Summary
- Adant sells colocation at Moscow's M9 carrier hotel from RUB 4,600 a month for a unit and RUB 135,000 for a 5 kW rack. The paid product is not floor space alone. It is the transfer of power, cooling, physical-security and repair risk from a customer server room to a specialist located where many carriers can be reached.
- Registry-derived accounts report 2024 revenue of RUB 76.2 million, gross profit of RUB 29.8 million and net profit of RUB 18.9 million. Those are healthy accounting margins for a six-person company, but they do not reveal cash conversion, site occupancy, customer concentration, asset ownership or the return on capital employed.
- AS199278 provides real operating evidence: one RPKI-valid IPv4 /22 is publicly originated. It also shows concentration. Current routing views identify RECONN as the only observed upstream, while no IPv6 route is originated and the older routing-policy record names several connections that are not visible as current transit.
- Adant can create value if its small team delivers faster human intervention and easier M9 cross-connects than larger competitors. Revenue growth alone is insufficient. The decisive tests are renewal, outage performance, current telecom authority, facility tenure, supplier redundancy and whether customers can buy equivalent resilience directly or from a larger multi-site provider.
The economic incentive: buy avoided downtime, not a rack
The smallest useful unit in Adant's offer is not one metal slot. It is an insurance-like bundle. A business places its own server in a controlled room and pays somebody else to keep electricity, cooling, access control and connectivity working while its office is closed. The customer benefits if the expected loss from an outage, theft, heat event or delayed repair is greater than the monthly fee. Adant benefits if the pooled cost of protecting many customers is lower than the sum of the rents it collects.
That bargain changes the allocation of risk, but it does not eliminate risk. The customer still owns the equipment, applications and data. It pays the capital cost of the hardware, arranges backups and absorbs business interruption if the server or its provider fails. Physical migration is costly because it requires a maintenance window, transport, recabling, route changes and testing. Adant therefore gains a retention advantage after installation. The same friction that protects recurring revenue can trap a dissatisfied customer, so contract terms, response times and exit assistance matter as much as the headline price.
Adant's current site says it places equipment from 1U, performs cross-connects, provides physical security, maintains climate, offers access to many carriers and supports customers when equipment or lines need attention. It advertises more than 15 years of colocation activity. These claims describe a coherent product: local intervention at a connectivity-rich building. They are not measurements of uptime or repair speed. No public service-level schedule, incident history, credit formula or remote-hands price was found on the reviewed page, and the footer says the information is not a public offer. A buyer therefore cannot infer the enforceable service from the marketing page.
The economic burden falls differently across customer types. A small software company may pay Adant because one server in an office cupboard is exposed to a single electrical circuit and consumer cooling. A regional carrier may pay for rack space and cross-connects because entering M9 directly at larger scale is administratively expensive. A public institution may value local control and physical access. A large cloud operator, by contrast, can negotiate wholesale capacity, employ its own engineers and spread workloads across sites. Adant's strongest customer is large enough to suffer from downtime but too small to reproduce carrier-hotel economics independently.
This creates a simple value test. Adant must solve the customer's continuity problem more cheaply than an office server room, more personally than a scaled provider and with less complexity than a self-managed M9 presence. Low rent can win an order; reliable repair and transparent escalation are what make the order economically durable.
Identity, ownership and the three-address question
The operating identity can be tied together with reasonable confidence. Russian registry-derived records identify the company as LLC Adant, registered on 21 July 2009 under OGRN 1097746396220 and tax number 7728704654. The RIPE organisation record for ORG-LA555-RIPE includes the same registration number, associates the organisation with the adant.ru network identity and names AS199278. That exact registration-number match is stronger than a shared name, which matters because many unrelated Russian companies use the name Adant.
RBC Companies reports that the legal entity remains active, is led by general director Alexander Kostikov and is owned 60 per cent by Sofia Skorenkova and 40 per cent by Nina Kostikova. Its registered capital is RUB 25,000. The small registered capital says little about operating resources, but the concentrated ownership matters. It allows decisions to be made quickly, while giving customers and creditors less public governance visibility than they would receive from a listed or institutionally owned provider.
The records expose three addresses that should be understood rather than casually merged. The current legal address is at 32 Nakhimovsky Prospekt. The RIPE record and older data-centre directories point to 20, building 2, Nauchniy Proezd. The current commercial page sells placement at MMTS-9 and lists 7 Butlerova Street, the M9 address. A change of legal address, a historic operating site and a current service site can all coexist legitimately. They nevertheless raise a commercially important question: does Adant own engineering capacity, lease wholesale racks, act as a retail reseller at M9, or combine those models?
