Summary

The buyer is outsourcing a burden, not buying a cabinet

The buyer archetype is a regulated Moscow enterprise IT director with a room full of aging servers in the basement of a head office. The room has a dedicated air conditioner, a UPS that passed its last battery test only after replacement cells were rushed across town, and a laminated visitor log that satisfies no serious audit. The servers still run billing, archive, HR, document-management and reporting workloads. Moving every workload to cloud would simplify procurement in one sense, but it would raise data-location, migration, legacy-application and internal-control questions. Building a new private server room would require electrical work, cooling design, fire suppression, access control, monitoring, spare-parts planning and a capital request that looks difficult in a sanctions-era budget. Deferring modernization is possible until the next outage. Another Russian data centre is possible. Regional office hosting is possible. A domestic cloud migration is possible. DATAPRO's offer is judged against all four substitutes, not against an empty rack brochure.

That is why the first economic unit must be defined narrowly. The customer is buying a Moscow data-centre rack, power and managed continuity contract. In public language this may appear as colocation, rack-place rental, an equipment-placement service or a machine-room service. In economic terms it is the right to put owned or customer-controlled equipment into an engineered environment where power, cooling, access, monitoring, carrier options and continuity processes already exist. The rack is the invoice line the buyer can understand. The burden being transferred is the hard part: electrical resilience, cooling resilience, physical access governance, audit evidence, remote intervention, supplier maintenance, carrier diversity, capacity reservation and an escalation path when the business is down.

DATAPRO's own website makes the capacity wrapper visible. The company describes itself as a leading independent data-centre operator in Russia and says its network consists of owned geographically distributed data centres, including four sites at Tier III to Tier IV levels (https://datapro.ru/about). The home page describes DATAPRO as Russia's largest independent operator of Tier IV-III certified data centres (https://datapro.ru/). Its official data-centre pages give the machinery behind the rack. DataPro Moscow I at Aviamotornaya Street lists realized capacity of 2,000 rack places, design capacity of 4,000 rack places, more than 20 MW of design power, two independent electrical inputs, 24/7/365 engineering staff, a carrier-neutral approach and 31 telecom operators on site (https://datapro.ru/data-centers/datapro-moscow-i). DataPro Moscow II in Khimki lists 1,500 realized rack places, 11 MW of total power, two independent electrical inputs, cooling redundancy, 20 telecom operators, two cross-connect rooms and three optical inputs (https://datapro.ru/data-centers/datapro-moscow-ii). Moscow III and Moscow IV on Ryabinovaya Street extend the same operating pattern, with public pages listing 30 MW and 9 MW of total power respectively, 24/7/365 engineering staff, security, carrier-neutrality and on-site telecom options (https://datapro.ru/data-centers/datapro-moscow-iii, https://datapro.ru/data-centers/datapro-moscow-iv).

The article's opening substitute has to return throughout the analysis. A private server room can look cheaper because the enterprise already owns the servers and already pays the lease. But the private room hides costs in the wrong ledgers. Power resilience is buried in facilities. Cooling failures appear as emergency maintenance. Security sits with office administration. Network diversity appears in carrier contracts. Compliance evidence becomes a scramble before each customer or regulator review. Remote hands becomes a taxi ride by the one engineer who knows which cable is which. The rack DATAPRO sells is not just a cabinet. It is a way to convert those scattered, fragile and sometimes invisible costs into a more formal service relationship.

The strongest thesis is not that every regulated buyer should use DATAPRO. It is that DATAPRO's rack is valuable when the buyer has enough Russian locality, uptime, evidence and physical-control demand to make an in-house room a false economy. The weaker thesis is that a rack is automatically superior because it sits in a data centre. That weaker claim does not survive procurement. A buyer still needs contract terms, power density, support-hour pricing, cross-connect costs, carrier list, security procedure, facility certificate status, incident history and exit rights. The public evidence supports the existence of the rack capacity. It does not support a blind conclusion about profitability or realized service quality.

Identity proof and unit-economics proof are different questions

DATAPRO's public identity can be established without confusing identity with economics. The company site uses the DataPro brand, the official pages name DataPro Moscow I through IV, and PeeringDB lists a facility named DataPro Moscow under the organization "DATAPRO" Limited Liability Company, with a website at datapro.ru, address 69 Aviamotornaya Street, Moscow, Russia, and a last-updated timestamp in September 2025 (https://www.peeringdb.com/fac/7597). That is meaningful identity and facility evidence. It helps tie the Moscow colocation location to the legal name used by public internet infrastructure databases.

It is not, however, a revenue statement. A PeeringDB facility entry proves that the facility is represented in an interconnection database and that networks report presence there. It does not prove how many racks are sold, whether a given rack is powered, whether a customer workload sits in a particular hall, whether the facility is profitable, or whether private service levels have been met. The same boundary applies to DATAPRO's official pages. The pages give design and operating claims: rack capacity, power, cooling, security, monitoring, carrier neutrality and certificates. They are load-bearing evidence for operating surface. They are not audited financial statements.

