Summary

  • NTCOM looks viable at the company level because public financial aggregators report revenue above one billion rubles and high profit in 2024, followed by a material but still profitable step-down in 2025. That evidence does not prove that each cheap recurring VPS invoice is attractive; it suggests the margin is likely held by a wider mix of hardware, integration, hosting, software, private-network and corporate-service work.
  • The stress point is not whether NTCOM can sell access or servers. Its own sites show that it can. The harder question is whether low monthly customer revenue leaves enough retained value after transit, peering, IP resources, storage wear, power, support, lawful-communications compliance, abuse handling and customer collections. The answer is conditional: yes where utilisation is high and support is controlled; no where customers buy the smallest packages and consume manual service.
  • The network evidence is real but narrow. AS41853, RIPE LIR status, two /21 IPv4 allocations, RPKI-valid routing evidence, Moscow M9 presence and open-peering posture support the claim that NTCOM is an operating network, not just a reseller label. The same evidence also shows a Moscow-centered footprint with no visible IPv6 scale and modest public traffic signals.
  • The judgment would change quickly if NTCOM disclosed service-line revenue, churn, top-customer concentration, support tickets per customer, wholesale-capacity contracts, capex, hardware refresh policy, licence renewals and the split between recurring telecom revenue and one-off equipment or integration revenue.

Start with the invoice, not the brand. NTCOM's hosting arm advertises a starter virtual-server plan priced at 209.70 rubles per month, with one Intel Xeon virtual CPU, 256 MB of DDR4 memory, 40 GB of SAS plus SSD-cache storage, one dedicated IP address, a 100 Mbit/s connection, unmetered traffic, a three-day test period and DNS hosting for three domains. The headline is designed to feel frictionless. It is also the correct stress test for the company.

A provider can look healthy in a registry and still lose money on the small customer who treats a low-priced recurring bill as a right to unlimited engineer time, unlimited abuse handling, unlimited network attention and no hardware amortisation.

The first ruble in that invoice is not NTCOM's retained service value. Some of it is pass-through capacity and infrastructure cost: upstream transit or settlement-free interconnection that still requires equipment, cross-connects and engineering; power and cooling for servers; storage media that wears out; rack or facility cost; licences and compliance systems; payment collection; and an IPv4 address that is no longer an abundant commodity. Some of it pays for customer acquisition and retention.

Some pays for the support desk, especially when a customer buys the cheapest plan and asks for migration, DNS, PTR records, operating-system rescue, abuse mediation or refund handling. The retained service value is what remains after those claims. NTCOM's public economics should therefore be read backwards from this invoice: what parts of the business can absorb that cost stack without turning every small customer into negative contribution?

The company does not present itself as only a low-end virtual private server shop. Its corporate site describes operations since 2006 across telematic services for corporate and private clients, broadband internet access and data transfer at MMTS-9 and MMTS-10, protected corporate networks, virtual hosting, VPS and VDS, dedicated and virtual servers, customer equipment placement in its own data center, cloud services, database backup, switching equipment sales and its own anti-DDoS protection under the SOPSA name. Its English-language page adds a supplier story around Cisco and NORSI-TRANS equipment.

Its Russian page also refers to PCI-SIG membership. Those claims matter because a provider that only sells 209.70-ruble VPS packages would need an extreme level of automation and density to survive. A provider that also sells access, equipment, integration, protected networks, dedicated servers, backup, DDoS mitigation and corporate support can use the smallest hosting plans as a funnel, a retention layer or a utilisation tool.

That distinction is the core of the judgment. NTCOM's recurring customer revenue probably leaves enough contribution when the customer is buying a bundle of business continuity rather than just a machine. The same revenue probably does not leave enough contribution when the customer is buying the lowest plan and asking for the highest-touch service. The provider's own documents show that NTCOM knows this. The NT-VPS service rules reserve broad discretion around technical feasibility, resource overuse, abuse, content violations, excessive traffic patterns and support boundaries. Those are not decorative legal clauses.

