SUVAN NET and the margin risk below cloud scale

Summary

• SUVAN NET is a real Tashkent hosting, domain and software operator rather than a company inferred from an internet registry entry. It trades through aHOST and AllDomains, holds a valid .UZ registrar contract, is currently the only ICANN-accredited registrar listed for Uzbekistan and reports more than 40,000 customers. Those facts support a recurring-service business, but public evidence does not disclose revenue, renewal rates, churn, margins or cash flow.
• Its network footprint is meaningful but economically ambiguous. AS35326 currently originates four IPv4 routes covering 2,048 addresses and one IPv6 /29; RIPE collectors observe one IPv4-side neighbour and one IPv6-side neighbour. The routes are registered, but none of the five returned a validating RPKI authorisation, the registered routing policy is stale, and public paths do not establish traffic, capacity, physical diversity or profitable demand.
• The strongest potential advantage is not raw server capacity. It is the combination of local billing, domain accreditation, reseller automation, support and hosting. The downside is visible in public tariffs: local hosting is priced above the company's German alternative, international bandwidth can cost several times the entry VDS, competitors advertise more compute and bandwidth near the same price, and a June 2026 power failure interrupted part of the cloud estate. Positive cash after renewals, supplier costs and resilience investment would change the judgment; resource ownership alone does not.

The incentive is to make small scale feel local, not small

SUVAN NET's management problem is familiar to every regional infrastructure provider. It cannot outspend the largest cloud platforms on processor generations, global regions, software ecosystems or procurement. It can still remain relevant if a local customer values one accountable provider for a domain, server, support ticket, invoice in Uzbek soum and compliance question. The strategy is to turn proximity into lower friction.

That proposition can earn money in several ways. Domain renewals recur annually. Shared hosting and virtual servers recur monthly or annually. A customer who buys a domain may add hosting, email, an SSL certificate, server administration or a website. A web studio or hosting reseller may integrate once and place many orders through an application interface. A government body or local company may prefer a domestic contract, local support and data held in Uzbekistan. Each additional service can raise revenue without requiring a new customer-acquisition event.

The same bundle can conceal weak economics. Domains carry registry charges and support obligations. Hosting carries hardware, software licences, power, cooling, storage, backups and staff. International traffic carries an upstream bill. Accreditation adds audit, escrow, abuse and policy work. A low-priced account can consume little infrastructure in an ordinary month and then require expensive intervention during a failure. The customer sees one small invoice; SUVAN NET carries a stack of fixed and contingent costs behind it.

This is why the starting point is not the quantity of internet addresses. It is who pays, what they renew and what the company must spend to keep the promise. A resource holder can still be a price-taker if customers view its server as interchangeable with dozens of local and foreign alternatives. A smaller provider creates value only where control reduces a cost, supports a premium or makes the customer sufficiently reluctant to leave.

SUVAN NET has assembled more control than its sparse corporate profile initially suggests. Whether that control earns a return remains unproved because the company does not publish accounts or operating metrics. The public evidence is strong enough to identify the levers, and strong enough to show where management's downside sits.

A Tashkent company with three operating faces

The legal entity is identifiable. A corporate-data page reproducing Uzbek registry information gives the name, taxpayer number 301551793, a 23 March 2010 registration date, active status, a Tashkent address and data hosting and processing as the principal activity. It reports charter capital of UZS425.2 million, Artur Suvanov as director and sole founder, and IT Park residency from December 2019. The page warns that its information is unofficial and may lag source registers, so those ownership and capital details should not be treated as audited financial disclosure.

The operating identity is clearer on the company's own services. The aHOST site says aHOST is a SUVAN NET trademark and offers .UZ and international domains, shared hosting in Uzbekistan and Germany, virtual and dedicated servers, SSL certificates and reseller hosting. AllDomains' public offer identifies SUVAN NET as the provider of domain registration, renewal, transfer, restoration, privacy, DNSSEC and related support. CONTRACTS identifies its electronic-document platform as developed and maintained by SUVAN NET and points to aHOST for hosting and SITES for development work.

These are not three separate companies in the evidence reviewed. They are distinct customer propositions around the same legal operator. aHOST is the retail infrastructure and domain brand. AllDomains is the registrar and wholesale-reseller proposition. CONTRACTS is a software service that can consume the company's own hosting and operational capability. The corporate-development offer adds websites and applications. This boundary matters because the economics are broader than internet access and narrower than a hyperscale cloud.

