Summary

  • RELCOM-SPB still controls a real routed asset: AS6672 is visible with two IPv4 prefixes, one IPv6 /32, valid RPKI, RIPE LIR status and observed upstream or neighbour relationships. The economic question is whether that footprint is being sold at enough utilisation to cover the non-optional costs of upstream capacity, ports, support, power, compliance and equipment renewal.
  • The public evidence points to a small, mature operator with useful infrastructure rights but weak scale economics. 2025 revenue of roughly 11.66 million rubles, a small net loss, low headcount, discontinued domain registration, conflicting licence-status signals and no visible downstream base make the investment case depend less on address ownership and more on customer retention, rack and port occupancy, and the ability to convert scarce IPv4 into recurring contribution without burning support time.

A routed customer is the right place to begin because it exposes the whole income statement. A business buying RELCOM-SPB's "INFOCHANNEL" service is not just buying Internet access. It is buying a dedicated channel, addresses, DNS support, the ability to connect remote offices, and someone local who will answer the phone when a route, mail account, port or address assignment stops behaving. The public tariff language says the dedicated-channel service runs continuously, 24 hours a day, every day of the year. It also says lines can be built on RELCOM-SPB's own resources or through other operators' networks.

That one sentence contains the profit problem. When a connection uses the company's own access assets, the retained revenue can be meaningful. When it uses leased last-mile or upstream inputs, the company is partly a broker of someone else's plant, and the gross margin must pay for sales, technical support, routing, billing, regulatory compliance and customer churn risk before any owner cash remains.

The same logic applies to a customer interconnection. RELCOM-SPB describes an office-unification product built as symmetric dedicated links in a star topology, with each client point connected to the Relcom network and the remote LANs joined at the IP layer. Traffic between the client's own sites is not charged. A customer can also take Internet access through the same arrangement. Economically, that is a sticky product if the customer values a local operator, static addressing, an existing site relationship and simple support. It is a weak product if the customer is only buying a commodity circuit that another carrier can undercut.

The company therefore has to convert local familiarity and address-resource control into a price premium or a low-churn account base. Without that, every routed customer becomes a contest between small-provider service and larger-carrier input cost.

The route table confirms that the company still has more than a token network. RIPEstat saw AS6672 announcing three active prefixes in early August 2026: 212.113.96.0/19, 37.202.8.0/21 and 2a02:9d8::/32. The IPv4 total is 10,240 addresses. RPKI validation for the three active origin/prefix pairs is valid. RIPEstat's routing-status view showed near-full visibility from full-feed peers, with 327 of 328 relevant IPv4 peers seeing the route set and 320 of 322 IPv6 peers seeing the IPv6 route. Hurricane Electric's BGP page likewise shows three originated prefixes, all valid under RPKI, and 10,240 originated IPv4 addresses.

IPinfo's public ASN page classifies the ASN as hosting, reports 1,655 hosted domains, lists the same two IPv4 blocks, and gives router evidence in Saint Petersburg. These are not proof of profit, but they are proof that the company still operates a globally visible network rather than merely retaining an old legal shell.

The value of that network is bounded by observed topology. RIPEstat's neighbour view reported 12 unique observed neighbours in August 2026. IPinfo listed three upstreams: CJSC RASCOM, INETCOM CARRIER LLC and BiMajLink. Hurricane Electric's view showed a broader peer count, including an IPv4 peer list and a PITER-IX St. Petersburg exchange presence at 185.1.152.12. The older RIPE aut-num object, by contrast, still names imports from AS2118, the SPB route server, RETN and an AS48268 relationship.

RIPEstat's routing-consistency data makes the gap visible: some whois-import peers are not in the current BGP view, while many observed BGP neighbours are not in the whois policy. That is normal in small networks when policy objects lag operations, but it matters economically. A buyer of transit, colocation or routed access wants evidence that paths are diverse and maintained. Stale policy objects reduce signalling quality, even when the routes themselves are visible.

The company is also small. Russian contractor and company-profile records identify the business as an active limited liability company registered in 1997, with OGRN 1027810245869, INN 7826683201 and a legal address now shown as Kirochnaya Street 9, letter A, room 3-N in Saint Petersburg. Public profiles identify Ali Alimovich Urusov as director, and T-Bank's contractor page records two individual founders with 83 percent and 17 percent shares. RBC's company page reports average headcount of six employees. Rusprofile describes the company as a microenterprise.