They are an economic operating system. A low-ticket infrastructure provider needs contractual authority to stop customers from converting cheap capacity into unpaid support labour or network risk.

The identity boundary is also fairly clear. The directory entity is Limited Liability Company NTCOM, the Russian limited-liability company behind the NTCOM and NT-VPS public surfaces. Public corporate profiles identify the Russian legal form, registration in July 2006, INN 7706626963, OGRN 1067746863007, current director Georgiy Aleksandrovich Pankov and a stated 100 percent owner structure in Pankov's name. Several registries list the company as an active medium-sized business.

The same public material shows a useful address tension: NTCOM's official contact pages and NT-VPS footer continue to use the Bolshaya Novodmitrovskaya address, while recent registry and RIPE member records point to 17th Maryina Roshcha Drive, house 13, building 5. That does not prove operating weakness. It does mean the public web estate is not perfectly synchronised with the latest legal-address data, and for a regulated communications provider that is worth noting.

The control boundary is narrow. There is no public evidence of a dispersed board, institutional investor, strategic carrier parent or deep multinational capital base. Public profiles put control with a single individual, Pankov, who is also the director. That can be an advantage in a small infrastructure provider. Pricing, support policy, procurement, licence response and customer escalation can move without a corporate committee. It is also a concentration risk. If the owner's judgment is sound, the company can keep margin by refusing unprofitable accounts and by moving customers toward higher-value configurations.

If the owner's judgment is wrong, there is less visible governance buffer between a bad capital cycle and the network's service quality.

The financial surface is surprisingly strong, but it needs to be read carefully. RBC's company profile reports 2024 revenue of 1.738566 billion rubles, cost of sales of 782.846 million rubles, gross profit of 955.720 million rubles and net profit of 725.748 million rubles. That implies a gross margin around 55.0 percent and a net margin around 41.7 percent for 2024. Those are not ordinary margins for a pure commodity broadband or VPS provider.

They are more consistent with a mix that includes higher-margin software, hardware resale with favourable procurement, integration, protected networks, proprietary services or unusual one-off project economics. The point is not that the numbers are false. The point is that they do not look like the economics of a provider whose main engine is a low-price virtual-server plan alone.

The 2025 public figures make that caution sharper. Firmoteka reports 2025 revenue of 1.415960 billion rubles, down 18.6 percent, and net profit of 288.823 million rubles, down 60.2 percent, with 22 employees and a reported net margin of 20.4 percent. TBank similarly shows 2025 revenue around 1.41 billion rubles and profit around 288.82 million rubles. A company that can fall from 41.7 percent net margin to roughly 20.4 percent while remaining profitable may still be robust. It may also be exposed to project timing, equipment resale swings, cost inflation or the difficulty of maintaining extraordinary margins as the customer mix changes.

The margin is not a static truth; it is an outcome produced by a particular mix of invoices, utilisation and cost control in a particular year.

The employee count sharpens the interpretation. A 22-person company with more than one billion rubles of reported revenue is not behaving like a manual help-desk-heavy local ISP. It is either selling high-value equipment and projects, running a lean automated infrastructure operation, booking substantial revenue through a small commercial team, or some combination of those. That supports the argument that low-ticket recurring invoices are only one edge of the business. It also creates an operational question: if the firm wants recurring revenue to dominate, can it keep support labour from scaling linearly with customer count?

The NT-VPS site claims fast server creation, a custom control panel, KVM virtualisation and Ceph-based storage. Those are the right mechanisms if the business is trying to turn many small invoices into automated recurring gross profit. The risk is that automation claims are not the same as observed support productivity.