It would be misleading to call SUVAN NET a household broadband provider simply because it has an autonomous system and a data-network licence. No public residential access tariff or address-level fibre footprint was found. Its visible commercial products are hosting, servers, domains, certificates, reseller infrastructure, software and related support. The assigned regional-ISP category reflects network-resource and connectivity context, but the economic unit is usually a hosted account, domain, virtual machine, server or service contract rather than a household line.

The company's RIPE NCC member record independently aligns the legal name and Olmachi Street address and says the member services Uzbekistan. A current Uzbek .UZ registrar card gives the same taxpayer number and address and shows contract 02/R, dated 24 May 2019, as valid. An official open-data licence record lists licence AA 0006680 for data-network design, construction, operation and service provision. Together, those records establish an operating boundary that marketing copy alone could not.

They do not disclose the financial boundary. Public sources reviewed for this article do not show annual revenue, operating profit, cash, debt, capital expenditure, employee count, customer acquisition cost or revenue by brand. Charter capital is an ownership contribution, not sales. A customer count is not recurring revenue. The absence of accounts requires a conditional judgment rather than an invented valuation.

A renewal business disguised as an infrastructure catalogue

The most attractive part of SUVAN NET's model is recurrence. A domain has to be renewed. A website must stay online. A virtual server continues billing while the customer's application runs. Email, certificates and support can be attached to the same account. The customer's cost of moving is not only the next provider's price; it is the risk of changing nameservers, mail, files, databases, certificates, billing contacts and renewal dates.

The aHOST catalogue uses bundling explicitly. A one-year shared-hosting purchase can include a .UZ or another eligible domain for the first year. The customer arrives for one product and receives a reason to keep two. The hosting tiers increase disk space, supported sites, subdomains and databases, while email and traffic are presented as unmetered. This is a classic fixed-platform model: sell many accounts over shared servers and control panels, assuming that most customers use only a fraction of the stated ceiling.

AllDomains pushes the same logic one level up the chain. Its partner page targets legal entities, charges no subscription fee, requires the partner to keep enough prepaid balance for registrations and signs contracts electronically. Its home page targets registrars, web studios, hosting companies and digital agencies with a billing panel, REST integration and a WHMCS module. A reseller can put its own customer relationship on top of SUVAN NET's registrar connection.

That arrangement shifts working-capital risk toward the reseller. The partner pre-funds the balance before a registry transaction. SUVAN NET can receive cash before it has to provide the service, while the partner carries the risk of finding an end customer. It also reduces sales friction: one integration can generate many registrations and renewals. The strategic question is whether the margin per transaction remains large enough after registry cost, payment expense, support and compliance.

The company reports scale that would make recurrence meaningful. The current aHOST homepage says it serves more than 40,000 customers, including more than 6,000 legal entities and more than 500 state organisations. These figures are company-reported and the word "customer" is not defined. One customer may hold only a domain, another a substantial server, and an old account may not be active. No independent source supplies a current paid-account count.

Historical company material nevertheless suggests that domains have been central rather than incidental. A late-2022 aHOST notice said its customers held more than 53,000 .UZ domains and had registered more than 18,000 new names that year. Those claims are dated and cannot be carried forward as a current market share. They do show the company's own view of the engine: domain volume feeds a renewal base, and that base can be cross-sold infrastructure.

The model is durable only when renewal outweighs support and churn. A dormant domain may be cheap to administer. An abuse complaint, failed transfer, disputed registrant identity or expired business-critical name can consume skilled time. Hosting can be sticky, but a customer that outgrows the platform may move to a larger cloud. The portfolio creates optionality; it does not guarantee that every product creates value.

The domain credential is the strongest differentiation

SUVAN NET has a distinction that many regional hosts cannot claim. The IANA registrar-ID registry lists Suvan Net LLC as accredited under IANA number 4327, with its AllDomains RDAP endpoint. ICANN's country-filtered registrar list currently returns one Uzbek entry: Suvan Net LLC. A .UZ administration report from October 2024 described the company as recently accredited when an ICANN representative visited Uzbekistan.

Accreditation can improve the wholesale proposition. AllDomains says it works directly with global registries, offers more than 1,000 domain zones and automates registration, renewal and transfer. A local reseller can receive Uzbek or Russian support, contract with an Uzbek entity and settle in local currency without building its own registrar compliance stack. Direct registry access can remove an intermediary and leave room for a wholesale margin.

The advantage is operational, not absolute. ICANN accreditation is available to qualified companies globally, and a customer buying a generic top-level domain can choose thousands of registrars. The current AllDomains price table lists .com registration and renewal at UZS185,000 and .net at UZS197,000. UzCloud advertises .com at UZS180,000 before VAT. A difference of UZS5,000 is too small to support a moat by itself, especially when promotions, tax treatment, payment methods and renewal terms vary.

