinders its country isn’t structural inevitability, it’s a posture choice. India, with an NIR, tops the modernization ranking here; Hong Kong, without one, sits at the bottom. The institution isn’t the variable; what it does with its position is.

A national registry can lower barriers and give a country a legitimate channel into resources its economy depends on, or it can lock addresses inside borders, keep unverifiable records, and leave capacity idle while neighbors go without. Both outcomes are live today, region by region. The fix isn’t abolishing the national layer; it’s holding it to the same transparency and liquidity standard as the RIR above it.

This article is part of a bigger opinion piece. Read the full text on the IPXO’s blog.