Public financial summaries put 2025 revenue around 11.66 million to 11.67 million rubles, down from roughly 12.39 million rubles at the start of the year, with 2025 net profit around negative 215,000 rubles. RBC reports 2025 cost of sales of 7.612 million rubles and gross profit of 4.055 million rubles. That is a gross margin of about 34.8 percent before the rest of the operating burden. The after-cost answer is therefore not generous: the asset exists, but the scale does not leave much room for mistakes.

This is why the company's address-resource tariff is more than a side note. RELCOM-SPB's own numbering-resource page says it is a RIPE NCC Local Internet Registry. It offers registration and support of autonomous systems, provider-aggregatable IPv4 and IPv6 addresses, and provider-independent IPv6 blocks. It prices an AS registration at 1,000 rubles one time and AS support at 970 rubles per month. IPv4 recurring charges step from 170 rubles per month for a single /32 to 19,435 rubles per month for a /22, with intermediate prices for /31, /30, /29, /28, /27, /26, /25, /24 and /23.

The tariff also says IPv4 blocks are provided only where technically possible. IPv6 PA carries no monthly fee, while IPv6 PI /48 support is 690 rubles per month. For colocation and virtual-server customers, a /30 can be bundled. For dedicated access and office-interconnection customers, a /29 can be bundled. The pricing tells us how the company thinks about scarce resources: IPv4 is both a retention tool and a billable asset.

The monetisation problem is that 10,240 addresses do not automatically create a high-value business. At the posted tariff, a /24 sells for 6,900 rubles per month as address service, but that assumes the address block is available, assigned, billed and supported. A /29 included free with a dedicated access account can be more valuable if it helps retain a connection product with higher gross contribution.

The article's central judgement is therefore not "RELCOM-SPB has IPv4, so it is valuable." The judgement is that RELCOM-SPB has a limited stock of useful IPv4 and a long-lived AS, and that stock only becomes durable cash if it is attached to customers who also buy ports, circuits, racks, voice lines or operational support. If the addresses are used merely to keep old low-yield accounts alive, scarcity accrues to customers rather than to the operator.

Colocation is the second direct contribution test. RELCOM-SPB's 2026 colocation tariff offers a customer's 42U rack, up to 800 by 1220 millimetres and up to 4 kW of power, for 76,500 rubles per month. Per-unit placement in the company's rack is 3,000 rubles per U per month. Port charges are separate: 2,000 rubles per month for 10 Mbit/s Ethernet, 6,000 rubles for 100 Mbit/s and 15,000 rubles for 1 Gbit/s. An additional Ethernet port is 1,300 rubles per month, and the listed port subscription includes unlimited inbound and outbound traffic.

Access to the technical site is free during weekday business hours, but off-hour access costs 1,000 rubles per hour on weekdays and 1,200 rubles per hour on weekends and holidays. Cable channels also have setup and monthly charges. Those prices create a coherent small-colocation product. They also show where margin can disappear. A full rack at 4 kW creates power, cooling, space, switching and support obligations. If the customer takes a low-speed port and generates heavy traffic, the "unlimited" traffic term shifts the risk back to RELCOM-SPB's upstream and internal capacity planning.

The colocation price point should be read against the company's financial scale. One fully billed cabinet at 76,500 rubles per month would produce 918,000 rubles per year before power and support. Ten such cabinets would approach 9.18 million rubles, or most of the reported 2025 revenue. But the public record does not show a facility footprint capable of assuming that many full cabinets, and the company describes "technical sites" rather than a large commercial data-centre estate.

The realistic interpretation is more modest: colocation likely supplements access, routing and address services, and it can be highly accretive when it fills existing power and rack capacity. It is not, on the public evidence, a standalone data-centre growth engine. The required operating question is occupancy. Empty U-space earns nothing, half-used power still carries fixed overhead, and each customer access visit consumes staff time.