The national domain is also competitive. aHOST lists .UZ at UZS27,000 a year. Domains.uz lists the same UZS27,000, UzCloud lists UZS30,000 before VAT, and the .UZ administration's current registrar page names more than 20 active providers. SUVAN NET may have a large installed base, but the published retail price is a market price rather than a visible monopoly rent.

Accreditation also imposes a floor under cost. ICANN's fiscal 2026 fee schedule sets a $4,000 annual accreditation fee, a variable per-registrar component and a $0.20 charge on each qualifying annual increment of an add, renewal or transfer. Registrars must also operate registration systems, maintain RDAP, place data into approved escrow, handle abuse, preserve records and follow registry and consensus policies. AllDomains' offer promises escrow and provides a complaint process that can take up to ten business days.

At high volume, those fixed obligations become cheap per domain. At low volume, they erode the margin on every UZS185,000 sale. This is where SUVAN NET's reported customer base matters. If the company can migrate a meaningful share of existing aHOST customers and reseller partners onto direct AllDomains transactions, accreditation can be a valuable distribution asset. If volume is thin or registries retain most of the retail price, the credential becomes an expensive badge.

The evidence therefore supports a real strategic upgrade but not its return. The facts that would settle it are active generic-domain count, gross registrations, renewal rate, transfer-in and transfer-out volumes, registry cost per zone, support tickets per thousand domains and contribution margin after ICANN and escrow costs. None is public.

The network exists, but address space is not customer demand

SUVAN NET's resource footprint is substantial for a private regional host. The RIPE Database record for AS35326 links the autonomous system to the company, taxpayer number 301551793 and local internet registry organisation ORG-NL314-RIPE. The organisation record dates from March 2017; the current autonomous-system record was created in March 2019. Those dates show registry administration, not the formation date of the legal company.

At 08:00 UTC on 10 July 2026, RIPEstat's routing-status sample showed four IPv4 prefixes covering 2,048 addresses, one IPv6 prefix equivalent to 524,288 /48 networks, and two observed neighbours. IPv4 routes were visible to 326 of 327 returned full-feed peers and the IPv6 route to all 321. This is strong evidence that the network is globally visible. It is not an uptime guarantee and says nothing about traffic or customer count.

The announced-prefix list identifies 37.153.159.0/24, 85.204.79.0/24, 89.39.94.0/23 and 185.196.212.0/22. Together they contain 2,048 IPv4 addresses. It also shows 2a14:4a00::/29, an IPv6 allocation created in October 2023 but first visible in the returned two-week window on 9 July 2026, one day before this article's evidence date.

That new route could support long-term hosting growth, customer assignments and modern network operations. A /29 is extremely large relative to the current IPv4 footprint because IPv6 allocations are designed for hierarchical assignment rather than address scarcity. It does not mean 524,288 customers, networks or servers are in use. The first commercial question is how much of the allocation has been assigned to paying services, not how many theoretical /48s it contains.

IPv4 is different because the resource is scarce and each virtual server can consume a dedicated address. The aHOST VDS offer includes one address per server. A pool of 2,048 globally visible addresses gives management room to host many directly addressed services, subject to infrastructure, abuse reputation and internal use. It can reduce dependence on leasing addresses from another host. It still does not reveal utilisation, acquisition cost, revenue per address or whether some space is reserved.

The route-security evidence is weaker than the route count. Individual RIPEstat RPKI checks for all five announcements returned "unknown" with no validating route-origin authorisation. That is not the same as "invalid": the routes are not rejected as conflicting with a signed authorisation. It means the cryptographic origin protection was not present in the returned checks. For a registrar and host selling reliability, creating appropriate authorisations would be a low-ambiguity sign of operational discipline.

The registered policy is also stale. The autonomous-system record declares import and export relationships with AS31492 and AS12365, while RIPEstat's current consistency response sees AS31203 and AS34250 in routing but not in the registered policy. All five prefixes are present both in observed routing and the RIPE routing registry, which is positive. Updating the autonomous-system policy would bring the supplier description into line with current observation.

These details matter because resource ownership is sometimes mistaken for independence. SUVAN NET can originate its own routes and move address space between compatible upstream arrangements. That is useful control. It cannot reach the rest of the internet without external networks, power, routers, fibre and skilled operation. The number resources are tools used by the company; they are not the company and they do not earn money while idle.

Two neighbours do not yet equal two resilient paths

The current neighbour observation names AS31203 and AS34250. Public registry and routing sources identify the first as Sharq Telekom and the second as Uzbektelecom. In the returned sample, Sharq Telekom appears on the IPv4 side and Uzbektelecom on the IPv6 side. There is no evidence in that response of two independently observed upstreams for each protocol.