An older AllDC listing described an Adant facility at Nauchniy Proezd launched in 2009, with 40 racks, 0.6 MW total electrical capacity, 0.3 MW guaranteed capacity, N+1 arrangements and two fibre cables to M9. That record is useful history, not proof of the present configuration. Adant's current page is explicitly about racks and units at MMTS-9. The financial distinction is large. An owner-operated facility earns the spread after power, maintenance and depreciation but carries plant-replacement risk. A reseller can operate with less fixed capital but must share the economics with the building or wholesale provider and remains exposed to lease renewal and wholesale price changes.
The public ownership records also differ from the named RIPE technical contact. Historical RIPE data names Alexandr Aronov as the administrative and technical contact, while the company register names Kostikov as general director and two other individuals as owners. A network contact is not evidence of equity control, so there is no contradiction to resolve by assumption. The useful conclusion is narrower: commercial authority, technical authority and ownership appear to sit with different people. A serious customer would ask for an escalation map that shows who can approve network changes, who can authorize physical access and who is accountable under the service contract.
What Adant sells and where the margin can come from
Adant's present public offer is narrower and more intelligible than a generic technology-company label. It advertises individual-unit and full-rack colocation at M9. The unit offer starts at RUB 4,600 a month and includes a 10 Mbps connection described for remote equipment control, one IPv4 address and the possibility of IPv6. The rack offer starts at RUB 135,000 a month with 5 kW of allocated power; more power and more connections are available by quotation. The site also emphasizes cross-connects and direct access to multiple carriers.
There are four possible revenue layers. The first is space and power. The second is connectivity, including internet access, cross-connect administration and address use. The third is labour: installation, cabling, rebooting, diagnosis and other remote-hands work. The fourth is trust, expressed economically through longer contracts and lower churn. Public information does not break revenue into these layers, so any valuation based on one line alone would be false precision.
The offer's appeal comes partly from aggregation. M9 is valuable because many networks meet there. When MSK-IX opened its M9.PLUS facility in 2017, it described the building as Russia's most popular interconnection location and said more than 440 providers were present. MSK-IX's current corporate page reports a much broader network of more than 800 organisations across 45 sites and 10 cities, with traffic of 8.78 Tbps. Those figures belong to MSK-IX as a whole, not to Adant, but they explain why space at Butlerova Street can command a connectivity premium.
Location reduces the physical cost of reaching another carrier. It does not guarantee that Adant's retail price includes the desired carrier, port speed or cross-connect. MSK-IX itself says colocation arrangements generally combine rack space, support, connection options, cross-connects and out-of-band access, with details varying by site. A customer should therefore divide Adant's quote into building rent, metered or committed power, network port, traffic, cross-connect, address, support and one-time installation. A cheap bundle can become expensive if the parts most likely to grow are separately priced.
Adant's model can be attractive because many costs are shared. Security, cooling, staff presence and carrier access do not rise one-for-one with every additional unit sold. Once the provider has committed to a rack and an engineering shift, filling the next empty unit can have a high incremental contribution. The reverse is also true. An empty rack still incurs wholesale rent or capital charges, and a lightly loaded technical team still has to be available. Occupancy is therefore the silent variable behind reported profit.
Network-resource evidence: real operations, limited scale
The network evidence establishes that Adant is more than a trading name. BGP.Tools shows AS199278 as active and registered to the RIPE member ru.adant. It originates 185.17.76.0/22, equivalent to 1,024 IPv4 addresses, and the route is covered by a valid RPKI authorization. The ASN was registered in April 2013. These facts demonstrate control of a modest public routing surface and responsible use of a modern route-origin control.
They do not demonstrate a broad access network. The same current view shows one originated IPv4 prefix, no originated IPv6 prefix, one upstream and no visible downstream networks. IPinfo classifies the ASN as hosting, reports 84 hosted domains across 16 addresses and also identifies RECONN, AS12722, as the sole upstream. Hosted-domain counts are discovery estimates, not customer counts; customers can run several domains on one address, and colocation clients may bring their own address space. Still, the pattern is more consistent with a small hosting or colocation network than a large regional last-mile provider.