The infrastructure evidence is substantive. RIPE lists Limited Liability Company NTCOM as a Russian member serving the Russian Federation. RDAP and RIPE REST records link AS41853 to NTCOM's RIPE organisation object. IPinfo, RIPE and related routing sources show two allocated /21 IPv4 blocks: 91.102.152.0/21 and 93.189.40.0/21. Together they amount to 4,096 IPv4 addresses before deaggregation. BGP sources show the network originating 18 IPv4 prefixes and no visible IPv6 originated prefixes in several public views. IPinfo pages for the /21s identify the resources as RIPE-registered and RPKI-valid.

Hurricane Electric and PEER.AS show NTCOM's prefixes as visible in global routing. This is not evidence of national scale, but it is evidence of a real routed platform.

The routing posture also tells us what NTCOM is not. PeeringDB lists NTCOM with a Europe geographic scope, open peering policy, balanced traffic ratio, public facility presence at Moscow M9 and a traffic level in the 20-100 Mbps band. Hurricane Electric's view shows observed IPv4 peers including networks such as Citytelecom, Arelion, RETN and Russia's radio-frequency center, though observed BGP adjacency should not be treated as a contract list. IPinfo's traceroute evidence from Moscow reaches an NTCOM address in a small number of hops through Citytelecom. That is consistent with a Moscow-centered access and hosting network.

It is not consistent with a hyperscale cloud network or a nationwide retail broadband carrier. The economic implication is straightforward: NTCOM's defensible position has to come from local connectivity, customer intimacy, support familiarity, niche services and controlled cost, not from scale advantages.

The most important asset may be the combination of address space, local network know-how and a service stack that can be sold to small and medium organisations that do not want to assemble their own infrastructure. IPv4 addresses have become a real balance-sheet-adjacent resource for hosting providers. A dedicated IP attached to a small VPS looks trivial in a tariff table, but it carries opportunity cost. A provider that has 4,096 addresses can run a meaningful hosting and access business, yet every abusive customer, unpaid server, idle allocation or low-margin use consumes scarce inventory.

This is why the lowest VPS tier should not be treated as a simple monthly rent. It is a bundled claim on compute, storage, address space, bandwidth, support and compliance capacity.

NTCOM's own tariff design reveals the tradeoff. A starter VPS at 209.70 rubles per month is an acquisition product. The Optima plan at 689.70 rubles and the Pro plan at 1,889.70 rubles offer more memory, storage and CPU. A separate package page presents a 2,499-ruble monthly bundle that includes static IP, firewall, VPN, site, mail server and Nextcloud or Samba resources, with discounts for longer prepayment. The first plan tests the support burden; the larger bundles test whether NTCOM can lift customers from a commodity server into continuity services. The difference matters more than the nominal price ladder.

A customer paying a few hundred rubles a month can destroy margin with one support-heavy incident. A customer paying for a business bundle is more likely to justify manual attention, provided the support scope is disciplined.

That is why support labour is the hidden cost line. Reviews and public comments should be used cautiously, but they point to the same pressure. Hosting101's NT-VPS page shows a 4.5 out of 5 average across 45 reviews, with positive signals around anti-DDoS protection, responsiveness, own AS and IPv4 pools, and Ceph storage. The same review environment lists negative themes such as disk speed, control-panel usability, backup interruption, DNS and PTR friction and support delay. Hosting101's separate NT-COM data-center page has a very small voter base and includes downtime complaints.

Hostings.info shows no user reviews but lists tariff and service attributes. HostDB records the provider as NT-VPS with a small number of reviews and notes two upstreams with redundancy. None of these signals is audited. They are useful because they show the market's everyday invoice psychology: customers praise low price and quick support when things work, and they punish the provider when a cheap service requires manual intervention.

The support problem is not merely reputational. It is mathematical. If a customer pays 209.70 rubles per month, one non-trivial ticket can consume the monthly contribution before transit, power, depreciation, IP opportunity cost, billing and abuse handling are considered. Even at 689.70 rubles, the invoice requires high automation. At 1,889.70 rubles or 2,499 rubles, the provider has more room, but still not enough for unlimited concierge service. This is why NTCOM's documents matter.