This is a narrower resilience position than the headline count of two neighbours suggests. If IPv4 services depend on one observed external network, the second IPv6 neighbour does not provide automatic IPv4 failover. The same logic applies in reverse. Private links, backup sessions and routes below the collectors' visibility threshold may exist, but public evidence cannot assume them.

Physical diversity is a separate question again. Two commercial suppliers can share a duct, building entrance, power feed or cross-connect. A route collector identifies adjacent autonomous systems, not fibre paths. SUVAN NET does not publish committed bandwidth, port capacity, utilisation, latency, packet loss, failover results or physically disjoint routes. Its company-written profile on Trustpilot claims multiple independent internet channels and redundant power, but no technical disclosure supports the scope of that claim.

Uzbekistan's policy environment now creates an opportunity to improve the position. Uzbektelecom's 2024 sustainability report says private providers were to be allowed to connect directly to international internet channels from 1 January 2025, ending the prior monopoly arrangement. A May 2026 parliamentary report said national international-connectivity capacity had reached 4,400Gbps in 2025, up from 1,200Gbps in 2020.

Liberalisation does not make direct international capacity free. A smaller operator has to buy cross-border transport, ports, equipment, colocation, support and enough traffic to justify them. The current observations still place domestic Sharq Telekom and Uzbektelecom next to SUVAN NET rather than demonstrating a direct foreign carrier. Management should compare the cost of a second domestic path, direct cross-border capacity and a content or exchange connection against the revenue at risk in an outage.

For low-priced shared hosting, complete independence may be uneconomic. For government, payment or business-critical contracts, measurable diversity can justify a premium. The strategic answer is product segmentation: do not finance an undefined enterprise promise from entry-level hosting prices. Sell a tested resilience tier to customers whose downtime cost is high enough to pay for it.

Public prices reveal the cost squeeze

The current aHOST tariff ladder exposes the economic trade-off more clearly than its corporate description. Local shared hosting in the TAS-IX network starts at UZS33,150 a month for 500MB and two sites. A 1GB plan is UZS42,075, 5GB is UZS146,625 and 50GB is UZS701,250. The site presents traffic and mailboxes as unmetered while increasing the permitted sites, databases and subdomains.

The company's German-hosted alternative is cheaper before a scheduled August 2026 adjustment. It starts at UZS22,000 for the same 500MB and two-site headline limits. The 5GB tier is UZS98,000 and the 50GB tier UZS500,000. The Uzbek entry plan is about 51% dearer than the German one, and the local 5GB plan is about 50% dearer. That differential is an unusually clear measure of the local-residency and local-network premium in SUVAN NET's own shop.

The premium may be rational. A local site can load quickly for Uzbek users, transact with a local provider and keep data in-country where required. It may avoid international-path congestion. Yet the customer has to value those benefits because the foreign option is sold by the same company with the same account relationship. SUVAN NET is effectively competing against its own lower-cost overseas capacity.

The 6 July price notice adds the other side of the risk. It says German shared-hosting prices will rise from 1 August because foreign suppliers increased their prices and inflation rose. SUVAN NET can diversify location and lower some infrastructure costs by hosting in Germany, but it imports the supplier's pricing decision and currency exposure. A domestic invoice in soum does not remove a euro- or dollar-linked input.

The VDS table shows that international bandwidth can overwhelm compute revenue. The entry VDS costs UZS125,000 a month for one processor core, 1GB of memory, 10GB of SSD storage, one address and 10Mbps internet. Moving to 15Mbps costs an additional UZS400,000 a month; 20Mbps costs UZS800,000; 35Mbps UZS2 million; and 100Mbps UZS6.3 million. The first extra 5Mbps costs more than three times the base virtual server.

Those are retail add-on prices, not SUVAN NET's wholesale transit cost. They still reveal management's pricing signal: scarce international capacity is treated as a far more expensive input than a small slice of compute. A customer with bandwidth-heavy traffic can be unprofitable if the base plan's "unlimited" wording is interpreted as a guarantee rather than a traffic quota subject to the port ceiling.

The same page monetises labour. One hour of server work costs UZS200,000. Monthly administration packages begin at UZS1.5 million for up to ten hours and reach UZS9.6 million for 80 hours. This is sensible separation. The UZS125,000 virtual machine cannot include unlimited skilled intervention; support beyond initial setup has to be priced as a professional service.