The distinction between allocated and routed resources matters. RIPE allocation statistics associate Adant with an IPv6 /32, while current global routing views show zero IPv6 routes originated by AS199278. Adant's tariff page says IPv6 can be provided, but the observed ASN does not prove that this is delivered from Adant's own announced IPv6 space. It could be supplied through another network, held but not routed, or activated only for specific customers. A buyer that needs dual-stack service should test it and record the actual origin, path and failover behaviour.
The public RIPE routing-policy object lists import relationships with RASCOM, TransTeleCom, COMCOR, AS203694 and Foton Telecom, and export statements to several of them. That record was last materially changed in 2020. Observed routing in 2026 shows only RECONN as upstream. Policy declarations describe intended or historical relationships; they are not invoices, live circuits or proof of simultaneous multihoming. The conservative interpretation is that Adant has experience with several carriers but currently exposes a concentrated public path for its own /22.
RECONN itself is not a tiny endpoint. Its current network view shows multiple upstreams, many peers and dozens of downstreams. Adant can therefore inherit path diversity behind RECONN. That is not the same as independent first-hop redundancy. If the Adant-RECONN commercial or physical connection fails, diversity deeper inside RECONN does not repair the first link. The customer needs to know whether separate ports, fibres, routers and power domains exist, and whether a second carrier can announce the service during failure.
Address resources also carry cost and responsibility. RIPE's 2026 billing procedure sets an annual LIR contribution of EUR 1,800 and an additional EUR 50 per ASN assignment. These sums are small beside Adant's reported revenue, but the obligation is denominated in euros and Russian members face specific payment procedures. More important than the fee is operational stewardship: accurate contacts, abuse response, route objects, RPKI and customer allocation records. A scarce IPv4 pool can help sell colocation, but it can also attract abuse complaints and create reputational cost if assigned carelessly.
Pricing and the unit-economics test
The published prices allow useful bounds even though they do not support a complete profit model. One unit at RUB 4,600 produces RUB 55,200 of annual recurring revenue before discounts, tax treatment and extras. One rack at RUB 135,000 produces RUB 1.62 million a year. A customer filling many units should expect a discount: the full-rack price equals the retail price of about 29 individual units, while a typical cabinet can hold substantially more. That is rational because one customer, one contract and one locked rack can require less administration than dozens of small tenants.
Power is the central constraint. A 5 kW allocation consumed continuously would equal 43,800 kWh a year before cooling and electrical losses. Dividing annual rack revenue by that maximum IT load gives about RUB 37 of revenue per IT kWh. That is not an electricity tariff or a margin. It is the total revenue envelope from which Adant must pay facility rent, actual power, cooling overhead, network access, maintenance, labour and tax. If customers draw less than their allocation, the economics improve; if wholesale power is repriced or high-density loads require additional cooling, they worsen.
The current Russian market gives Adant some pricing room. Kommersant, citing iKS-Consulting and 3data, reported an average Moscow-region colocation price of RUB 160,300 a month at the end of 2025, excluding individual discounts and VAT, after an 11 per cent annual increase. Adant's RUB 135,000 starting rack price is about 16 per cent below that average. The comparison is directional, not exact: rack power, certification, carrier fees, tax and support differ. It nevertheless suggests that Adant is not attempting to charge a scale-provider premium.
At the small end, the price is ordinary rather than disruptive. DataHouse lists RUB 5,400 a month for 1U with a power supply up to 550 W and an unlimited 100 Mbps port, rising to RUB 6,100 for 950 W. Friend IT advertises 1U from RUB 4,000 with up to 200 W and 100 Mbps, and a 42U rack with up to 7 kW and 1 Gbps from RUB 125,000. Adant's RUB 4,600 unit sits between them, but its listed 10 Mbps link is described as remote-control access and may not be comparable with a production traffic port. The correct comparison requires a written configuration.
The company's 100-client claim provides another rough check. If 2024 revenue of RUB 76.2 million and the current client count referred to the same period and scope, revenue would average about RUB 762,000 per client a year, or RUB 63,500 a month. That is almost fourteen entry units, or roughly half the advertised rack rate. The arithmetic implies that the average relationship is larger than a single entry unit, that revenue includes other services, that client counts changed, or some combination of the three. It does not reveal the actual distribution.
