The Russian government plans to consider, by the end of September 2026, the issue of dividing data processing centers (DPCs) into two separate classes: facilities intended for artificial intelligence and all other sites. Formally, the goal is targeted distribution of state support and protection of the budget from misuse of benefits and subsidies. However, behind this seemingly logical initiative lies a whole range of serious problems that could negate all the regulator's efforts.
The essence of the initiative and driving forces
Preparation of the decision has been assigned to the Ministry of Digital Development, and the project's working group is headed by Sberbank's Senior Vice President Andrei Belevets. The idea of the reform has been discussed since 2025 on the instructions of Deputy Prime Minister Alexander Novak. The new rules should guarantee that subsidies and benefits do not go to ordinary commercial data centers that merely formally hide behind the "AI infrastructure" label.
As part of the "Guillotine 2.0" project, the Ministry of Digital Development, the Ministry of Construction, and the Analytical Center under the government must officially clarify the term "DPC" itself by September 1. It is noteworthy that earlier the Ministry of Digital Development gave a negative review of such an initiative, and the article on computing infrastructure ultimately did not make it into the relevant law signed by Vladimir Putin on July 26.
The economics of the issue: who is lobbying for the division
The main argument of supporters of the changes is the colossal scale of required investment. The architecture of AI sites is fundamentally different: servers with graphics processors consume enormous amounts of energy, and the facilities themselves can be built far from end users. The cost of a regular 10 MW DPC is estimated at approximately 20 billion rubles, while an AI site of similar capacity would cost 40 billion rubles.
Representatives of the company "Atomdata" (a Rosatom structure) insist on the urgent introduction of targeted support. According to their calculations, the market's need for new capacity by 2030 will reach 2.5 GW — twice the current volume of all commercial sites in the country. These figures look impressive, but this is exactly where the problems begin.
Criticism: criteria are vague, the market is not ready
The largest operator, "RTK-DPC" from the Rostelecom group, has opposed the new bill. Company representative Dmitry Panyshev confirms the presence of specific characteristics in specialized sites: high connected power and special engineering systems. However, such high-tech infrastructure can easily be used for ordinary traditional computing as well, which makes directly tying subsidies to the wording "for AI" unjustified. The operator proposes evaluating facilities solely by their level of reliability.
Market analysts also point to the complete absence of clear technical criteria. Leading consultant at iKS-Consulting Stanislav Mirin notes that there is no clear boundary for rack parameters: one cannot simply call racks of 15, 20, or 50 kW high-load. The use of liquid cooling also cannot serve as the sole factor. Powerful equipment is often used for standard non-AI tasks: three-dimensional graphics rendering, modeling complex physical processes, and precise strength calculations.
My expert view: The initiative looks premature. There are no objective criteria for a rigid division yet, and a limited circle of companies is engaged in real tasks in this area. The authorities could support such organizations on a case-by-case basis without creating a new class. The industry does not currently count on large-scale state assistance — last week, the ANO "Digital Economy" requested only 72 billion rubles per year from the government for tax and financial benefits for technology development, construction, and modernization of sites. Creating a separate DPC class is more an attempt to streamline future spending than a real mechanism for stimulating the industry.