The margin test therefore has three parts. Compute density must be high enough that many virtual servers share hardware without unacceptable contention. Bandwidth must be oversubscribed carefully enough that ordinary usage fits contracted capacity. Support must be bounded so that one demanding customer does not consume several months of subscription revenue. No public metric shows SUVAN NET's performance on any of the three.

Competitors set uncomfortable reference prices

SUVAN NET does not price in isolation. Hostmaster's current VDS table advertises UZS119,900 a month for two processor cores, 2GB of memory, 30GB of SSD storage, one address and 100Mbps internet and TAS-IX access. That is UZS5,100 less than aHOST's entry VDS while advertising twice the cores and memory, three times the storage and ten times the international port speed.

Airnet asks UZS200,000 for two cores, 4GB of memory, 50GB of SSD storage, 250Mbps internet and 1Gbps domestic exchange access. At UZS75,000 above aHOST's entry price, it advertises much more capacity. Hostmaster's UZS299,900 tier advertises four cores, 4GB of memory, 100GB and 100Mbps, while aHOST asks UZS320,000 for two cores, 2GB, 100GB and 10Mbps.

Published specifications are not measured performance. Providers can differ in processor age, storage contention, support, backups, address reputation, availability and oversubscription. A 100Mbps port does not mean a customer can sustain 100Mbps at all times. The comparison nevertheless matters because many buyers screen on the visible table before testing quality. SUVAN NET has to explain why local support, reliability or bundled domains justify a weaker headline ratio.

Shared hosting is equally exposed. Hostmaster advertises a 1GB plan with unlimited sites, SSL and traffic for UZS19,900 a month and a seven-day test. aHOST's 1GB local plan is UZS42,075 and limits the account to five sites. Domains.uz advertises a 1GB plan for UZS5,000, though a low price says nothing about workload limits or service quality. A price-sensitive brochure-site customer has several substitutes.

Larger local cloud platforms compete on resilience rather than only price. UzCloud says it operates five Uzbek data centres separated by more than 300 kilometres, serves more than 2,000 customers and offers a 99.9% service commitment. Its hosting starts at UZS120,000 for three months with 5GB. Those are company claims, but they establish the competitive message: geographic separation, sovereign hosting and a broad product stack.

Global cloud and European dedicated-server companies are substitutes for customers that do not require domestic storage or local contracting. The 2026 change in Uzbek data law widened that option for many workloads. A developer can also split the stack: register a domain with SUVAN NET, place compute abroad, use a global content-delivery network and keep only regulated records locally. Every layer that can be unbundled limits SUVAN NET's pricing power.

The company should therefore avoid competing as a generic unit of processor, memory and disk. Cloud-scale providers and aggressive local hosts can usually win that comparison. Its defensible offer is operational convenience for a defined customer: local contract, domain, DNS, hosting, support and compliance in one account. The convenience must be measurable enough to survive a visible hardware-price disadvantage.

The June outage shows who carries the reliability bill

On 4 June 2026, aHOST reported interruptions affecting part of its cloud infrastructure after a power failure caused an accident in one cluster. A follow-up notice said restoration was completed before midnight that day and that the affected services had returned. The notices are useful because they identify a real failure domain and a restoration action rather than making an abstract availability claim.

The incident does not establish poor long-term reliability. One disclosed outage cannot supply an annual availability rate, and the start time is not stated. It does show that a cluster and its power dependencies can interrupt customer services despite the company's public claims about redundancy. The quality question is what failed over automatically, what did not, how long each customer was unavailable and what was changed afterward.

Customers carry the immediate cost of downtime: missed orders, unavailable websites, interrupted email, failed integrations and staff time. SUVAN NET carries restoration labour, replacement or repair cost, possible service credits and churn. Hardware and power suppliers receive payment before the company knows whether the affected customer will renew. The owner carries the residual loss when the subscription does not cover the incident.

This asymmetry is why cheap hosting can be dangerous. The monthly fee is capped; the restoration obligation is not. A UZS33,150 account that needs hours of manual work after a cluster failure cannot fund itself. The economics depend on automation, shared repair across many accounts, clear responsibility boundaries and enough gross margin across the whole cohort.

The company does not publish an availability history, incident count, recovery-time objective, recovery-point objective, backup success rate, power autonomy, service-credit schedule or data-centre certification. Its CONTRACTS page says documents are stored in a reliable domestic data centre meeting security requirements, but does not name the standard or independent assessor. These gaps do not prove weakness. They prevent a buyer from pricing reliability.

The most useful next disclosure would be a short incident report. It should state the failure's start and restoration times, affected services, customer impact, backup behaviour, root cause, remediation and whether the relevant power path is now independent. Publishing monthly availability by product would make local accountability a product rather than a slogan.