Voice is a legacy option with similar economics. RELCOM-SPB's telephone page offers Saint Petersburg city telephone lines with 812 numbers, plus zonal, long-distance and international communication through licensed operators. The tariff effective from December 2025 prices number allocation at 1,000 rubles, monthly number fees at 250 rubles for the 812 578-09 range and 300 rubles for the 812 380-77 range, connection of a telephone line at 700 rubles, and line subscriptions at 1,000 rubles per month for unlimited local outgoing calls or 500 rubles per month with metered local outgoing calls.

Outgoing local calls under the metered plan are priced at 0.60 rubles per minute. That can help a small customer that still wants a local number, but it is not a high-growth service. It is a retention layer. It may make a business account less likely to leave, but it also pulls the company into legacy numbering, interconnection and customer-service work.

There is also evidence of retrenchment. RELCOM-SPB's domain-registration page tells subscribers that the company stopped providing domain-name registration service from November 1, 2025. That decision is not necessarily negative. For a small operator, registrar relationships, customer disputes, billing edge cases and support obligations can consume attention without producing enough margin. But the discontinuation changes the shape of the bundle. The company once looked more like a full local Internet provider: access, mail, domains, telephony, addresses and hosting-adjacent services.

The current public product set leans toward network access, IP resources, colocation and telephone lines. A narrower product set can be disciplined. It can also indicate that the operator is choosing which legacy services are worth maintaining as scale falls.

Customer concentration is the largest unknown. Rusprofile reports that RELCOM-SPB has acted as a supplier in a small number of government procurements, with seven concluded contracts and about 1.2 million rubles of supplier value, including named public-sector customers in Saint Petersburg. That is useful because public institutions can value stable local service and may tolerate a small operator if the relationship is proven. It is not enough to imply a large enterprise base.

IPinfo's hosted-domain count of 1,655 and 2ip's ISP page are market signals that there are domains and users associated with the ASN, but they are not audited customers, not revenue accounts and not churn evidence. The company's own customer pages still include mail and client-settings instructions, which suggests legacy subscriber operations, yet public pages do not disclose active subscriber count, contracted bandwidth, top customers, average revenue per account or churn.

The public procurement signal should therefore be read carefully. A few small contracts can validate that the company can sell to institutions and manage formal procurement documents. They can also create concentration if a microenterprise depends on a handful of institutional renewals. When annual revenue is about 11.66 million rubles, even a 500,000-ruble account matters. One lost office-interconnection customer, one rack customer moving to a larger data centre, or one public institution rebidding to a national carrier can move the income statement.

RELCOM-SPB's retained contribution depends on renewals, not on a vague "regional ISP" label.

The supplier side is equally unforgiving. The company states that dedicated lines can be built using its own resources or other operators' networks. The routing record shows upstream exposure to carriers such as RASCOM, INETCOM and BiMajLink, while the HE view also shows exchange participation and peer visibility. Supplier diversity is a strength if it lets the company route around price or quality problems. It is a weakness if the customer price is too low to absorb upstream changes.

Upstream transit, exchange ports, cross-connects, access leases, replacement routers, switches, optics, power equipment and backup power do not scale down in perfect proportion to revenue. A five- or six-person operator can be efficient, but it also has limited slack when equipment ages or a carrier reprices.

The difference between gross margin and owner cash is the discipline point. Using RBC's public 2025 numbers, revenue of 11.667 million rubles and cost of sales of 7.612 million rubles leave gross profit of 4.055 million rubles. That gross profit has to cover wages, rent or premises costs, support, maintenance, fees, professional services, banking, taxes beyond cost of sales, bad debt, compliance work and replacement capital. Net profit of negative 215,000 rubles means the after-cost system did not clear the hurdle in 2025. It may have been close enough that one renewal, one price increase or one cost cut changes the result.

But close is not durable. A durable regional-network business should be able to fund router refresh, UPS replacement, spares, remote hands, recordkeeping and regulatory response without relying on owners to accept negligible returns.

The strongest economic argument for RELCOM-SPB is that its network rights are hard to recreate. It is a RIPE member and LIR, it holds ORG-RA30-RIPE, it has AS6672, it has valid ROAs, and its route objects date back through long operational history. The 212.113.96.0/19 route object still carries Relcom Corp. description fields and older Saint Petersburg address references. The 37.202.8.0/21 route was created in 2012. The IPv6 route6 object was created in 2009. The organisation object was created in 2004 and modified in 2026.

