On 08/01/2026 02:24, Chubasco Diranga wrote:
Subject ------------------------------------------------------------------------
Dear John, CPWG colleagues,
Thank you for the detailed data and analysis. The figures you shared clearly illustrate two structural realities that many regions— particularly AF and Pacific Island countries—continue to face:
These are the figures for the AP (Asia Pacific) region, Chubasco, The important thing to bear in mind is that they are the local hosting figures for gTLD websites on IP addresses associated with the countries. Cloud hosting also adds an extra layer of complexity. There is often a parallel market in each country's ccTLD. What happens in the early phase of a country's Internet infrastructure development is that much of the hosting will be on the hosting infrastructure of other countries. This is exactly what happened with Ireland and other European countries in the early 1990s. As the infrastructure developed, more local providers appeared and much of the hosting moved back to the local providers. Some providers will maintain their reseller accounts and web hosting in other countries even when the infrastructure has developed. The reason is often based on pricing. This leads to some countries having larger markets (US/DE/UK/FR/AU/SG) that include some of these providers. Godaddy's main DNS provides DNS service for approximately 11,882 web hosting providers. These are effectively resellers and customers of Godaddy that do not host their websites on Godaddy's web hosting. This is part of the reseller layer that Alan Greenberg has mentioned on the calls. Resellers with their own DNS are often easier to associate with a country as the domain name WHOIS could, in the past, be checked. While some registrars are larger providers that may have customers in other countries, resellers tend to be incredibly local with their customer base because they are often web developers who are developing websites for customers in their locality and they know their customers. The catch, and there always is one, is that they may often host outside their country's Internet infrastructure. One of the best examples is that of Nigeria (NG). It has 1,598 gTLD websites on IP addresses associated with Nigeria. It has 84,497 gTLD domain names on identified Nigerian hosting providers. Local website hosting markets AP region December 2025 AP Japan JP 3,482,598 AP Turkey TR 1,191,467 AP India IN 1,119,015 AP Korea (South) KR 659,976 AP Singapore SG 2,132,016 AP Iran IR 323,947 AP Bangladesh BD 65,176 AP Philippines PH 52,288 AP Kazakhstan KZ 13,835 AP Israel IL 39,166 AP Cyprus CY 37,169 AP Saudi Arabia SA 25,031 AP Cambodia KH 40,818 AP Indonesia ID 326,831 AP Jordan JO 1,531 AP Viet Nam VN 305,056 AP Malaysia MY 164,210 AP Georgia GE 2,239 AP Uzbekistan UZ 1,322 AP Mongolia MN 844 AP Pakistan PK 18,634 AP Iraq IQ 729 AP Nepal NP 4,926 AP Hong Kong HK 6,604,549 AP China CN 3,031,950 AP Australia AU 1,837,150 AP Taiwan TW 227,125 AP Thailand TH 158,195 AP United Arab Emirates AE 65,921 AP Armenia AM 37,810 AP New Zealand NZ 29,293 AP Bahrain BH 3,001 AP Macau MO 2,158 AP Laos LA 1,370 AP Lebanon LB 1,278 AP Kyrgyzstan KG 1,127 AP Kuwait KW 1,071 AP Azerbaijan AZ 1,046 AP Marshall Islands MH 912 AP Palestine PS 897 AP Brunei BN 815 AP Syria SY 677 AP Oman OM 566 AP Qatar QA 544 AP Sri Lanka LK 489 AP Turkmenistan TM 370 AP Myanmar (Burma) MM 362 AP Yemen YE 190 AP Maldives MV 133 AP Tajikistan TJ 109 AP Afghanistan AF 85 AP East Timor TL 63 AP Papua New Guinea PG 56 AP Vanuatu VU 51 AP Bhutan BT 27 AP Fiji FJ 23 AP Samoa (western) WS 10 AP Palau PW 7 AP Tonga TO 5 AP Cook Islands CK 2 AP Niue NU 2 AP Micronesia FM 1 AP Kiribati KI 1 AP Korea (North) KP 1 There are approximately 32,980 identified web hosting providers so far and they account for 196,608,845 gTLD websites out of 197,786,326 websites (99.4%). The providers of the remaining 1,177,481 websites are still being checked. Many of these are smaller resellers. There is also another feature of developing markets in that the local phone and telco companies become the first generation of ISPs and providers. People will also host websites on their ISP connection. The figures for China and Hong Kong might look a bit strange. Some of the large Chinese Cloud hosting operations and Chinese and Hong Kong resellers use US flagged IP addresses. The Chinese market also has a much larger ccTLD (CN) market. China has approximatley 15.96 million gTLD registrations and most of the websites of those domain names are on US flagged IPs. Hong Kong has a large number of websites on IP addresses that were originally allocated to the African region. Some of the countries in the AP region have large numbers of gTLD websites and no ICANN registrars. Their gTLD regisrtation activity is 100% outsourced.