The same exercise on cost is instructive. Reported 2024 cost of sales of RUB 46.4 million equals about RUB 3.87 million a month. Spread across 100 claimed clients, that would be RUB 38,700 per client per month, leaving roughly RUB 24,800 of gross profit before overhead and tax. Averages hide the tail: a few rack-heavy customers may account for much of both revenue and cost. Without contract-level gross margin, a healthy average can coexist with unprofitable small accounts or dangerous dependence on one large buyer.
Revenue growth is not yet proof of value creation
The 2024 accounts are encouraging. RBC Companies, using published accounting data, reports revenue of RUB 76.221 million, cost of sales of RUB 46.443 million, gross profit of RUB 29.778 million and net profit of RUB 18.905 million. That implies a gross margin of about 39.1 per cent and a net margin of about 24.8 per cent. Revenue rose 24 per cent from RUB 61.481 million in 2023. For a small infrastructure service, those margins indicate that customers paid materially more than the recorded direct cost of serving them.
But revenue growth can come from four sources with very different value. More occupied units represent volume. Higher prices represent market power or inflation recovery. More electricity and carrier charges passed through represent accounting expansion with little incremental profit. A richer service mix represents genuine product improvement if customers renew. Public accounts do not separate these effects.
Nor is one year's rebound a trend. A separate registry-derived company page reports 2021 income of RUB 68.559 million and profit of RUB 30.986 million. Compared with 2024, nominal revenue was only about 11 per cent higher while reported profit was about 39 per cent lower. Accounting labels and periods should be checked against the original statements before drawing a hard time series, but the direction warns against equating the 2024 revenue increase with compounding owner value.
Value creation requires a return on the capital actually committed. Public summaries do not disclose enough balance-sheet or cash-flow detail to calculate it. If Adant owns generators, UPS equipment, cooling and racks, depreciation and replacement capital are central. If it leases space at M9, lease liabilities, deposits and wholesale indexation are central. If customers prepay, cash conversion may be excellent; if large customers pay late while suppliers demand advance payment, reported profit can conceal working-capital strain.
The six-person headcount reported in the company records makes the operation look extremely lean. Revenue per employee was roughly RUB 12.7 million in 2024 and net profit per reported employee roughly RUB 3.15 million. That can indicate effective automation and the use of specialist facility partners. It can also mean contractors perform important work, or that a small number of employees carry key-person and shift-coverage risk. The figure cannot tell us whether an engineer is physically available at 03:00, which is the service moment that determines whether accounting profit reflects durable value.
The judgment therefore remains conditional. Adant looks profitable, not necessarily highly valuable. To cross that gap, the company would need to show sustained renewal, stable cash generation after maintenance spending, manageable customer concentration and an operating record that converts low apparent scale into trusted service rather than fragility.
Costs, capital and the reliability burden
Colocation costs divide into those that run continuously and those that arrive suddenly. Continuous costs include wholesale space or building depreciation, electricity, cooling, security, connectivity, facility maintenance, insurance, software, taxes and labour. Sudden costs include a failed UPS component, router replacement, generator overhaul, fibre repair, sanctions-driven vendor substitution or emergency transport. Reliability is created by paying for both categories before a failure occurs.
Power and space are likely the largest direct inputs. The Russian market backdrop is adverse for buyers of both. Kommersant reported that day-ahead electricity prices in the European Russia and Urals zone rose 17.5 per cent year on year through late December 2025. It also reported limited available data-centre capacity and hard price positions from operators. A small reseller can pass increases to customers only when contracts permit; otherwise the gross margin absorbs them until renewal.
Capital is expensive. The Bank of Russia recorded a 14.25 per cent policy rate on 9 July 2026. Even a profitable company should hesitate before financing new electrical plant at commercial rates materially above that benchmark. Leasing M9 capacity can avoid a large construction bill, but it exchanges financing risk for counterparty and renewal risk. That trade is sensible only if the wholesale term is longer or more predictable than the customer commitments sold against it.
Hardware replacement has a foreign-exchange and availability component. Network operators depend on routers, optics, UPS modules, batteries, cooling parts and management software. Some may be domestic or sourced through alternative supply chains, but many global products contain foreign technology. European sanctions restrict exports to Russia of categories that include advanced semiconductors, electronic components, software, laptops and hard drives. US rules restrict certain enterprise software support and related services while preserving authorizations for internet communications. These measures do not show that Adant is sanctioned, and internet access is not generally prohibited. They do raise lead times, compliance work and the risk that a familiar device cannot be replaced on the old terms.