That continuity can be valuable to customers that care about stable addressing, reverse DNS, mail reputation, routing policy, and a local operator that knows its old network. The weakness is that the same history can become inertia. Old route descriptions, inconsistent addresses and stale whois policy do not by themselves break service, but they create a maintenance debt.

The address-record discrepancy is not trivial. RELCOM-SPB's own requisites page and Russian contractor profiles show the current legal address at Kirochnaya Street 9, letter A, room 3-N. RIPE's member page and organisation object still show the older Shpalernaya address. T-Bank reports a May 13, 2025 legal-address change from Shpalernaya to Kirochnaya. This is a practical control issue. Banks, upstreams, registries, procurement officers and counterparties often check addresses across systems. A small operator cannot afford avoidable friction in diligence, especially if it sells LIR and address-resource services to other customers.

Updating public registry and route-contact records does not create revenue by itself, but it protects sales that depend on institutional trust.

The licence evidence is another control issue. RELCOM-SPB's own licence page lists three communications licences: telematic services, data transmission excluding voice transmission, and local telephone service. It says the first two were extended to September 9, 2026 by a 2021 licensing order, and the local telephone licence, now under the L030-00114-77/00068131 identifier, was extended to December 3, 2028 by a 2023 order. T-Bank's contractor page, however, presents three active communications licences while also logging February 4, 2025 events saying activity under two communications licences was suspended.

That may reflect a reporting nuance, a later reinstatement, a distinction between licence objects and activity types, or a lag in one public source. The economic point is not to accuse the company of a defect. The point is that a customer buying regulated communications services needs clear licence continuity. If the two data/telematic licences are impaired, the service mix and renewal base change materially. If they are not impaired, the public record should be reconciled.

Regulatory exposure is not theoretical. A 2013 Saint Petersburg city court decision involving RELCOM-SPB concerned an order to restrict access to online gambling sites, with discussion of border-router filtering and obligations under Russian information law and telematic-services rules. The case is old, but its relevance is structural. Small Internet operators in Russia are not merely technical resellers; they are enforcement points. They must handle blocking, filtering, lawful requests, reporting and licence obligations. That work does not necessarily come with separate customer revenue.

For a large carrier, compliance cost is spread over a bigger base. For a microbusiness with roughly 11.66 million rubles in revenue, each compliance demand competes with customer support and engineering time.

Geopolitical and equipment risk sits on top of that. Public sources do not disclose RELCOM-SPB's router vendors, switch vendors, optical inventory or maintenance contracts. The operating model nevertheless depends on replaceable equipment, stable power systems and carrier connectivity. Sanctions, import constraints and local supply-chain delays can raise replacement cost or lengthen repair cycles for Russian telecom operators. The company's colocation tariff exposes it to UPS and power infrastructure. Its access and office-link products expose it to field equipment and third-party last-mile reliability.

Its LIR and routing function exposes it to router and security maintenance. None of these risks can be quantified from public data. That is exactly why the reported net loss matters: thin after-cost economics reduce the ability to absorb an unexpected hardware cycle.

Competition is visible in substitution rather than in a single named rival. A Saint Petersburg business needing Internet access, VPN-like office connectivity, rack placement, local numbers or static addressing can turn to national carriers, city fibre operators, hosting companies, data centres, cloud providers, voice-over-IP providers and larger registrars. Some will be cheaper. Some will have better online ordering. Some will have deeper redundancy. RELCOM-SPB's defence is locality, continuity, technical simplicity and address-resource control.

That defence works best for customers who already trust the operator and whose requirements are too specific for a mass-market plan but too small for a national-carrier enterprise contract. It works poorly for price-only customers, web-hosting customers that can move to cloud infrastructure, or voice customers migrating to application-based communication.

Unofficial market signals support the "small but still alive" reading. 2ip lists Relcom-SPb as an Internet provider with AS6672 and repeats the dedicated-channel description. IPinfo reports hosted domains and pingable routers. Hurricane Electric sees peer and exchange data. 2GIS lists the company at Kirochnaya Street 9 as a telecommunications company. A phone-number catalogue attributes about 200 Saint Petersburg numbers in the 812 area to RELCOM-SPB, a tiny share of the numbering space. None of these signals should be treated as audited revenue. Together they suggest that the company is not dormant, but also not operating at scale.