The public record also contains a useful technical signal. PeeringDB shows DataPro Moscow with 13 networks and three local exchanges: CLOUD-IX MSK, Global-IX and GNM-IX (https://www.peeringdb.com/fac/7597). The networks listed at the facility include RETN, Filanco, GNM, HOSTKEY, T1Cloud, Ruform/RUTUBE, TimeWeb and others, as rendered through the facility page and API data (https://www.peeringdb.com/fac/7597). T1Cloud's own PeeringDB page gives a more detailed example: AS206805 is listed as T1Cloud, also known as LLC "T1Cloud" and T1 Oblako, and its interconnection facilities include DataPro Moscow, Moscow M9 and Moscow TehnoGorod (https://www.peeringdb.com/net/37248). This supports the idea that DataPro Moscow is part of a live Russian cloud and hosting interconnection environment.

Again, it is not a customer-revenue proof. Networks can list a presence for interconnection, transit, hosting, cloud operations, backup or many other reasons. Some may be tenants, some service providers, some network partners. The right use of PeeringDB is bounded: it shows market participation, carrier density and industry self-reporting around the facility. It does not show DATAPRO's contract values or margins.

The public CNews IaaS Enterprise 2025 table gives another market clue. It lists T1Cloud's data centres as including DataPro in Moscow, alongside Rostelecom and IXcellerate, and it also lists Selectel's data-centre estate as including DataPro in Moscow, alongside its own or other sites (https://www.cnews.ru/reviews/rejting_provajderov_iaas_enterprise_2025/review_table/2ea0bce54bdc5e36f7ba959308088831be7fb73a). That is useful because it shows DataPro appearing in the infrastructure stack of other Russian cloud providers. It does not prove end-customer dependence on DATAPRO. It does show that the rack is not merely a retail cabinet proposition; it can be part of another provider's cloud supply chain.

The disciplined reading is therefore two-step. First, identity and operating surface are visible: DATAPRO has official Moscow-area data-centre pages and an industry database facility record. Second, unit economics remain private: utilization, contracted power, price per kW, remote-hands margin, cross-connect revenue, power cost pass-through, outage penalties, audit-driven retention and customer concentration are not public. Any serious judgement must keep those questions separate.

Compliance capacity is the pricing logic

For the regulated Moscow buyer, the rack price is not only rent plus electricity. It is a price for auditable capacity. Russian personal-data law, critical-information-infrastructure rules, sectoral security demands and customer contract reviews all push certain workloads toward demonstrable control over where and how systems are operated. Russia's Federal Law on Personal Data, 152-FZ, is the core personal-data statute; its public legal text identifies the federal law, the obligations chapters and later amendments, including the 2014 amendments associated with data-localization requirements (https://www.consultant.ru/document/cons_doc_LAW_61801/). The critical-information-infrastructure law, 187-FZ, creates another regulatory frame for significant information systems and their security obligations (https://www.consultant.ru/document/cons_doc_LAW_220885/). A colocation provider does not make a customer compliant by magic. It sells physical and procedural evidence the customer may need to assemble its own compliance case.

That distinction matters. A DATAPRO rack is not a legal opinion. It is not a substitute for the customer's data map, access policy, threat model, encryption controls, backup policy, incident runbook or regulator notification process. But a data-centre rack can answer hard evidence questions that a small private server room often cannot answer convincingly. Where is the equipment? Who can enter? Is access logged? How is video retained? Are the engineering systems monitored around the clock? Are there independent power inputs? What certificates can the site show? Are telecom providers available? Can the customer's duty team sit near its own equipment? DATAPRO's public pages speak directly to those questions.

The official Moscow I page lists Uptime Institute Tier III Design, Facility and Operation certificates as well as ISO/IEC 27001, ISO 9001 and ISO/IEC 20000-1 labels, though the page uses older or misspelled standard versions in places and should be treated as public self-presentation rather than a fresh certificate audit (https://datapro.ru/data-centers/datapro-moscow-i). Moscow II lists Uptime Institute Tier IV Design and Tier IV Facility (https://datapro.ru/data-centers/datapro-moscow-ii). Moscow III and IV pages show Tier IV status language and certificate sections, with notes about Uptime Institute Tier IV Design certification in process on the rendered page text (https://datapro.ru/data-centers/datapro-moscow-iii, https://datapro.ru/data-centers/datapro-moscow-iv). Uptime Institute's own Tier Certification overview explains why these labels matter commercially: Tier Certification verifies whether a facility is designed, constructed and operated to Tier requirements, and Tier III means components and distribution paths can be removed for planned work without impacting operations, while Tier IV adds fault tolerance for individual equipment failure or distribution-path interruption (https://uptimeinstitute.com/tier-certification, https://uptimeinstitute.com/tiers).

The pricing logic is not the certificate alone. It is the combination of certificate evidence with local execution. A Moscow enterprise that has to satisfy a bank, insurer, public-sector customer or foreign partner may need to show both locality and process. It needs a site address, an access procedure, a power design, a cooling design, a security procedure, records, maintenance windows, incident escalation and a customer-facing document pack. A cheap office room can rarely provide that. A large public cloud may provide far richer automation, but it may not solve all legacy-hardware, licensing, data-location or physical-control concerns. DATAPRO's value lies in the middle: the customer keeps enough control over its own equipment while outsourcing the facility layer to an operator whose public pages are built around audit-friendly facts.