Labour has its own asymmetry. Customers buy availability around the clock, while Adant's sales page and older listings distinguish business-hour sales from continuous technical coverage. A six-person legal headcount cannot staff every specialty on every shift without partners, overtime or on-call arrangements. That is not evidence of inadequate coverage. It means the staffing model is financially material and should be documented. Cheap remote hands cease to be cheap when only one person knows the cabinet, the route or the customer's access procedure.
Finally, number-resource administration has a small cash cost but a large tail risk. A bad allocation can generate abuse traffic, blocklisting and customer support work. An inaccurate contact can slow incident handling. A route leak can damage many tenants at once. The RPKI-valid /22 is positive evidence that Adant has taken one important control seriously. Equivalent evidence for route filtering, change approval, backups and incident review is not public.
Suppliers and upstream leverage
Adant's supplier chain begins with the physical site. The current website says service is delivered at MMTS-9, but it does not name the contractual landlord, wholesale colocation provider or length of tenure. MSK-IX offers colocation at M9.PLUS directly, and other carriers and resellers operate in the same building. This gives Adant choice in principle while also giving its customers alternatives.
The second supplier layer is power and cooling. Adant advertises dual power redundancy, five independent inputs and climate control. Those are company statements about the offer, not an independently published topology. The customer should distinguish building feeds from the two circuits that actually reach its rack, verify whether both pass through independent UPS paths and ask how generators are tested under load. Five upstream utility inputs do not help if a common distribution board, lease boundary or human procedure remains a single point of failure.
The third layer is connectivity. The building's carrier density is strategically valuable, yet Adant's own public route currently reaches the internet through RECONN. This may be an efficient wholesale choice. RECONN's larger network can buy transit and peering at better rates than Adant could on its own, and the public testimonial on Adant's page says the two companies have worked together since 2016. But RECONN appears both as a named partner and as the observed upstream. That makes service quality and bargaining power unusually concentrated in one relationship.
Older RIPE policy text names several carriers, and Adant says direct access to many Russian and international operators is available. A customer may therefore be able to buy an independent circuit in the same building even if Adant's bundled internet remains single-homed. That is one of M9's advantages. The practical question is who signs the carrier contract and who controls the cross-connect. If the customer contracts directly, it gains route independence but takes on more administration. If Adant bundles the circuit, the customer gets convenience but inherits Adant's supplier selection and margin.
The fourth layer is equipment supply and maintenance. A larger operator may stock more spares, negotiate vendor support and move workloads between owned sites. Adant may compensate with shorter decision chains and familiarity with each rack. The economically relevant measure is mean time to restore, not the logo on the chassis. No public spare policy, maintenance schedule or restoration statistics were found.
Supplier resilience should be evaluated end to end. Two carriers sharing one duct are not independent. Two power feeds sharing one switchboard are not independent. Two engineers depending on one access credential are not independent. Adant's location makes redundancy purchasable; the commercial contract determines whether the customer actually bought it.
Customers, concentration and market dependence
Adant says it serves 100 clients and works across industries, including government bodies. It publishes testimonials attributed to the general director of Kvant-Telecom, a department head at the Institute for High Energy Physics, a client director at Mastertel and an executive director at RECONN. The Institute reference says the relationship dates to 2011, the RECONN reference to 2016, and the Kvant-Telecom reference describes more than ten years of work. Long relationships would be valuable because they imply recurring demand and operational trust.
The evidence is controlled by Adant, so it should be treated as named reference material rather than independent satisfaction data. Even so, the mix is revealing. Carriers value cross-connects and equipment access. A scientific institution values continuous access and bandwidth. Government-related users may value domestic location and familiar support. These are customers for whom moving costs are real and a local relationship can matter more than a sophisticated self-service portal.
The 100-client figure also implies concentration risk can be meaningful despite appearing diversified. If all clients were equal, each would represent one per cent of revenue. They will not be equal. A full rack produces almost thirty times the starting revenue of one unit before extras. Ten rack-heavy accounts could therefore dominate a nominally broad customer base. The public accounts provide no top-five share, churn, arrears or contract duration.
Market dependence cuts both ways. Moscow colocation capacity has been tight, which supports price and occupancy. The same shortage pushes customers to reconsider architecture, regional sites and cloud alternatives. Kommersant reported that the five largest Russian data-centre operators held 74 per cent of rack capacity in 2025. Adant is operating beside providers with more capital, multiple sites, certifications and integrated cloud services. It is unlikely to win a scale contest. Its advantage has to be responsiveness, flexible small orders and access to a specific interconnection location.