The public Internet leaves traces of a niche provider, not a growth carrier.

The judgement depends on utilisation. If AS6672's routed address space is attached to paying access, colocation and routed-resource accounts, then RELCOM-SPB can remain economically relevant despite its size. A /22 IPv4 block at the posted recurring tariff is 19,435 rubles per month, a /24 is 6,900 rubles per month, and a full customer rack is 76,500 rubles per month before port charges. A few well-priced, low-support customers can move the annual margin.

If, instead, much of the address space is tied to legacy accounts with low pricing, bundled free allocations, unresolved billing or high support load, then the network position is an accounting mirage. The active BGP table proves reachability. It does not prove pricing power.

One way to test the business is to ask what happens after upstream and port costs. Suppose a customer buys a 1 Gbit/s colocation port at 15,000 rubles per month with unlimited traffic. If average usage is low and the traffic mix is helped by peering or local exchange paths, that port can be attractive margin. If the customer pushes heavy outbound traffic over paid upstreams, the port can become underpriced. The same is true for the dedicated-channel service. A line built on owned resources can produce retained contribution; a line using another operator's network must leave enough spread after the access lease.

RELCOM-SPB's public tariffs do not disclose cost per Mbit, port capacity, oversubscription, peering ratios or traffic mix. The rational stance is to value the offering, then discount it for unknown utilisation and input-cost exposure.

Support labour is the second hidden cost. A small operator can deliver better human service than a large carrier because the engineers know the network and customers. But that advantage has a ceiling. Mail support, DNS zones, route objects, AS support, telephone lines, customer site visits, after-hours facility access, billing and compliance all draw on the same narrow team. RBC's six-employee figure and Rusprofile's microenterprise classification make that constraint explicit. A support-heavy customer can consume the margin from several quiet accounts.

A support-light address-resource customer can be profitable even at low nominal billing. The company should therefore be judged by support minutes per ruble, not just by number of services offered.

The third hidden cost is replacement capital. The published tariffs include no depreciation schedule, equipment age, network inventory or power-system condition. Yet a company selling 24/7 access and colocation must replace routers, switches, optics, cables, UPS batteries, PDUs and monitoring systems. It must maintain spares or accept longer outages. A business with negative net profit may still be healthy if owners recently invested or if cash flow differs from accounting profit. But without evidence of investment, the prudent assumption is that replacement capital is a pressure point.

Regional ISP economics often fail when prices cover today's transit bill but not tomorrow's equipment cycle.

The customer-value proposition remains coherent if the company is disciplined. RELCOM-SPB can sell continuity to organisations that want a Saint Petersburg operator, stable IP resources, a local technical contact and a bundle of access, office interconnection, DNS, mail, colocation and telephone options. The product catalogue is old-fashioned, but old-fashioned does not mean useless. Many small institutions value fixed service, not a software portal. The LIR function gives the company a reason to talk to customers with routing needs. The 812 telephone ranges and local support reinforce that municipal business feel.

The key is to price those customers as managed technical relationships, not as commodity broadband lines.

The risk is that the company has allowed its historical position to become a low-yield annuity. Domain registration has stopped. Public revenue is small and down year on year. The income statement tipped negative in 2025. Routing records are active but policy records are not perfectly aligned with observed BGP. Address records lag across RIPE and Russian company registries. Licence public records need reconciliation. Those are not fatal flaws, but they are the exact symptoms of a business where the inherited network still functions while the commercial system has not been refreshed at the same pace.

In telecom, that gap usually narrows either through price discipline and operational cleanup or through gradual erosion.

The facts that would change the judgement are concrete. The positive case would improve if RELCOM-SPB could show high rack occupancy, low power cost relative to colocation revenue, a diversified base of paying access customers, a high share of routes carried over settlement-free or low-cost paths, current licence confirmations, updated RIPE records, low churn, and a clear schedule for equipment replacement.

A disclosed set of multi-year institutional renewals would matter more than a generic claim of "many clients." Evidence that IPv4 blocks are actively billed at current tariffs, rather than bundled into old accounts, would also change the economics. So would proof that the 62.76.254.0/24 whois-only route is intentionally unused rather than stranded.