This is why compliance capacity should be understood as a capacity product. A rack that can support high-density equipment, audited access and documented facility controls is a scarce input for buyers with Russian workloads. The customer is not only paying for today. It is paying for the option to survive the next customer audit, the next internal risk review, the next regulator question, the next migration delay and the next incident without admitting that core systems still depend on an improvised server room.

Power density makes the rack real

Power turns the rack from furniture into infrastructure. DATAPRO's public pages are power-heavy because power is the constraint buyers cannot fake. Moscow I lists more than 20 MW of design power, two independent power inputs, an IT redundancy scheme shown as 4/3N, dynamic UPS equipment from KINOLT and Hitec Power Protection at 1,670 kVA, Siemens and Eaton distribution substations and 1,600 kVA dry transformers (https://datapro.ru/data-centers/datapro-moscow-i). Moscow II lists 11 MW of total power, two independent inputs, IT redundancy shown as 4/3N, Vertiv/Emerson UPS at 1,200 kW, 2N power redundancy for cooling systems, 1,520 kW diesel rotary UPS units, 10 kV switchgear and 1,600 kVA transformers (https://datapro.ru/data-centers/datapro-moscow-ii). Moscow III lists 30 MW of total power; Moscow IV lists 9 MW (https://datapro.ru/data-centers/datapro-moscow-iii, https://datapro.ru/data-centers/datapro-moscow-iv).

Those numbers are not just engineering color. They tell the buyer what burden is being transferred. A private server room may have one building feed, a limited breaker panel, a UPS sized for a few cabinets, a generator only for office life-safety systems, and no clean way to expand. When a new storage array or GPU server arrives, the office room becomes a negotiation with facilities. In a purpose-built data centre, the buyer is supposed to reserve a slice of already-planned electrical capacity. That does not eliminate power risk, but it changes the operating model from improvisation to capacity management.

The cost base is also visible. Power systems are capital-intensive and maintenance-intensive. A dual-input, monitored, redundant data-centre power chain has switchgear, transformers, UPS systems, generators, fuel arrangements, protective rooms, fire protection, controls, testing and staff. It must be serviced before the customer knows whether every rack will sell at the desired margin. The customer sees a rack charge and power commit. The operator sees capex, maintenance, spares, power procurement, thermal limits, compliance documentation and incident risk.

This is where sanctions-era procurement becomes part of the economics. DATAPRO's own pages name foreign or internationally branded equipment families: Schneider Electric EcoBreeze, Vertiv/Emerson UPS, Siemens and Eaton distribution equipment, KINOLT and Hitec Power Protection (https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/data-centers/datapro-moscow-ii). That does not mean DATAPRO lacks support or spares. It does mean buyers should ask how the operator maintains Western-origin or internationally sourced equipment after 2022, what local substitutions are approved, which spare parts are stocked on site, how warranty or service channels work, and how changes affect certificate or audit evidence. The public record does not answer those questions. It makes them central.

The private server room substitute looks worse under this lens. The same procurement constraints affect the enterprise, but the enterprise has less scale. It may have to find one replacement UPS board, one air-conditioning compressor, one fire-suppression vendor, one network contractor and one facilities engineer under urgency. DATAPRO can pool those problems across a larger estate. Whether it does so efficiently is private. The economic hypothesis is that pooled expertise, spares, service relationships and operating discipline make the rack cheaper than the buyer's own emergency maintenance cycle.

Cooling converts electrical capacity into uptime capacity

Power without cooling is a liability. DATAPRO's public cooling claims are therefore part of the same purchased unit. Moscow I lists Schneider Electric EcoBreeze modular cooling and precision air-conditioning systems (https://datapro.ru/data-centers/datapro-moscow-i). Moscow II, III and IV list N+1 cooling redundancy, independent cooling systems and 125 kW modular cooling structures, with cooling-system power backed by 2N redundancy on the Moscow II-IV pages (https://datapro.ru/data-centers/datapro-moscow-ii, https://datapro.ru/data-centers/datapro-moscow-iii, https://datapro.ru/data-centers/datapro-moscow-iv). DATAPRO's media archive also preserves a PCWeek-linked interview summary about achieving average annual PUE in the 1.25 to 1.3 range and the cooling-design criteria behind that target (https://datapro.ru/media/cod-v-megapolise-ehkonomiya-na-vozdushnom-ohlazhdenii).

Cooling is a good example of why the rack is compliance capacity, not only uptime capacity. A customer audit may ask whether environmental conditions are monitored, who responds to alarms, how maintenance is handled and whether critical indicators are supervised around the clock. DATAPRO's pages repeatedly state that critical indicators of engineering systems are fed into a single monitoring system and watched 24/7/365 (https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/data-centers/datapro-moscow-iii). That evidence matters for customers whose own office cooling is one compressor and a facilities ticket.