By defining resource limits, content rules, abuse escalation, traffic conditions and technical feasibility, the company is trying to keep the invoice from being rewritten by customer behaviour after sale. The margin survives only if those rules are enforced with consistency.

Customer concentration is not visible enough to call. Public procurement records cited by TBank and other aggregators show two 44-FZ contracts: a server contract for 1.600351 million rubles and a communications-channel contract for 840,000 rubles. Xfirm and other profiles identify a 2022 server supply to the National Research University Moscow Power Engineering Institute, with an initial maximum contract price above the final award. These records prove that NTCOM can sell into public-sector or institutional procurement. They do not prove dependence. The combined visible 44-FZ amount is tiny next to 2024 and 2025 reported revenue.

If anything, the public-procurement trail suggests that known government contracts are not the main revenue engine. The unknown is private enterprise concentration: one or two large equipment, integration or protected-network customers could still explain a large share of revenue and profit.

Supplier dependence is easier to see in shape than in contract detail. NTCOM's official pages mention Cisco. NT-VPS names Cisco, Dell, HP, Intel and Supermicro as equipment foundations. The English NTCOM page mentions NORSI-TRANS for technical controls related to surveillance activities. Public BGP views show connectivity through or alongside networks such as Citytelecom, Arelion and RETN, though those observations should be read as routing evidence rather than definitive purchase contracts. Each category creates a different margin exposure. Hardware suppliers define replacement cost and lead time.

Upstream or peering partners define capacity economics and resilience. Compliance equipment and procedures define the right to operate. None can be ignored in an invoice-margin test.

The hardware angle is especially important in Russia. Server replacement is not just a technical lifecycle question. It is also a currency, sanctions, warranty, import, spare-part and substitution question. NTCOM's price ladder uses older Intel Xeon positioning on small plans, which may be economically rational if depreciated equipment is still serviceable. But customers do not pay less for downtime just because the provider sweats assets longer. The company has to balance utilisation against reliability. Replace too early and cash leaves the business. Replace too late and support tickets, outages and churn rise.

Public sources do not disclose NTCOM's depreciation policy, server age, storage endurance, power density or spare-part stock. That absence is not a minor footnote; it is one of the main uncertainties.

Capital structure is also opaque. The public accounts show assets and capital in the hundreds of millions to more than a billion rubles, but they do not give a clean service-line capex schedule. A hosting and access provider has at least three capital rhythms: network upgrades, compute and storage refresh, and facility or colocation commitments. NTCOM claims a data-center capability and public presence at Moscow M9. Market sources describe Russian data-center and colocation prices rising, capacity tightness, Moscow's importance and large operators expanding. For NTCOM, a tight Moscow data-center market is double-edged.

It can support pricing power for local hosting and colocation, but it can also lift power, rack, cross-connect and expansion costs. A small provider cannot assume that a cheap server invoice will cover future replacement if the capital cycle moves against it.

The market backdrop is favourable in demand and hostile in competition. ComNews reports that Russia's cloud-services market grew strongly in 2025 to 416.5 billion rubles, citing iKS-Consulting estimates and an expectation of continued growth. Yandex Cloud reports 2025 cloud revenue of 27.6 billion rubles, up 39 percent, with more than 51,000 customers and strong enterprise demand. Mordor Intelligence frames Russia's data-center market as growing through 2031, with Moscow dominant and IT and telecom as major end-user segments.

Kommersant reports Moscow and Moscow-region colocation price increases in 2025, with leading operators such as Rostelecom structures, IXcellerate, Atomdata, DataPro and Selectel controlling a large share of rack capacity. Those conditions help infrastructure sellers, but they also put NTCOM in the path of larger platforms with better procurement scale and broader compliance packaging.