1.
the absence or scarcity of locally accredited ICANN registrars, and
Originally, Network Solutions was the only registrar for .COM (and .NET and .ORG). There was a change to a multiple registrar system. The problem is that the requirements for accreditation were defined then and in a very different Internet market. I think that there is a requirement for a $70,000 line of credit for a prospective registrar. That is not feasible for a lot of small resellers and web developers. I am not sure that it was even based on any economic analysis at the time. It created a kind of two tier domain name sales business with the registrars and their resellers. By the mid 2000s, the ccTLDs started to compete with the gTLDs. The registrar requirements for the ccTLDs are often more attuned to their local markets so web developers and resellers often become registrars in their local ccTLD rather than ICANN accredited registrars. And this has cascaded to such an extent that the gTLDs have become "legacy" TLDs in many markets with the local ccTLD being larger than the number of locally registered gTLDs. The .COM is still the top TLD for a global market. At a local level, other than the US market, it is competing with the local ccTLD.
2.
the loss of meaningful market visibility following GDPR and WHOIS/ RDAP changes.
In many small or developing markets, expecting registrar accreditation as the primary path to local gTLD ecosystem development is neither realistic nor cost-effective. At the same time, relying on WHOIS/RDAP fields to understand local markets is increasingly unreliable due to privacy and proxy services.
Apart from scraping all WHOIS data, the commonest way (pre-GDPR) of estimating market share for countries was to take a sample of domain names and check the country field of the registrant. In an ideal situation, as expected by some, WHOIS data was properly formatted and accurate. In reality, people will intentionally obscure their data for all sorts of reasons. They might even put a two character code for a local province or region in to the country field instead of the code for the country. That's the measurement problem. The economic problem for gTLDs in developing markets is that the gTLDs are not seen as being local in the same way as the local ccTLD. As the ccTLD grows, people increasingly think of it as being *their* TLD. The .COM has immediate recognition but other legacy gTLDs have much smaller registration numbers. People may recognise .ORG more than .NET. The 2012 round of new gTLDs had this recognition problem in that they depended on the registrars and resellers to promote them. (A big mistake. registrars and resellers promote what sells.) Some registries, especially some of the city and regional gTLDs, were very good at promoting their gTLDs. There is a very good economic reason for that and it is the same reason why ccTLDs are so effective. The market for those geo gTLDs was concentrated in geographical or linguistically. This is also why IDN gTLDs are more like ccTLDs than gTLDs. ICANN's gTLD registrar model was great for the much smaller Internet of the 1990s. It is too late to change it now for the 2026 round and the real driving force for some of the geographically or linguistically focused 2026 gTLDs will be the resellers in those markets.
A practical complementary approach may be to introduce a *Trusted Local Business Verification Layer*, operating alongside (not instead of) registrars and registries.
It is an interesting idea. The ccTLDs will have the advantage of being associated with the country in a way that most gTLDs are not. That is one of the most effective sales tools for ccTLDs in that they can claim to be local and more trustworthy than the gTLDs. Some of them use it in their marketing.
In simple terms, this would be a *Trusted Local Business Register* managed by recognised local institutions (e.g. chambers of commerce, SME authorities, ICT regulators), which verifies legitimate local entities independently of WHOIS data. Registrars and registries—particularly new gTLD applicants in the 2026 round—could voluntarily reference this verification to:
The hardest part would be getting all the interested parties to agree. Each of them will have their own reasons for pushing themselves as being in charge of such a venture.
demonstrate genuine local market engagement,
One of the biggest mistakes made in the 2012 round by some registries and their advisors was underestimating the sheer hard work, expertise and money necessary to get a new TLD established. Some of the claims that I read in media during the 2012 round made it clear that expertise and knowledge about what it takes to establish and grow a TLD was in very short supply while there was a glut of snakeoil. The registries, the registrars and the resellers all play their parts in promoting a new TLD. Successful TLDs often build a community as well as being locally engaged.