The buyer should still be skeptical. Public pages do not show actual PUE time series, rack inlet temperatures, thermal incidents, maintenance records, seasonal derating, chilled-water risks, air-side contamination controls or high-density utilization by hall. Cooling claims are design and operating-surface evidence. They are not a guarantee that every customer rack can draw any desired load. A serious procurement process would ask for contracted kW per rack, density limits, hot-aisle/cold-aisle rules, blanking-panel policy, sensor data, escalation windows and whether the customer's actual hardware profile fits the facility.

The economic point remains. Cooling shifts from a small-office incident to a professionally managed service. That is especially important for customers caught between modernization paths. If the buyer is not ready to rewrite the application, move to a domestic cloud platform, change licenses or abandon owned hardware, a colocation rack can buy time. It turns a cooling emergency into a capacity reservation.

Security and remote hands sell audit confidence

Physical security is not decorative. It is the part of colocation that many enterprise buyers underestimate until the first audit or incident. DATAPRO's Moscow I page says security works 24/7/365, the building and territory have more than 350 video cameras with motion detectors, video records are stored for at least 90 days, important zones are controlled by an access-control system with contactless-card readers and mechanical locks, and there are entrance vestibules to machine halls (https://datapro.ru/data-centers/datapro-moscow-i). Moscow II and IV list more than 150 cameras, while Moscow III lists more than 450, along with the same 90-day video-retention and access-control language (https://datapro.ru/data-centers/datapro-moscow-ii, https://datapro.ru/data-centers/datapro-moscow-iii, https://datapro.ru/data-centers/datapro-moscow-iv).

This changes the compliance conversation. An in-house room often has a key, a badge reader, a camera that also covers the corridor, and a spreadsheet of authorized staff. A professional data centre should have a visitor process, video retention, access-zone segregation, machine-room vestibules, security staff and customer access records. The buyer is not just trying to prevent theft. It is trying to prove who could touch the machine, when, and under which procedure.

Remote hands and nearby customer offices add another layer. DATAPRO's pages repeatedly say the customer can place its duty service near its own equipment and can rent office and warehouse space (https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/services/other/rent). The office-and-warehouse rental page frames office rental as a solution for customers that place large volumes of equipment and do not want data-centre staff to access it, allowing the customer's personnel to be located there permanently or temporarily (https://datapro.ru/services/other/rent). That is economically important. Some customers want the provider to intervene. Others want physical proximity while preserving customer-only handling of sensitive equipment. DATAPRO's public material suggests it sells both the facility layer and the surrounding working environment.

For a regulated enterprise, this can be the difference between a viable colocation plan and an audit objection. The enterprise may not want external technicians replacing drives in a banking-security appliance, but it may also not want to dispatch a specialist across Moscow in a traffic jam at 2 a.m. A nearby duty room, office lease or warehouse option can reduce that contradiction. It also turns colocation into more than rack rent. DATAPRO can monetize office space, warehouse space, customer duty-team proximity and operational convenience around the rack.

The margin risk is labor. Remote hands, escorted access, ticket handling, audit support, security checks and customer coordination consume staff time. If the operator prices only the cabinet and gives away too much labor, the rack can become a low-margin support contract. If it charges every intervention aggressively, customers may feel trapped. Public pages do not disclose remote-hands rates, response times, ticket volumes or support gross margin. The public evidence supports the existence of the labor layer. It does not prove its economics.

Carrier access makes the rack a platform rather than a closet

A rack with one internet provider is a remote server closet. A rack with carrier choice is infrastructure. DATAPRO's public pages repeatedly stress a neutral approach to the customer's choice of communications provider. Moscow I says 31 telecom operators provide services on site (https://datapro.ru/data-centers/datapro-moscow-i). Moscow II, III and IV each say 20 operators provide services, with cross-connect rooms and optical inputs called out on the later pages (https://datapro.ru/data-centers/datapro-moscow-ii, https://datapro.ru/data-centers/datapro-moscow-iii, https://datapro.ru/data-centers/datapro-moscow-iv). PeeringDB adds a market check: DataPro Moscow has local exchanges and networks listed, including RETN, Filanco, GNM, HOSTKEY, Ruform/RUTUBE, T1Cloud, TRUNK MOBILE, TimeWeb and others (https://www.peeringdb.com/fac/7597).

Carrier neutrality matters because customers buy optionality. A regulated enterprise may need one carrier for corporate WAN, another for internet transit, another for private interconnection, another for backup replication, and another for a cloud provider. A hosting company may need peering and transit choices. A cloud provider may need exchanges and upstream diversity. A content platform may need low-latency access to Russian networks. Carrier density is not the same as reliability, but it gives the buyer more routes out of a procurement problem.

It is also a source of competitive advantage against the private server room. A head-office room is limited by the building's fiber risers, landlord rules, street works and carrier willingness. A data centre can aggregate demand and justify cross-connect infrastructure. That does not mean every cross-connect is cheap. Buyers need cross-connect pricing, installation lead times, monthly fees, meet-me-room rules, optical diversity and carrier contract terms. But the existence of 20 to 31 on-site operator claims on DATAPRO pages and the PeeringDB facility record make carrier access a real part of the economic unit.

This is why the article treats ASNs and PeeringDB rows as evidence only. RETN, HOSTKEY, T1Cloud or Ruform are not the subject of the article. Their presence signals that DataPro Moscow is part of a live ecosystem. It does not reveal DATAPRO's internal network architecture or the customer contracts behind any listed network. The buyer should use the carrier list as a diligence starting point, not a shortcut.