The substitute question is brutal. A small business that wants one VPS can go to a national hosting provider, a cloud marketplace, a regional data-center provider or a large Russian platform. A company that needs domestic cloud, backup, DDoS protection or private connectivity can choose a national carrier or an enterprise cloud. A customer that is technical enough can rent bare metal or build in a larger cloud and avoid a smaller provider's support constraints. NTCOM's defence cannot be generic hosting.

It must be something more specific: fast local response, Russian-language support, familiar billing, address-space availability, Moscow interconnection, custom integration, protected networks, anti-DDoS know-how, and willingness to handle awkward small and medium customer needs that large platforms push into ticket queues.

The competitive picture also explains the 2024 margin puzzle. If NTCOM's revenue were primarily commodity VPS, larger platforms should pressure margins. If revenue includes equipment, system integration, proprietary anti-DDoS, protected networks or project work, higher margins are more plausible. The company's own OKVED history and current descriptions reinforce this ambiguity. Public profiles variously show wholesale software, computer production, printed circuit boards, communications equipment and data-processing or hosting activities. In April 2026, some public histories show a change toward computer production as a main activity.

These classifications do not tell us what customers actually bought in 2024 and 2025. They do tell us that NTCOM should not be analysed as a single-product access ISP.

Regulatory risk is not abstract. NTCOM operates under Russian communications licensing. Public profiles list three active communications licences with end dates in November 2026, January 2027 and February 2027. That creates a near-term renewal calendar. A licence problem would not merely raise compliance cost; it could impair the ability to sell the very services that make the invoice recurring. The company's official page says its telematics and data-transmission services are licensed by the federal communications regulator.

NT-VPS documents also give the operator broad rights to suspend or disconnect customers for illegal content, abuse, malware, network attacks, spam and rights violations. That is operationally necessary. It also means NTCOM must staff or automate a governance function around customer behaviour. Abuse and rights complaints are not edge cases for hosting; they are part of the cost model.

The lawful-interception and security-control surface deserves a hard, careful read. NTCOM's English about page refers to NORSI-TRANS as a vendor of technical controls for carrying out surveillance activities. In Russia, communications providers operate in a regulatory environment where compliance with lawful-interception and data-control requirements is not optional. That does not mean NTCOM is doing anything unusual or improper. It means the company has compliance obligations that a pure software reseller would not have.

The cost of those obligations should be allocated to the customer invoice, either directly through price or indirectly through higher-value service tiers. If the low-end hosting price cannot absorb that compliance surface, then margin must come from larger business accounts.

Geopolitical exposure is also present through hardware and network positioning. NTCOM names foreign hardware brands on its hosting site and claims Cisco-based network equipment on its company site. Russian technology import conditions have changed materially since 2022. Public sources do not provide NTCOM-specific purchase contracts or sanctions effects, so it would be wrong to assert a particular shortage. It is still fair to identify the exposure: any provider relying on branded servers, switches, CPUs and storage has to solve maintenance, replacement and firmware risk under a more constrained procurement environment.

A provider with high retained profit can buy ahead or shift suppliers. A provider whose low-end customers consume support and capital faster than revenue cannot.

Collections and payment friction sit below the more dramatic issues but matter just as much. Hosting providers sell many small recurring invoices. Some customers prepay, some churn, some abandon servers, some trigger abuse, and some dispute service quality. NT-VPS advertises discounts for longer prepayment on at least one package. That is economically rational: prepayment reduces collections risk and gives the provider working capital before hardware and support are consumed. The danger is that discounts move revenue forward while support and replacement obligations remain in the future.

Prepayment is healthy only if churn and complaint cost are controlled. Otherwise it becomes a way to borrow from tomorrow's margin.

The public reviews show exactly why customer psychology can break the model. Positive users praise speed, low price, stable operation, anti-DDoS protection and support. Negative users complain about slow support, refund refusal, disk speed, awkward panel functions and backup or DNS limitations. The right inference is not that either side is true in full. The right inference is that the provider's value proposition sits in a sensitive zone: customers are price-conscious, technically dependent and often not tolerant of operational nuance. A bigger platform can absorb some support anger through scale and brand.