Demand quality matters as much as demand quantity. A customer placing legacy network gear may remain for years but consume little incremental service. A growing digital business may buy more power and connectivity but demand stricter assurance. High-density artificial-intelligence equipment increasingly seeks 10 kW or more per rack; Adant's public base rack is 5 kW, although more power is said to be available. A 5 kW product can be well matched to telecom, enterprise and storage equipment without participating fully in the highest-growth compute segment.
The best customer portfolio would combine long-lived low-churn tenants with enough expansion to cover input inflation, while avoiding dependence on one carrier, public body or high-power buyer. Public data does not allow that portfolio to be reconstructed. Revenue growth shows that someone paid more in 2024. It does not show who, for what, or under what renewal risk.
Competition and realistic substitutes
The first alternative is an office server room. It avoids colocation rent and gives staff immediate physical access. For one non-critical server with good building power and modest heat, it can be rational. It becomes false economy when the office depends on one electrical path, lacks fire suppression, runs consumer cooling or has no qualified person available during an outage. Adant wins when the customer's expected downtime and staff burden exceed RUB 4,600 a month plus connectivity and migration costs.
The second alternative is another small Moscow colocation provider. Friend IT advertises a lower starting unit price and a lower rack price with higher listed bandwidth and power. DataHouse charges somewhat more for a unit but publishes clearer power tiers, a 100 Mbps port and an SLA claim. ITSoft offers low base unit prices plus separately priced power, addresses and connectivity, which makes the cost more configurable. These offers show that Adant cannot rely on price alone. It must prove that M9 location, cross-connect convenience and human response outweigh package differences.
The third alternative is a scaled provider such as Selectel, DataHouse, IXcellerate or the Rostelecom data-centre group. Scale can provide multi-site architecture, audited processes, deeper staffing, cloud adjacency and stronger purchasing power. Selectel's rack documentation describes six sites, 24/7 support, 5 kVA included rack power, independent feeds and optional connections to other operators. A small customer may receive less personal attention and pay for capabilities it does not need, but a regulated or rapidly growing customer may prefer the documented expansion path.
The fourth alternative is direct M9 participation. MSK-IX's M9.PLUS offer combines colocation with exchange and private-network services in the same building. A carrier with its own network team can contract closer to the source and remove a retail layer. Adant remains useful if it aggregates small requirements, handles paperwork, performs hands-on work or supplies a simpler bundled bill. Its margin is justified only by services the direct option does not provide as cheaply.
The fifth alternative is a dedicated server, virtual private server or domestic public cloud. These remove the customer's hardware purchase and make capacity easier to change. They can be superior for variable workloads, standard applications and teams without hardware expertise. Colocation remains stronger when the customer already owns equipment, needs unusual hardware, wants predictable heavy utilization or requires direct carrier cross-connects. The comparison must include hardware depreciation, software licences, backup, egress, administration and migration, not just monthly compute rent.
The sixth alternative is geographic diversification. A customer can place primary and backup systems with different Moscow providers, or move the secondary site to St Petersburg or another region. This costs more than one Adant rack but actually reduces building-level and city-level risk. CNews reported that customers were considering regional centres and hybrid deployments as Moscow capacity tightened. For a critical service, Adant should be assessed as one component of resilience, not as a substitute for a second location.
These alternatives clarify the company's economic position. Adant is not selling the cheapest possible compute or the largest possible facility. It is selling a small, reachable operating layer at a valuable interconnection point. That can be a defensible niche, but only if response quality is real and the customer understands which risks remain outside the bundle.
Regulation, geopolitics and operating risk
Adant operates where telecom, hosting, personal-data and sanctions rules overlap. The exact obligations depend on what it supplies. Pure space and power are different from internet access, managed hosting or customer data processing. The company's registered primary activity is wired telecommunications, its current page includes an IP address and network access, and older directories describe hosting. A buyer should map each contracted line to the legal entity and authority that supplies it.
One public record deserves specific follow-up. A company-data service drawing from Russian registers shows three entries on 5-6 December 2025 in which a licensing authority reported licence annulments. The accessible record does not identify the licence numbers, services, reasons or whether replacement authority exists. Annulment can follow expiry, surrender, restructuring or enforcement; it is not by itself proof that the current colocation offer is unlawful. It is, however, too material to ignore. Current telecom licences, partner licences and the contractual basis for any bundled connectivity should be obtained before a long commitment.