The customer mix can make the rack sticky

Customer stickiness in colocation often comes from inconvenience rather than affection. Once a buyer has racked equipment, cabled carriers, documented controls, passed an audit, trained support staff, and built runbooks around a site, moving becomes a project. That is good for the operator if service quality holds. It is dangerous if the operator underinvests, because customers can tolerate friction until a single incident makes migration politically easier.

DATAPRO's public market signals point to several possible customer types. First, cloud providers can use the facility as an infrastructure location. The CNews IaaS Enterprise 2025 table names DataPro as a Moscow data-centre location for T1Cloud and Selectel in their cloud infrastructure listings (https://www.cnews.ru/reviews/rejting_provajderov_iaas_enterprise_2025/review_table/2ea0bce54bdc5e36f7ba959308088831be7fb73a). Second, hosting and content networks appear in PeeringDB's DataPro Moscow facility list, including HOSTKEY, Eurobyte, TimeWeb, Ruform/RUTUBE and ATLEX (https://www.peeringdb.com/fac/7597). Third, enterprise or network-service users appear through T1Cloud, TRUNK MOBILE, GNM, Filanco and other network names in the same facility record (https://www.peeringdb.com/fac/7597, https://www.peeringdb.com/net/37248).

These are market signals, not audited customer lists. They show what kinds of buyers or network actors have reason to be around the facility: cloud, hosting, content, transit, exchange, enterprise and telecommunications. That diversity can improve DATAPRO's resilience if no single customer dominates. It can also increase operational complexity because a cloud provider, a content network and a regulated enterprise have different remote-hands, access, density, incident and audit needs.

The customer mix matters for the rack economics. A cloud provider may buy larger blocks, demand aggressive price per kW, bring its own operations discipline and request predictable power. A regulated enterprise may buy fewer racks but pay for audit support, access controls, documentation and office proximity. A hosting provider may be sensitive to power price, abuse handling, IP reputation and remote intervention. A network operator may value cross-connects and exchange presence. The same rack product can therefore carry very different margins depending on the customer.

The public evidence does not show which mix dominates. That is a key missing proof category. A buyer or investor would want the percentage of racks sold to cloud providers, hosting operators, enterprises, state-linked customers, financial institutions and content platforms. It would also want renewal rates by cohort. A rack that supports a regulated customer after a difficult audit may be very sticky. A rack used for price-sensitive hosting can churn if power or network pricing moves.

Competition is mostly about who absorbs the hidden work

DATAPRO competes against several substitutes at once. The first is the customer's own room. The second is another Moscow or Russian colocation site. The third is domestic public cloud or private cloud. The fourth is a regional office hosting workaround. The fifth is delayed modernization, which is always a competitor because many IT departments survive by doing nothing until risk becomes visible.

Russian market context shows why the competition is serious. CNews's IaaS Enterprise 2025 table lists many providers competing on SLA, protected data-processing levels, certificates, operating systems, virtualization stacks, pricing and data-centre locations, including MTS Web Services, T1Cloud, Rosukrep, ITGlobal.com, Selectel, RTK-Data Center, Aiteko.Cloud, Linx Cloud and VK Tech (https://www.cnews.ru/reviews/rejting_provajderov_iaas_enterprise_2025/review_table/2ea0bce54bdc5e36f7ba959308088831be7fb73a). Those providers are not all direct colocation competitors. But they compete for the same workload decision. A customer that can move to a domestic cloud does not need to own the server. A customer that needs owned hardware may still choose another data centre. A customer with low audit burden may keep the office room alive.

This is where DATAPRO's rack has to be framed carefully. It is not always cheaper than cloud. Cloud can win when the workload is elastic, the application can be modernized, licensing is cloud-friendly, compliance can be satisfied through the provider's document pack, and the customer values automation over physical control. Colocation wins when the workload is hardware-specific, latency-specific, license-specific, data-location-sensitive, appliance-heavy or migration-constrained. The rack is a bridge for customers that cannot or should not abstract everything away.

Another Russian data centre can win if it offers better geography, price, carrier mix, cross-connect terms, remote hands, uptime record, power density, customer support or regulatory documentation. DATAPRO's public pages show a strong footprint, but they do not show price. A buyer comparing DATAPRO with IXcellerate, DataSpace, Linx, Selectel, Rostelecom, 3data or a regional provider must normalize the quote: included power, usable kW, remote-hands charges, cross-connect fees, access rules, audit documents, office space, migration support, penalties, payment terms and exit costs.

The hidden work is what decides the competition. If DATAPRO absorbs more of the compliance, continuity and procurement burden than the alternatives, it can command a premium. If it merely rents rack space and leaves the customer to assemble the evidence, it competes on price. The official pages suggest a premium story because they emphasize certificates, ownership, security, monitoring, engineering staff, carrier neutrality and ancillary office space (https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/services/other/rent). The private metrics decide whether customers actually pay for that story.