Growth in the market does not decide who captures it

The demand backdrop is favourable. Uzbekistan's National Statistics Committee says communication and information services reached UZS79.5 trillion in 2025, 22.7% more than a year earlier. Tashkent accounted for UZS57.8 trillion, reinforcing the importance of the capital where SUVAN NET is based.

The national domain is also expanding. The .UZ administration reported 159,562 active domains on 9 July 2026 and separately reported net growth of 15,212 since the start of the year. More domains create more potential renewals, DNS accounts, mailboxes and websites. They also attract more registrars and hosts.

The broader software economy can feed infrastructure demand. IT Park said 481 export-oriented companies joined in the first seven months of 2025, including 232 foreign companies, and that members exported to more than 90 countries. Exporters need domains, development environments, email and production systems. Some will want local support; many will benchmark against global infrastructure and require higher reliability than a simple website.

Supply is growing at the same time. The Ministry of Digital Technologies signed a June 2026 memorandum with Oracle and DataVolt around cloud infrastructure and a modern data-centre ecosystem. UzCloud markets multi-site sovereign infrastructure. Established registrars bundle hosting. Local VDS providers compete aggressively on port speed and compute. The market can grow rapidly while an undifferentiated provider's price falls.

SUVAN NET benefits from an installed domain base and years of local operation. It can sell to customers that already trust its billing account. Its reported 6,000 legal-entity customers are potentially more valuable than a much larger group of one-domain individuals because businesses can buy servers, administration and multiple renewals. That conclusion remains conditional because the company does not define active status or revenue per cohort.

The correct management response is to separate growth from value creation. More registered domains are valuable if renewal contribution exceeds acquisition and support. More virtual machines are valuable if utilisation and bandwidth produce a positive gross margin. More government customers are valuable if contract pricing covers compliance, service and payment delay. A headline customer count without those denominators can rise while return on capital falls.

Regulation creates a local advantage, then narrows it

For several years, strict data localisation gave Uzbek hosts a strong demand tailwind. The 2021 regime required personal data of Uzbek citizens to be collected, systematised and stored using technical means in Uzbekistan. That made a foreign-only cloud difficult for businesses holding customer records and strengthened the value of a domestic data centre.

Law ZRU-1125 changed that position on 27 March 2026. The Ministry of Justice's public explanation says mandatory domestic storage now applies to biometric data, genetic data and data of people using telecommunications operators active in Uzbekistan. Other personal data may be stored and processed abroad when the foreign jurisdiction offers equivalent protection, approved contractual or corporate rules are used, or approved international standards are met.

This is a strategic loss of regulatory shelter for generic hosting. A local retailer, software company or exporter may now be able to place more workloads abroad if it satisfies the new conditions. SUVAN NET's own German hosting becomes more usable, but global competitors gain the same opening. Domestic location must compete on latency, support, contract convenience and resilience rather than relying on a blanket legal requirement.

The local advantage remains stronger for telecom-related data and sensitive categories. Government bodies and regulated customers may impose additional requirements. A company that can document domestic storage, access controls, backup and incident handling can still win business. The key word is document: a local street address and an Uzbek IP do not themselves establish compliance.

Domain accreditation adds a different regulatory stack. ICANN and registry policies govern registrant records, transfers, abuse and data escrow. Uzbekistan's .UZ administration governs national-domain registration and registrar contracts. The company holds a data-network licence. Each credential opens a revenue channel and creates a continuing cost of compliance.

Geopolitics enters through suppliers and routes. German hosting exposes SUVAN NET to foreign data-centre and currency costs. International domains expose it to registry pricing in foreign currencies. Cross-border connectivity depends on carrier and transport arrangements. Domestic law can change the relative value of local infrastructure quickly, as the March reform demonstrated. Management should treat regulatory advantage as a renewable contract with policy, not a permanent moat.

Customer concentration is the largest commercial unknown

The company-reported customer mix sounds broad: more than 40,000 customers, 6,000 legal entities and 500 state organisations. Public procurement records confirm at least some government demand. A 2024 government purchasing report names SUVAN NET in a UZS5.22 million web-hosting transaction, and the Anti-Corruption Agency reported a roughly UZS2.02 million domain purchase for 2024. These examples establish a real public-sector channel, not its scale.

No source shows the ten largest customers, revenue share by customer, revenue by brand, or the split between domains, shared hosting, servers, development and software. Forty thousand low-value domain holders could produce less revenue than a small number of managed-server customers. Six thousand legal accounts could be active businesses or historical billing profiles. Five hundred government organisations could buy one domain each or substantial infrastructure.