distinguish legitimate local use from speculative registrations, and
Speculative registrations are an important indication of interest in a TLD. They are a gamble on the success of a TLD when they appear in a newly launched TLD. There are two main examples of what can go wrong with a new TLD when specuative activity becomes a problem. The first, obviously, is the .EU ccTLD fiasco. The people advising the European Commission didn't seem to understand the domain name industry. The European Commission awarded the contract to run the ccTLD to a small ccTLD registry (with a few other ccTLD registries as part of the consortium) that had no obvious experience in running a gTLD. The .EU ccTLD was effectively a gTLD with 27 different languages. A few hours after the ccTLD opened for general registration, the registrations backend fell over. It also lauched with no transfer mechanism in place so that people couldn't sell and transfer domain names to each other. The ccTLD was absolutely plundered by non-EU speculators to such an extent that demand, other than brand protection, collapsed. It never recovered. Twenty years later, the .EU ccTLD typically has less than 5% of the domain name markets of most EU countries. It is not a First Choice TLD and it never became a replacement for the .COM in the EU market. The main axis for EU markets is the local ccTLD and .COM. These two TLDs generally represent 80% or more of the local EU country markets. Many of the domain names registered speculatively were later deleted. The .EU fiasco was the result of unrealistic expectations, low industry knowledge and poor execution. Properly executed, the .EU ccTLD could have been a viable alternative to .COM in the European Union. The second example was Uniregistry's 2012 round gTLDs. It demonstrated the importance of having some speculative activity in a new TLD. Many of the "good" speculative domain names were registered via a registry affiliated registrar (this feature had been approved before the 2012 round launched). That actually reduced the demand for some of these gTLDs and a flood of registrations did not happen. This is because speculative sctiviy actually drives some of the demand in the launch of a new TLD for the first six months or so. Speculation gets people interested and helps drive demand. Many speculative registrations will be deleted without ever having a developed website. A Junk Dump typically follows the launch of a new TLD a year later. The attention that a reasonable leel speculation gains for a TLD helps get it noticed and builds awareness. This case was an example of somewhat unrealistic expectations, too much one-sided knowlege (which domain names would have been valuable in .COM) and a collision with the brick wall reality of the domain name marketplace. Low awareness = low demand = low development = low registration volume. Without development and usage in a TLD, domain names have no value beyond the registration fee. Specuative registrations depend on that development and usage even more than the registries. Too much speculation in a new TLD depresses development. That is why the management of some registries don't want excessive speculation. Some people may think that all speculation in a new TLD is bad. Its not. In moderation, it helps the new TLD to build a stock of initial registrations. Some 2012 round new gTLD registries were more careful and only retained a smaller number of potentially high value domain names for auction. To consider all speculative registrations as being bad is dangerously simplistic for a registry because they provide unpaid marketing for the TLD. Those registrants will often be promoting the domain names that they have for sale and that's unpiad marketing for the TLD.
*
partially restore market visibility lost through privacy regimes, without exposing personal data.
Such a model does not require ICANN registrar accreditation, does not alter existing contractual frameworks, and is GDPR-safe by design. It may be especially relevant in AF and Pacific regions where reseller models dominate and local hosting ecosystems are still emerging.
ICANN isn't really s player at the local level though it might help with connecting people. It is up to the registries, registrars and resellers to do the hard work. A local approach with local engagement would be much more effective. The profit margin on domain names is low and most of the organisations that would potentially be interested will also be selling other services such as web hosting, web development, brand protection, and search engine marketing. Civil society organisations will probably want some involvement too.
From a CPWG perspective, this kind of locally governed verification layer could:
*
support underserved-region applicants in the 2026 round,
*
provide better post-WHOIS market signals, and
*
strengthen public-interest justifications without adding regulatory burden.
I believe this approach aligns well with CPWG’s focus on practical market realities and could be explored as a voluntary best practice or pilot concept rather than a policy mandate.
It might be a good idea to establish what kind of local trust and verifcation schemes already exist. Some of them may be deeply involved with the local ccTLDs. Regards...jmcc -- ********************************************************** John McCormac * e-mail: jmcc@hosterstats.com MC2 * web: http://www.hosterstats.com/ 22 Viewmount * Domain Registrations Statistics Waterford * Domnomics - the business of domain names Ireland * https://amzn.to/2OPtEIO IE * Skype: hosterstats.com ********************************************************** -- This email has been checked for viruses by Avast antivirus software. www.avast.com