The negative case would sharpen if public records showed licence impairment, unpaid tax or enforcement issues, loss of key upstreams, reduced BGP visibility, invalid RPKI, declining prefix reachability, large customer exits, persistent public procurement losses, or a further revenue decline without cost reduction. It would also worsen if a large share of customers take free address allocations while buying low-margin access through third-party last mile. In that situation, IPv4 scarcity would not protect RELCOM-SPB; it would subsidise customers and raise operational complexity.

The capital question is sharper than the revenue question because the public tariff book creates obligations before it creates surplus. A company selling a 24/7 dedicated channel has to maintain routing, monitoring, access continuity and customer response even when the customer pays a modest monthly fee. A company selling colocation has to keep power stable, cooling sufficient, ports available and technicians reachable. A company selling telephone service has to preserve numbering, billing and regulated voice interconnection.

A company selling LIR assistance has to maintain accurate registry records and answer customers who are often asking for work that is small in invoice value but high in precision. None of those obligations is optional. They are the base layer beneath the 2025 gross profit. The public accounts show enough gross profit to operate a lean service business, but not enough to make mistakes invisible. A router refresh, UPS battery cycle, failed switch, unplanned carrier increase, legal response or two engineer departures can absorb a large part of annual net capacity. That is why the right metric is not reported revenue, or even gross margin.

It is recurring contribution after the service-specific burdens that keep the promise alive.

The company has one attractive lever that many small service firms lack: a finite inventory of addresses with operating credibility behind it. IPv4 scarcity is not merely a market slogan when the operator can point to valid ROAs, active route objects and an LIR status. Customers that need static address space for servers, mail systems, VPN endpoints, cameras, industrial equipment or branch-office routers may prefer to rent from a local operator that can both assign the addresses and solve routing support problems. But the pricing must reflect scarcity.

A /29 bundled free with a dedicated access product can be a sensible retention tool if the circuit itself is profitable. It is weak economics if the bundle was priced years ago and never repriced after addresses became scarcer. The same applies to /30 colocation bundles. Free addresses are not free to the operator; they carry registry work, routing reputation, abuse handling and opportunity cost. RELCOM-SPB should know which bundled blocks are attached to profitable accounts and which merely lock scarce resources into relationships that cannot carry the support burden.

The 62.76.254.0/24 inconsistency is a useful example of why registry hygiene has an economic value. RIPEstat shows the prefix in whois but not in BGP, while the three other prefixes are both in whois and in BGP. There may be a benign explanation: a retired customer, a reserve block, a route kept in documentation, or a network no longer intended for public announcement. But from the outside, it is an unresolved asset-state question. A routed /24 can be a billable resource, an interconnection input or a customer retention tool. A non-routed /24 can be a reserve asset if deliberately held, or dead weight if forgotten.

For a large carrier, one ambiguous /24 is noise. For RELCOM-SPB, where reported annual revenue is modest, the difference between billable, reserved and stranded address space is material. The company should be able to classify every block by route state, customer state, billable state, abuse state and renewal state. That classification is not back-office neatness; it is the bridge between network position and cash.

Customer renewal strategy also determines whether supplier bargaining helps or hurts. A small operator with several upstream or neighbour options can avoid being wholly dependent on one carrier, but only if its customer contracts allow input changes to be passed through or offset by technical routing choices. If customers are on fixed low prices while transit, exchange, cross-connect or leased access costs move upward, supplier diversity slows the pain but does not remove it. If contracts are renewed annually with clear service definitions, address charges and speed tiers, the operator can protect contribution.

The public tariff pages suggest RELCOM-SPB already understands modular pricing: rack space, U-space, port speed, additional ports, after-hours access, cable channels, address block size and line type are separately priced. The next question is whether old accounts actually pay in that modular way. A tariff page is a price list; an old customer book can be a collection of exceptions. The company's financial result implies that exceptions, utilisation or cost inflation may be large enough to matter.

The unofficial hosted-domain count is worth using as a clue, not as proof. IPinfo's count of domains associated with the ASN suggests that address space is not empty. It may reflect legacy hosting, customer mail, reverse-DNS patterns, parked domains, old websites or third-party hosted names. It does not reveal whether the accounts are profitable or even directly billed by RELCOM-SPB. A hosted-domain count can remain high long after revenue has migrated elsewhere, especially in a legacy regional network where old DNS and mail arrangements persist. That is why the customer pages for mail and administrative settings matter.