Russian law also imposes duties on certain hosting activities. The government's implementation of the hosting-provider register and the current reviewed text of Federal Law 149-FZ require covered providers to notify the authority, enter the register and identify or authenticate customers before providing connected computing capacity. Colocation where the customer owns and controls the server may not map identically to managed hosting. The service description and legal interpretation matter. The absence of an easily found Adant entry during this review is a question, not a finding of non-compliance.
Personal-data rules can create demand for domestic infrastructure. Federal Law 152-FZ governs automated processing and security of personal data, while the 2014 amendments addressed processing in information networks and domestic database use. Local colocation can help a customer choose where systems sit. It does not by itself make the customer's applications compliant, secure or properly documented. Adant's own privacy page identifies LLC Adant as the website data operator and describes measures for website inquiries; that policy is not a certification of customer workloads.
Sanctions create both direct and indirect risk. EU restrictions cover broad technology and electronics categories. US rules introduced restrictions on certain IT consultancy, design, support and cloud services for enterprise and design software, while expressly preserving many transactions related to telecommunications and internet communications. The relevant outcome depends on the product, vendor, end user and payment path. No reviewed source identified Adant itself as a sanctioned party. The prudent concern is supply continuity, bank screening and support access, not an unsupported allegation.
RIPE NCC has separately explained that internet number resources received an exemption from specific EU sanctions concerning Russia, while its regular sanctions process still applies in other cases. Its 2026 billing procedure specifically addresses Russian invoicing and requires euro payment. That preserves a route for registry continuity but leaves foreign-exchange and banking execution as operating tasks.
Domestic network policy is another uncertainty. Operators can be required to restrict access to prohibited content, retain specified communications data or cooperate with lawful requests depending on their service role. These obligations can raise equipment, storage and compliance costs. They can also transfer risk to customers whose traffic or counterparties create scrutiny. A well-written contract should allocate notice, suspension, data-handling and exit rights instead of relying on general assurances.
Unofficial market signals: useful mainly for what they do not prove
Independent public commentary about Adant is unusually thin. Business directories repeat old Nauchniy Proezd contact details and broad labels such as hosting, network installation or equipment sales. One directory shows no customer review; another invites the first review. A current 2GIS listing exists at the Nakhimovsky address, but the publicly indexed result provides little operating detail. This sparse trail does not prove either excellent service or hidden trouble. Business-to-business colocation customers often complain privately, through account teams or in closed technical groups rather than on consumer sites.
The most favourable qualitative material is on Adant's own page. Named testimonials praise responsiveness and long relationships. Because Adant selected and published them, they are commercial references, not a representative survey. They are still useful leads: a prospective customer can ask for permission to contact the named organisations and verify scope, duration, incident handling and whether the person quoted still stands behind the statement.
Industry participation offers a modest legitimacy signal. The Russian Data Center Industry Association's public membership list included LLC Adant in its October 2025 register alongside larger operators, suppliers and engineering companies. Membership shows engagement with the sector. It does not certify uptime, solvency or technical standards.
There is also evidence of senior participation in industry events and a social-media post by a person presenting herself as Adant's finance director. Such material can indicate an operating business with visible personnel, but it is weak evidence for performance. Social profiles can be outdated, self-described or detached from the contracted service. No rumour or isolated comment should override routing data, signed accounts, current licences, a facility inspection or customer references.
The absence of a public status page is more consequential than the absence of online praise. Customers cannot independently inspect incident frequency, maintenance windows or restoration times. A provider can have an excellent record without publishing one, but opacity forces the buyer to demand private evidence. In this case unofficial signals should raise the required level of due diligence, not drive a positive or negative verdict.
Bull case, base case and downside case
The bull case is a disciplined, capital-light niche. Adant buys or controls well-located M9 capacity, aggregates it for customers too small to contract efficiently, and adds high-value local support. Tight Moscow capacity supports occupancy and repricing. Long relationships lower churn. The six-person team keeps overhead low, while RECONN and facility partners provide scale behind the scenes. Under this case, 2024's 39 per cent gross margin and 25 per cent net margin are evidence of a good service franchise rather than temporary accounting strength.