A small provider has less room. Its best protection is to set expectations before the invoice is paid and to reserve manual help for accounts that justify it.

There is one clear positive indicator: NTCOM appears to have survived multiple cycles. The company dates to 2006, AS41853 was registered in 2006, its IPv4 allocations date from the mid-2000s, and the NT-VPS public presence has multi-year review and directory history. Longevity matters in infrastructure because bad operators often fail when abuse, hardware replacement and support debt accumulate. NTCOM's continuing active status, RIPE membership and profitable public accounts argue against dismissing it as a thin reseller. But longevity does not settle the current invoice question.

A provider can survive for years on project work while low-priced recurring services merely fill idle capacity or feed leads.

The strongest evidence for retained margin is the 2024 and 2025 profit line. Even after the reported fall in 2025, the company remained profitable at a level that many small infrastructure firms would envy. The strongest evidence against clean recurring-margin confidence is the mismatch between that profit line and the public service menu. A 20.4 percent 2025 net margin can coexist with a weak VPS plan if other segments carry it. The 2024 net margin above 40 percent almost demands that interpretation.

The safe conclusion is therefore not "NTCOM's recurring revenue is weak" or "NTCOM's recurring revenue is strong." The safe conclusion is that NTCOM's aggregate company margin is strong enough to support recurring services, but the low-end invoice by itself is not a defensible profit engine unless utilisation and support discipline are excellent.

The customer concentration question remains unresolved. Public procurement is too small to matter at the company scale. Hosted-domain counts from IPinfo indicate hundreds of domains, but hosted-domain counts are not customer counts and can be distorted by parked domains, customer concentration on a few IPs or service artefacts. The NT-VPS site claims large numbers of installed VPS instances and more than 150 dedicated servers, but those are company marketing claims and should not be treated as audited capacity. If the real recurring base is broad, churn risk is diversified.

If the real base depends on a few integration accounts or reseller relationships, reported revenue can move sharply when one project ends. The 2025 decline makes this a live question.

The supplier concentration question is similarly unresolved. Public routing views show a handful of observed network adjacencies. PeeringDB reports an open peering policy and a Moscow M9 facility entry. Company pages name major hardware brands. None of that tells us contract duration, committed information rate, payment terms, spare equipment, repair terms or SLA penalties. If NTCOM buys capacity flexibly and keeps enough alternatives, it can preserve margin by routing around price and quality problems. If capacity or facility access is tied to a narrow supplier set, the retained value from customer invoices can be squeezed quickly.

For a provider of this size, a single upstream or facility cost reset can matter.

The regulatory and court surface should be watched but not over-interpreted. Public profiles show active licences, a personal-data operator registry reference, a 2022 communications-control inspection in some aggregator records and arbitration matters including a 2026 neighbouring-rights claim and 2025 contract disputes. These are not, by themselves, evidence of systemic weakness. Russian companies of meaningful size often have court and inspection traces. The economic relevance is that disputes and compliance events consume management time and sometimes cash.

A provider whose support desk and compliance desk are already thin cannot afford too many such drains. A provider with strong retained profit can absorb them.

What would change the judgment? First, a service-line revenue split showing that most revenue is recurring hosting and access, not project or equipment revenue, would make the margin story stronger if profit remains high. Second, a churn table by tariff would show whether low-end customers are profitable acquisition or noisy leakage. Third, support tickets per active server and average first-response time would test whether labour is controlled. Fourth, top-ten customer revenue concentration would reveal whether 2025's decline was a broad market move or a project roll-off.

Fifth, capex and depreciation schedules would show whether the 2024 profit was harvested by under-replacing equipment. Sixth, wholesale-capacity and facility contracts would show whether bought connectivity and power costs are variable, fixed or exposed to renewal shocks.