The sanctions-era cost base is a proof problem

The operating risk in Russia's data-centre market is not only demand. It is maintenance under constraint. A data centre is a long-lived asset full of equipment whose supply chain, warranty status and service relationship can change faster than the building. DATAPRO's public pages name international equipment brands across cooling, power distribution and UPS systems (https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/data-centers/datapro-moscow-ii). The buyer should therefore ask a practical question: does the operator have the spare parts, service knowledge, substitution approvals and documentation needed to keep the site reliable without creating audit or certification uncertainty?

This is not a DATAPRO-specific accusation. It is a market condition. Since 2022, Russia-facing technology procurement has been shaped by sanctions, export controls, supplier withdrawals, indirect import channels and domestic substitution. For data-centre customers, the impact is concrete. A replacement breaker, UPS module, cooling controller, server part, storage shelf, network optic or support contract can become a lead-time and documentation issue. A private server room has little bargaining power in that environment. A data-centre operator may have more scale, but it also carries more critical equipment.

Compliance buyers should not accept generic reassurance. They should ask for maintenance history, spare-parts inventory policy, vendor substitution policy, incident records, generator-test results, UPS maintenance windows, cooling spare strategy, control-system patching, documented change management and whether certificate claims remain current after equipment changes. Uptime Institute emphasizes that Tier Certification verifies application of the Tier Standard across design, construction and operations and that operational certification concerns the ability to plan for continuous production during normal operating conditions (https://uptimeinstitute.com/tier-certification). If the equipment base changes under procurement pressure, the operator's evidence pack matters more, not less.

The same logic applies to energy. Moscow I's public page lists more than 20 MW design power, Moscow II lists 11 MW, Moscow III lists 30 MW and Moscow IV lists 9 MW (https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/data-centers/datapro-moscow-ii, https://datapro.ru/data-centers/datapro-moscow-iii, https://datapro.ru/data-centers/datapro-moscow-iv). Energy is not only a utility line. It is tariff exposure, capacity planning, generator maintenance, fuel logistics, cooling load and customer metering. The rack price has to recover that cost while remaining competitive against cloud and other colocation sites.

This is why the article treats DATAPRO's rack as compliance and uptime capacity. The buyer is paying the operator to carry a bundle of hard-to-replicate risks. The proof question is whether DATAPRO can price those risks profitably while still giving the customer a cheaper and safer outcome than the substitutes.

The quote should be read as a risk-transfer schedule

A buyer should not read a DATAPRO rack quote as a simple rent table. It should be read as a schedule of which risks remain with the customer and which have been transferred to the operator. The obvious lines are rack size, power commit, extra power, cross-connects, internet or carrier charges, remote hands, storage space, office space and contract term. The less obvious lines are the ones that decide whether the rack actually solves the server-room problem: who installs equipment, who may open the cabinet, who signs access records, who changes cabling, who keeps spare optics, who responds to temperature alarms, who escorts an auditor, who documents a maintenance event, who pays for emergency work, and who decides when a load cannot be added to a cabinet.

DATAPRO's public pages make several of those questions visible but not complete. Moscow I says the customer can use office and warehouse space, that 31 operators provide services on site, and that the customer's duty service can be placed near its own equipment (https://datapro.ru/data-centers/datapro-moscow-i). Moscow II-IV repeat similar language around customer duty-service proximity, office rental, warehouse rental and cloud-service availability, while the separate office-and-warehouse page explains the use case more plainly: customers placing large amounts of equipment may want their own personnel nearby rather than allowing data-centre employees to access the equipment (https://datapro.ru/data-centers/datapro-moscow-ii, https://datapro.ru/data-centers/datapro-moscow-iii, https://datapro.ru/services/other/rent). That is the commercial boundary a regulated buyer has to price. If the customer insists on customer-only handling, it may need its own duty staff, office space and stricter access process. If the customer accepts operator remote hands, it needs response-time, liability and procedure detail.

The quote should also expose the difference between reserved capacity and consumed capacity. An in-house server room usually fails because no one knows the real cost of reserving capacity. The air conditioner is there until it breaks. The UPS is there until the batteries fail. The spare circuit is there until a dense server consumes it. In a colocation quote, reservation becomes explicit. The buyer may reserve a rack and a power envelope before equipment arrives. That reservation has a cost even if the servers are lightly used. The operator has to keep upstream power, UPS capacity, generator capacity, cooling, floor space, monitoring and staff ready for use. If the buyer expects growth, the value is in having capacity waiting. If the buyer's workload is shrinking or ready to move to cloud, the same reservation can become waste.

That is why a good DATAPRO procurement memo would separate three prices. The first is the cash price, including rack, power, network, cross-connects, support, office space and taxes. The second is the avoided cost of the private server room: electrical remediation, cooling replacement, fire-suppression upgrades, access control, monitoring tools, audit remediation, overtime, emergency supplier calls and downtime risk. The third is the opportunity cost of not modernizing: applications that remain tied to owned hardware may be safer in a professional data centre, but they may also delay architecture changes that cloud or managed platforms would force. The right conclusion can be "lease the rack for five years," "lease the rack only for the legacy systems," "move the predictable workloads to domestic cloud and keep appliances in colocation," or "replace the application and avoid new hardware." DATAPRO's rack is the answer only for the workloads whose risk-transfer value exceeds those alternatives.