Contract durability varies by product. A domain normally renews annually and can be transferred. Shared hosting can move after files, databases and mail are migrated. A managed server is stickier because configuration knowledge sits with the provider. An electronic-document system can become deeply embedded in process, but may also bring higher support, security and regulatory obligations.

AllDomains' partner terms reduce contractual friction but also reduce lock-in. There is no subscription fee, and the partner pays per registration from a prepaid balance. That is attractive for acquisition. A reseller can also stop placing new orders if price, service or integration disappoints. The application interface becomes durable only when it is reliable and switching it is more expensive than the saving elsewhere.

The portfolio can reduce concentration if the cohorts are genuinely independent. Domain demand, government hosting, web studios and software clients need not fail together. It can also create hidden concentration in infrastructure: many products may depend on the same cluster, power source, control panel, staff or upstream. The June incident showed that one cluster could affect multiple services.

The decisive disclosure would be recurring revenue and gross margin by cohort, with the top-ten customer share and annual renewal or churn. Without it, reported customer scale is evidence of distribution, not proof of durable economics.

Unofficial signals are positive and too small to settle the case

Public review platforms show favourable sentiment. Trustpilot displayed a 4.6 score from 28 reviews, with 96% at five stars. WHTop displayed an average of 9.6 out of 10 from 82 reviews, 81 recommendations and one opposed review. Reviewers often mention fast support, an easy panel and convenient domain management.

These are useful demand signals because convenience and support are exactly where a regional provider should differentiate. They are not an uptime study. The samples are tiny relative to the company's claimed 40,000 customers, many reviewers have only one review, and many WHTop posts are clustered in late March and early April 2026. The platforms themselves warn that user opinions are subjective.

The company's public Telegram channel had 952 subscribers in the observed preview and markets hosting, domains, site creation and direct international-registry connections. That establishes an active customer-acquisition channel, not conversion or retention. Sparse negative commentary cannot be interpreted as low outage rates because many customers never post publicly.

The official June outage notices deserve more weight than generic praise because they identify an actual operating event. The responsible conclusion is therefore balanced: users publicly value the support and interface, while the disclosed power incident shows that reliability still has to be measured. Reviews support the local-service thesis; they do not prove the margin or resilience thesis.

Evidence register

The main records answer different parts of the economic question and should not be combined into claims they cannot support.

Evidence	What it supports	What it does not prove
Corporate-data record	2010 registration, taxpayer number, activity, reported capital, management and IT Park status	Audited ownership, revenue, cash or profit
RIPE member page	Current RIPE membership, name, address and Uzbek service context	Hosting sales, transit products or profitability
AS35326 registry search	Autonomous-system identity, local internet registry status and registered policy	Current suppliers, traffic, capacity or physical routes
Routing status	Four IPv4 routes, 2,048 addresses, one IPv6 /29, high visibility and two observed neighbours	Uptime, utilisation, contracts or customer demand
Routing consistency	All five routes present in routing and registry; current neighbours absent from registered policy	Route security, physical diversity or supplier terms
aHOST home and tariffs	Current product range, list prices, local/German location choice and company-reported customer counts	Paid users, realised price, revenue or margin
aHOST VDS details	Compute, address, international-bandwidth and administration prices	Wholesale cost, utilisation or delivered performance
AllDomains home	Reseller positioning, automation, 1,000-plus zones and company identity	Transaction volume or direct-registry margin
IANA registrar record	Current accredited status, IANA 4327 and RDAP endpoint	Commercial success or compliance quality
.UZ registrar card	Valid national registrar contract and matching legal identity	Current .UZ portfolio or market share
June interruption	Power-related failure in one cloud cluster and affected services	Annual availability or precise outage duration
June restoration	Company-reported restoration before midnight on 4 June	Independent verification or permanent remediation
Hostmaster VDS prices	A close local price and specification substitute	Comparative real-world performance
UzCloud profile	Competitor's multi-site, domestic-cloud and service-commitment positioning	Independently measured availability or customer quality
2026 data-law summary	Narrower mandatory local-storage categories and foreign-processing conditions	Compliance of any specific SUVAN NET customer

What would change the judgment

The current judgment is that SUVAN NET has a defensible distribution and workflow position in domains, but has not shown that its infrastructure earns more than the cost of staying relevant. Several disclosures would turn that conclusion into a stronger positive or negative one.

First, publish the revenue bridge. Separate .UZ domains, generic domains, shared hosting in Uzbekistan, shared hosting in Germany, virtual servers, dedicated servers, administration, certificates, software and development. Show recurring revenue, recognised revenue, gross margin and renewal or churn for each. This would reveal whether accreditation is improving mix or merely adding compliance cost.