Seventh, licence-renewal confirmation through 2027 would reduce regulatory overhang. Eighth, IPv6 adoption or a plan for address scarcity would show whether NTCOM is preparing for future hosting demand without overusing scarce IPv4. Ninth, objective uptime and incident records would test the review-page complaints. Tenth, a clear statement of the legal address across corporate, RIPE and hosting surfaces would remove a small but unnecessary credibility gap. None of these facts needs to be exotic. They are ordinary operating facts. Their absence is why the correct answer must remain conditional.

The final judgment is direct. NTCOM is not an empty shell and not merely a forwarding website. It has real Russian corporate standing, real routed resources, real public services, real licences, visible Moscow infrastructure ties, a long operating history and reported profitability. But the recurring invoice is not automatically attractive. The low-priced VPS invoice only works if the customer consumes a mostly automated slice of existing infrastructure and does not demand repeated manual service.

The better economic model is to use low-end hosting as a controlled entry point while retaining margin in higher-value services: protected networks, dedicated servers, backup, DDoS mitigation, business bundles, integration, equipment and local support for customers who want continuity more than the absolute lowest price.

NTCOM therefore passes the invoice test only with discipline. It must separate bandwidth and hardware pass-through from retained service value. It must keep utilisation high without letting noisy customers consume the support desk. It must treat IPv4 addresses as scarce inventory, not freebies. It must renew communications licences, keep lawful-service obligations funded, and manage hardware replacement under Russian procurement constraints. It must use Moscow interconnection as a strength without being trapped by Moscow power and rack-cost inflation.

Above all, it must avoid the classic small-provider mistake: pricing like a commodity platform while servicing customers like a bespoke integrator. If it avoids that mistake, recurring revenue can leave enough contribution. If it does not, the apparent strength of the company accounts will be carrying a weak invoice engine rather than proving one.

The invoice can be broken into three layers. The first is pass-through utility: power, space, network access, address allocation, storage wear and the basic right to keep a virtual or physical server reachable. The second is platform labour: provisioning, monitoring, billing, abuse response, DNS, reverse DNS, backups, virtualisation templates, kernel or console rescue and the control panel. The third is judgment labour: the engineer or account manager who decides whether a customer problem is a fair support request, an abuse risk, a product-design flaw, a collection dispute or a customer who should be moved to a higher plan.

The first layer scales with capacity. The second scales only if automation is good. The third does not scale cheaply. NTCOM's economic survival depends on keeping most low-end invoices in the first two layers and charging explicitly when a customer enters the third.

This is where utilisation becomes more important than the tariff table. A server host with unused CPU, memory, disks and IP addresses can sell a low-priced VPS and still earn contribution because the marginal cost is small. The same invoice becomes unattractive once the platform approaches the next hardware purchase, rack-power threshold or storage replacement cycle. NTCOM's public pages do not disclose utilisation, so the tariff must be read as a capacity-management instrument rather than a pure price promise. If the starter plan fills idle capacity and pushes a minority of customers into larger bundles, it is rational.

If it creates a large population of marginal accounts that require support and block address inventory, it is a subsidy. The public financials cannot distinguish those cases, but the 2025 decline makes the distinction material.

Churn has the same asymmetry. A customer on a cheap monthly service can leave quickly after a single bad support experience, but the provider still bears the fixed investment in platform, address space, documentation, billing and abuse procedures. Longer prepayment discounts improve cash timing, but they do not automatically improve economics if customers churn emotionally after renewal or if discounted revenue underfunds future support. For NTCOM, the best churn defence is not simply lower price. Larger competitors can copy low price or bundle it into broader cloud suites.

The better defence is predictability: customers should know what is self-service, what is paid support, what is excluded, what response time is realistic and what continuity service costs. Ambiguity sells the first invoice and damages the second.

Collections risk is modest per account but meaningful in aggregate. A single unpaid starter VPS is small. A pattern of abandoned servers, abuse-related shutdowns, disputed refunds and low-value tickets can produce a hidden receivables and labour drain. TBank's 2025 profile reports debtor and creditor balances, but the public data does not break receivables down by customer type or ageing. The invoice test therefore has to assume that collections discipline is part of the product. Automated suspension, clear prepayment terms, no-confusion refund rules and fast abuse handling are not just administrative details.