The base case is a profitable but opaque reseller. Adant earns a reasonable spread on space, power and connectivity, with some sticky customers and some legacy revenue. Its low public price helps fill units, but wholesale increases and customer bargaining prevent profit from keeping pace with revenue. The single observed upstream is acceptable for non-critical bundled access because customers needing more can buy another carrier at M9. Value exists, but it belongs to a closely held microbusiness with meaningful key-person and contract-renewal risk.
The downside case is a mismatch between promises and control. Adant may not own the physical capacity it markets, may face repricing or non-renewal at M9, and may depend operationally on one upstream and a very small staff. Customers may discover that power, traffic, support or cross-connects cost more than the headline. Licence changes could complicate bundled telecom services. A major failure could force customer migrations at the worst time, while Adant's low registered capital and limited public asset visibility provide little comfort.
None of these cases can be selected confidently from revenue alone. The bull case requires evidence that the provider controls its inputs for at least as long as its customer promises, restores incidents quickly and retains customers without excessive discounting. The downside case becomes less likely if Adant produces current licences, a long facility agreement, independent route diversity, audited uptime and a customer-concentration schedule.
Facts that would change the judgment
The first decisive fact is the physical contract. A document showing Adant's ownership or long-term right to specific M9 racks, power and cross-connect capacity would establish whether revenue rests on durable control. It should identify renewal, indexation, termination, access and disaster provisions. The present website does not answer those questions.
The second is a current service schedule. It should state what the RUB 4,600 unit and RUB 135,000 rack actually include: usable power, A and B feeds, traffic port, committed bandwidth, cross-connects, IPv4 and IPv6, support minutes, access hours, installation, overage and tax. It should also contain measurable uptime and response targets with credits large enough to influence behaviour.
The third is facility performance. Twelve to thirty-six months of power, network and support incidents, including maintenance and customer-affecting near misses, would show whether reliability is engineered or merely described. Independent certification would help, but observed performance and restoration evidence matter more than a label.
The fourth is supplier redundancy. Live route collectors should show two genuinely independent first-hop upstreams if multihoming is sold. Fibre maps, router diagrams and power-path tests should demonstrate that common points are understood. If RECONN remains the only bundled upstream, the contract should say so plainly and make a second-carrier option practical.
The fifth is regulatory authority. Current telecom licence extracts, an explanation of the December 2025 annulment entries, any partner-carrier agreements and the company's status under the hosting-provider regime would remove a material uncertainty. The question is not whether paperwork exists somewhere; it is which entity is legally responsible for each paid service.
The sixth is customer economics. Revenue by product, occupancy, top-ten concentration, annual churn, renewal price, bad debt and average contract term would reveal whether growth reflects durable demand. A provider with 100 customers and no client above ten per cent is very different from one where two racks fund the business.
The seventh is cash and capital. Original 2023-2025 financial statements, cash flow, balance sheet, lease commitments, maintenance spending and owner distributions would allow a real return-on-capital calculation. Profit that finances replacements is value; profit distributed while equipment ages is not.
The eighth is operating depth. Shift coverage, contractor dependencies, escalation roles, spare inventory, recovery drills and succession for key technical knowledge would determine whether a six-person structure is efficient or brittle. Named customer references should be checked specifically on nights, weekends and difficult incidents.
The ninth is IPv6 delivery. A visible Adant-originated IPv6 route, valid route authorization, customer allocation policy and tested failover would turn "possible" IPv6 into operating evidence. Until then, the active /22 and its valid RPKI state are the stronger facts.
Judgment
Adant appears to have found a rational place in Moscow's connectivity economy. It is small, profitable on the latest available accounts, linked to a real ASN and IPv4 block, and positioned at a building where interconnection has genuine economic value. Its entry prices are competitive without being implausibly low. The company can create value for customers that need a few units, local hands and carrier access but cannot justify their own facility team.
The same evidence limits the conclusion. The public network is small and currently concentrated behind RECONN. The latest detailed financial year is 2024. Facility ownership and wholesale tenure are unclear. A six-person headcount may be efficient or exposed. Licence-annulment entries need explanation. Public uptime, contract and customer-concentration data are absent.
The cash-flow test is therefore not whether Adant can collect growing rent. It is whether the company can keep enough of that rent after power, site, transit, labour and replacement costs to preserve the redundancy customers believe they bought. Revenue growth creates value only when maintenance capital is funded, suppliers remain replaceable and customers renew because failures are handled well. Until those facts are available, Adant merits a cautiously positive operating assessment and a demanding contractual one.