The same logic should discipline customer audit language. A rack in a certified data centre is useful evidence, but it is not the whole evidence file. The customer still needs its own controls. Uptime Institute says Tier Certification verifies the application of Tier Standards to ensure a facility is designed, constructed and operated to specifications, and its scope includes electrical systems, structural factors, building characteristics, mechanical systems, management and operations, site location, on-site power production, occupational safety, physical security, equipment maintenance and related fields (https://uptimeinstitute.com/tier-certification). That evidence helps the customer, but it does not replace identity-and-access management, backup testing, vulnerability management, encryption, application logging, incident response or data-retention policy. If a salesperson collapses those layers into "the site is certified, therefore the workload is compliant," the buyer should push back.

The strongest DATAPRO case is therefore a hybrid case. Keep hardware where hardware control is needed. Use the facility where physical controls and power/cooling evidence are hard to reproduce. Use cloud where application economics and elasticity matter more than cabinet control. Use office or duty-room rental only where customer-only handling is worth the extra labor. Use the carrier-neutral site where network diversity is valuable enough to pay cross-connect charges. The rack becomes an operating option, not an ideology.

This quote discipline also protects DATAPRO's side of the bargain. A data-centre operator can destroy margin by accepting customers whose operational burden is not priced. A customer that wants unlimited escorted visits, frequent emergency remote hands, bespoke audit documents, unusual cabling rules, high-density racks, special network changes and customer-only handling should not be priced like a simple cabinet tenant. If DATAPRO's public premium story is real, the operator should be able to charge for the complexity it absorbs. If it cannot, then compliance capacity becomes a marketing label rather than an economic engine.

Redundancy has to earn its renewal

The renewal decision is where DATAPRO's rack either proves itself or becomes an expensive habit. In the first year, a customer may move because the private server room is no longer defensible. By the third year, the customer knows whether the data centre has actually reduced work. Has the audit process become easier? Did the operator produce certificate, access, maintenance and environmental evidence quickly? Did carrier ordering get faster? Did remote hands reduce emergency travel? Did the customer avoid outages that would have hit the old room? Did power and cooling support the actual hardware rather than only the quoted average rack? Those questions decide renewal more than the rack list price.

Power, cooling and carrier diversity each have their own renewal metric. For power, the buyer should track unplanned power events, maintenance windows, A-feed/B-feed behavior, generator-test communication, power-quality incidents, cabinet load warnings, and the difference between contracted and actually usable capacity. DATAPRO's public pages make power central, with more than 20 MW at Moscow I, 11 MW at Moscow II, 30 MW at Moscow III and 9 MW at Moscow IV, plus two independent electrical inputs on each official site page (https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/data-centers/datapro-moscow-ii, https://datapro.ru/data-centers/datapro-moscow-iii, https://datapro.ru/data-centers/datapro-moscow-iv). Renewal depends on whether that power architecture translated into fewer business interruptions and cleaner capacity planning for the customer.

For cooling, the buyer should track thermal alerts, hot-spot remediation, density exceptions, seasonal restrictions, response times and whether the operator made the customer change equipment layout, blanking, cabling or load distribution. DATAPRO's cooling claims are credible as operating-surface evidence because the pages name modular cooling, N+1 schemes, precision air conditioning and 24/7 monitoring (https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/data-centers/datapro-moscow-iii). Renewal evidence is more granular. A customer renewing a high-density rack wants proof that the cooling system served its actual load, not just that a certificate or design target exists.

For carriers, the buyer should track cross-connect installation time, recurring cross-connect charges, carrier outage isolation, path diversity, latency to key Russian networks, the ease of adding a second provider and whether a carrier dispute required DATAPRO intervention. Moscow I's 31-operator claim and the 20-operator claims on the other sites are commercially important because carrier diversity can lower the customer's switching cost and improve resilience (https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/data-centers/datapro-moscow-ii). PeeringDB's DataPro Moscow page gives an external signal of network density through local exchanges and listed networks (https://www.peeringdb.com/fac/7597). But a renewal committee will ask whether those choices were usable in procurement practice, not just visible in a database.

The SLA should be read the same way. A service-credit clause can make a contract look disciplined, but the meaningful question is whether the customer had to invoke it. If it did, how quickly was the incident acknowledged, how clearly was root cause explained, and did the credit matter relative to the customer's business loss? The buyer should compare DATAPRO not with an idealized data centre, but with the old private room and with a plausible cloud or local-colocation alternative. If the old room would have required six emergency visits and DATAPRO required none, the rack earned part of its renewal. If cloud would have avoided hardware dispatch entirely, DATAPRO has to justify why physical control was still worth the added operating work.

Remote hands are the most revealing metric because they expose the labor hidden inside colocation. A buyer should track tickets per rack per month, emergency tickets, median and tail response time, first-touch fix rate, repeat visits, after-hours charges, customer-caused incidents, provider-caused incidents, access delays, and whether the operator's staff could follow customer runbooks without improvisation. DATAPRO's public material supports the existence of a 24/7/365 on-site engineering and monitoring model (https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/data-centers/datapro-moscow-iv). The private evidence that would change the judgement is whether that labor model improves reliability without turning every change into a bespoke bill.