They decide whether small recurring accounts behave like annuity revenue or like a queue of mini-disputes.

The in-house alternative is real for NTCOM's better customers. A small company that needs one website will not build infrastructure. A technically capable customer that needs VPN, mail, storage, private networking and backups may decide to run equipment in a larger colocation hall, rent from a national provider, or move workloads to a domestic cloud. That customer compares NTCOM not only with other budget VPS hosts but also with the cost of control.

NTCOM wins when the customer values local competence, known people, Moscow connectivity, Russian-language support, custom networking and a provider willing to combine services without enterprise bureaucracy. NTCOM loses when the customer sees only a virtual server and can buy an equivalent instance from a larger platform with better documentation and perceived resilience.

The national-carrier alternative is harsher for corporate connectivity. Large Russian carriers and cloud platforms can sell bundled access, security, hosting, data localisation and account management. They also have procurement scale in hardware, facilities and compliance. NTCOM cannot beat that by pretending to be larger than it is. Its economic wedge is narrower: customers that are too specialised, too local, too price-sensitive for enterprise cloud, or too dependent on pragmatic support to be happy inside a national platform. Those customers can be profitable, but only if NTCOM prices the messy parts correctly.

A provider that quietly absorbs custom work to preserve a relationship will soon discover that relationship revenue is not the same as contribution.

The regulatory calendar should be treated as a management task, not a background fact. The active communications licences visible in public profiles run into late 2026 and early 2027. A well-run provider will have renewal files, compliance evidence and regulator communications under control long before the dates arrive. A poorly run provider will treat renewal as paperwork until it becomes a sales risk. Customers buying continuity do not care whether the problem is administrative or technical; they care whether the service remains legal and reachable.

For an infrastructure provider, the licence renewal process is part of the retained-service value. It must be paid for by the invoice even though the customer never sees it on a line item.

The same is true of abuse and rights complaints. Cheap hosting often attracts customers who want disposable infrastructure. The provider can make money from that only if abuse systems identify and remove bad accounts faster than they consume network reputation, staff time and legal attention. Public court and review signals do not prove an abuse problem at NTCOM, but NT-VPS's own rules show why the company reserves suspension and removal rights. Those rights protect the network's other customers. They also protect the margin.

If a provider is slow to remove abusive customers, upstreams, search engines, mail systems and regulators can impose costs on the whole platform. If it is overzealous, good customers leave. The retained margin sits between those two errors.

The final operating recommendation is therefore simple and severe. NTCOM should not measure recurring success by customer count, server count or invoice count. It should measure contribution per support hour, contribution per IPv4 address, contribution per rack-unit or power unit, contribution per abuse event and contribution per renewal cycle. Those measures would show whether recurring revenue is financing the operating surface or merely adding noise. They would also explain the 2024-to-2025 margin reset better than revenue alone. If contribution per support hour is rising, NTCOM is becoming a stronger infrastructure business.

If it is falling while revenue remains high, the company is buying top-line scale with future service debt.

On the evidence available, my base case is that NTCOM has enough economic substance to keep operating, but not enough public transparency to call the recurring engine self-proving. The business should be valued as a disciplined Moscow infrastructure and integration provider with a hosting arm, not as a pure-play regional ISP and not as a hyperscale cloud challenger. Its best invoices are probably not the cheapest ones. They are the invoices where access, servers, security, backup, private networking and support are bundled tightly enough that the customer pays for continuity rather than raw capacity.

That is where NTCOM can retain service value after bought capacity and equipment replacement. The weak invoices are the ones where the customer buys capacity at commodity price and then consumes bespoke attention. NTCOM's future margin depends on making the first category larger and the second category smaller.

Sources