They show a support surface that can persist even when the commercial value per account is low. For management, the issue is not whether customers exist. It is whether the remaining customers pay for the operational work they require. Legacy mail support, DNS troubleshooting and address reputation work can be sticky, but they can also become a silent subsidy if sold as part of an underpriced access bundle.

Competition should be understood at the task level. For a raw Internet port, the substitute is a larger carrier with more backbone capacity and a lower unit cost. For basic web hosting, the substitute is a hosting provider or cloud platform. For domain registration, RELCOM-SPB has already exited the service, implicitly accepting that larger registrars can do the job more efficiently. For local numbers, the substitute is a voice provider or hosted PBX. For an office interconnection, the substitute is a carrier VPN, SD-WAN overlay or managed service from a larger integrator.

RELCOM-SPB wins only where the customer values the combination: local access, static addressing, a reachable technical contact, a small colocation option, and continuity from a provider that has handled the account for years. This is a defensible niche, but it cannot be defended with commodity pricing. If the company prices as though it were selling mass-market bandwidth, it inherits the cost structure of a telecom operator without the scale benefits of one.

The regulatory burden is likewise part of customer economics. The 2013 court record is old, but it shows the company being treated as an operator capable of filtering at the border router. That is the role a state assigns to network operators, not a one-off support ticket. Russian communications providers have to operate inside a changing environment of blocking, licensing, reporting and information-control obligations. The work is not always visible in financial summaries. It appears as staff time, legal review, router configuration, correspondence, logging and risk management.

A customer paying for access does not necessarily pay a separate line item for that burden. For RELCOM-SPB, the right response is to include compliance workload in the contribution model. Accounts that generate abuse handling, legal notices or special filtering obligations have a higher cost than quiet accounts at the same bandwidth. In a small company, regulatory work competes directly with revenue work.

The judgement can therefore change quickly on evidence that looks mundane. A clean updated RIPE organisation address would lower diligence friction. A current regulator extract resolving the licence contradiction would remove a material uncertainty. A route-policy update matching observed upstreams and exchange peers would improve wholesale credibility. A published maintenance window or network-status practice would signal operational discipline to colocation and routed-access customers. None of these requires a large capital programme. They are control improvements.

By contrast, buying new capacity, adding services or chasing low-margin customers before that cleanup would increase complexity without fixing the core contribution question. For RELCOM-SPB, the best growth is probably renewal quality: fewer exceptions, clearer address charges, cleaner registry records, better customer-level contribution data and a sharper refusal to carry unprofitable legacy arrangements.

The immediate managerial answer is not expansion for its own sake. RELCOM-SPB should first make the existing footprint earn. That means reconciling licence and address records, tightening whois policy to match real routing, pricing IPv4 scarcity explicitly, measuring contribution by customer after access, transit, port, power and support costs, and refusing to carry accounts that are loyal only because old bundled terms are underpriced. It means treating colocation as a power-and-support product, not just rack space. It means checking whether each 1 Gbit/s "unlimited" port has traffic behaviour that matches the tariff.

It means turning AS support and LIR services into professional-service retainers where customers pay for correct registry work, not casual favours.

That stance is not glamorous, but it is the practical route for a micro-operator. RELCOM-SPB does not need to become a national network to be viable. It needs enough retained contribution from a finite group of customers to fund reliability, compliance and replacement capital. The public evidence says the operating footprint is real, the legacy position has value, and the product catalogue still maps to actual small-business and institutional needs. The same evidence says the margin of safety is thin. After upstream capacity, ports, support, power and equipment renewal, the durable cash yield is unproven.

My judgement is therefore conditional and hard-edged: RELCOM-SPB has a defensible network position only if management converts routing history and IPv4 scarcity into priced, renewed, low-support customer relationships. The company should not be valued as a scaled regional carrier, and it should not be dismissed as dormant infrastructure. It is a small Saint Petersburg network with scarce resources, active BGP, old customer-service DNA and 2025 financials that show how little room remains when legacy telecom services are not repriced.

The control question is whether the next renewal cycle is used to raise contribution or merely to preserve revenue. If it is the former, the footprint can keep producing cash. If it is the latter, upstream, power, support and replacement capital will keep eating the network position from the inside.

Sources