The private metrics that would change the judgement

The public record supports a meaningful thesis but leaves the final judgement open. The first private metric is sold usable power. Rack counts are useful, but data-centre economics increasingly depend on kW per rack and on how much of the designed load is contracted. DATAPRO's pages list rack counts and total MW figures, but they do not show average contracted kW, peak load, headroom, power utilization or how many high-density racks can be supported simultaneously (https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/data-centers/datapro-moscow-iii). A facility with many racks but constrained power sells a different product from a facility with fewer racks and high contracted density.

The second metric is audit conversion. How often does DATAPRO help customers pass bank, public-sector, critical-infrastructure, personal-data or enterprise-security audits? Which certificates are current, independently verifiable and accepted by customers? What documents are included without extra charge? How many customers require site visits? How often does an audit produce remediation requests? Public certificate labels are useful, but the commercial value comes from customer acceptance.

The third metric is incident history. A buyer needs power incidents, cooling incidents, network incidents, security incidents, fire-system events, access-control failures, remote-hands response times and customer-impact minutes. Uptime language is meaningful, but it is not the same as an incident log. Service credits and guarantees matter only after the buyer knows how often the guarantee is tested.

The fourth metric is support economics. Remote hands, escorted access, office proximity, warehouse use, cross-connect installation and customer audit support can be profit centers or margin leaks. DATAPRO's public pages say the customer can locate duty staff near equipment and rent office or warehouse premises (https://datapro.ru/services/other/rent). That is commercially interesting. It also means labor and real-estate services surround the rack. The operator's margin depends on whether those services are priced with discipline.

The fifth metric is customer concentration. PeeringDB and CNews show cloud, hosting, network and content signals around DataPro Moscow, but not the revenue mix (https://www.peeringdb.com/fac/7597, https://www.cnews.ru/reviews/rejting_provajderov_iaas_enterprise_2025/review_table/2ea0bce54bdc5e36f7ba959308088831be7fb73a). One anchor customer can improve occupancy and finance expansion, but it can also create renewal risk. Many small customers diversify revenue but raise support overhead. A premium regulated-enterprise base is different from a price-sensitive hosting base.

The sixth metric is spare-parts resilience. A site built with complex international equipment needs maintenance evidence under current procurement conditions. The buyer should ask not only whether equipment was well specified, but how it will be maintained now. That answer could materially raise or lower the value of DATAPRO's rack.

Final judgement: the rack is worthwhile when it beats the server-room fiction

Return to the regulated Moscow enterprise with the fragile head-office server room. Keeping that room alive can feel prudent because the servers are already paid for and the team knows the environment. But the room is often a fiction. Its cost is scattered across facilities, IT overtime, audit remediation, emergency procurement, customer anxiety, network limitations and modernization delay. Its uptime depends on people and components that were never meant to carry regulated workloads indefinitely.

DATAPRO's rack is compelling when it breaks that fiction. It lets the buyer keep hardware control while buying a professionalized facility layer: Moscow locality, owned data-centre sites, rack capacity, multi-megawatt power, cooling, security, monitored engineering systems, carrier access, certificate evidence, remote labor and nearby duty-team options (https://datapro.ru/about, https://datapro.ru/data-centers/datapro-moscow-i, https://datapro.ru/services/other/rent, https://www.peeringdb.com/fac/7597). That is not a generic company profile. It is an economic claim about burden transfer.

The claim is strongest for workloads that cannot move cleanly to cloud, cannot tolerate the office room, and need Russian evidence in front of customers or regulators. It is weaker for workloads that can be rebuilt on a domestic cloud platform, for low-risk systems where downtime is tolerable, or for customers whose main concern is lowest possible monthly price. It is also weaker if DATAPRO's private metrics show poor utilization, high incident rates, weak spare-parts resilience, slow support, stale certificates or customer churn.

The public evidence favors DATAPRO on operating surface. Its own pages show a sizeable Moscow-area data-centre platform. PeeringDB confirms industry database presence and a carrier/cloud/content ecosystem around DataPro Moscow. CNews shows other Russian cloud providers listing DataPro as Moscow infrastructure. Uptime's framework explains why Tier evidence matters for maintenance and fault tolerance. Russian personal-data and critical-infrastructure law explain why locality and audit evidence can become economic value rather than mere compliance theater (https://www.peeringdb.com/fac/7597, https://www.cnews.ru/reviews/rejting_provajderov_iaas_enterprise_2025/review_table/2ea0bce54bdc5e36f7ba959308088831be7fb73a, https://uptimeinstitute.com/tier-certification, https://www.consultant.ru/document/cons_doc_LAW_61801/).

The proof gap is profitability and realized service quality. That gap should not erase the thesis. It should discipline it. DATAPRO's rack is worth paying for when it replaces a bundle of hidden risks with documented, local, power-backed, cooled, secured and staffed capacity. If the buyer merely wants a cheaper place to put servers, the market will offer many arguments. If the buyer needs compliance capacity and uptime capacity in Moscow, the rack is the real product.