Second, define the customer count. Report paid active customers, domains under management, active hosted accounts and active servers. Separate individuals, legal entities and government bodies, and disclose the top-ten revenue concentration. Forty thousand becomes economically meaningful only when the renewal and revenue denominators are known.

Third, disclose domain economics. Report annual adds, renewals, transfers in, transfers out, redemption events and gross margin by major zone. Show how many AllDomains partners are active and what share of transactions arrives through the application interface. A stable renewal rate above customer-acquisition needs would support the strongest part of the thesis.

Fourth, disclose infrastructure utilisation. Publish physical host count, saleable versus used processor and memory capacity, storage utilisation, address utilisation and peak bandwidth. Revenue per physical server and per occupied IPv4 address would show whether the resource footprint is productive. The new IPv6 /29 should be accompanied by customer-adoption and assignment data rather than a capacity headline.

Fifth, publish cash and capital needs. Audited revenue and operating profit are useful; operating cash after maintenance capital is better. Show power, transit, foreign data-centre, registry, software-licence, hardware and payroll costs. Identify foreign-currency exposure and the amount of German supplier increases passed through to customers.

Sixth, make reliability measurable. Report monthly availability by service, material incidents, recovery time, recovery point, support response, backup success and service credits. Explain the June power failure, the redundancy that operated, the redundancy that did not, and remediation. State whether critical clusters have independent utility, generator, battery and distribution paths.

Seventh, prove network diversity. Publish contracted upstream count by IPv4 and IPv6, aggregate capacity, peak utilisation, failover tests and whether principal paths are physically disjoint. Update the RIPE registered policy to match current neighbours and create appropriate RPKI authorisations. If direct international connectivity has been bought under the liberalised regime, state its capacity and economic purpose.

Eighth, disclose service-level segmentation. Entry shared hosting should not silently finance enterprise obligations. Define a standard product, a managed product and a resilient product with different support, backup, route and restoration commitments. Report the gross margin of each. A customer with a high downtime cost should be able to buy a measurable outcome rather than infer it from the word "cloud."

A positive judgment would follow if active domains and customers renew at high rates, AllDomains transaction volume grows, recurring gross profit covers ICANN and RIPE costs, local hosting retains customers despite its premium, and the company produces positive operating cash after resilience investment. It would be strengthened by two observable upstreams for each protocol, valid route-origin authorisations, a current routing policy and no repeat of the June failure mode.

A negative judgment would follow if customer growth is mostly low-value registrations, domain transfers out accelerate, hosting prices cannot recover supplier increases, international bandwidth remains too expensive for modern workloads, or utilisation is too low to absorb hardware and power. Continued cluster outages, rising customer concentration, weak cash conversion or a large stock of idle number resources would show that control has become cost rather than advantage.

Relevance has to be renewed like a domain

SUVAN NET has done more than collect internet credentials. It has built a local customer base, operated hosting in two countries, maintained an autonomous network, become a national-domain registrar and crossed the more difficult threshold into ICANN accreditation. These are real capabilities and they can reinforce one another.

The economic hierarchy matters. The IPv4 pool supports servers but does not create demand. The IPv6 allocation supports future growth but is not evidence of adoption. Two observed neighbours create external reach but do not establish resilient dual-stack paths. A local data centre satisfies some buyers but costs more in the company's own tariff book than its German alternative.

The most promising asset is the workflow around the infrastructure: one local relationship for domain registration, renewal, DNS, hosting, support and billing, with a reseller interface that can multiply distribution. That is harder to compare in a simple processor-and-memory table. It can justify a premium if customers renew and if support solves problems faster than a distant alternative.

Management should resist treating breadth as strategy. Every extra product adds a supplier, control panel, policy, failure mode or support obligation. The portfolio creates value only where products share customers and infrastructure without sharing too much risk. The June cluster failure is a reminder that technical concentration can sit beneath commercial variety.

The fairest conclusion is conditional. SUVAN NET has enough differentiated demand potential to avoid being a pure infrastructure price-taker, particularly in domains and local reseller services. It has not published the evidence needed to show that this potential becomes value creation. Current price comparisons suggest that generic hosting and virtual servers are already price-taker products, while direct registrar status offers the best chance to earn a return above commodity compute.

Relevance is not secured by an allocation or accreditation once. It is renewed when customers pay again, suppliers are covered, incidents are absorbed and cash remains for the owner. Until SUVAN NET discloses those outcomes, its resource-holder status is a credible operating foundation and a visible fixed-cost burden, not proof of margin.