The pool of tenants for computing capacity in Russia remains highly restricted: cryptocurrency miners, the public sector and corporate customers subject to data localisation requirements. Meanwhile, the truly big catch, international operators of artificial intelligence systems, whose demand for data-centre services could help Russia recoup investments in the sector, is passing the country by. Besides favourable climates and electricity tariffs, such clients require reliable institutions and access to equipment based on the latest chips, an area from which Western sanctions have effectively cut Russia off. Yet even if Russia focuses solely on developing its own ‘sovereign’ artificial intelligence, it still faces major challenges, not least overcoming a structural shortage of power generation.
A jurisdiction may be attractive to cryptocurrency mining farms while remaining entirely unappealing to AI operators.
Russia is a prime example of such asymmetry, according to experts from the Roscongress Foundation in their report, Criteria for Selecting Data Centre Locations and Russia’s Position.
Electricity prices, for example, are a key factor in the economics of data centres. Industrial tariffs for large consumers vary almost tenfold worldwide, ranging from USD 0.028 per kilowatt-hour in Iran to USD 0.295 in Ireland.
In Russia, according to the report, the average industrial tariff amounts to approximately USD 0.046 per kilowatt-hour. In the Irkutsk region it stands at USD 0.025–0.03, a level close to the lowest globally and comparable with countries such as Paraguay.
Another criterion is the number of hours per year when temperatures remain below 15°C. This determines how long a data centre can operate in free-cooling mode, a technology that uses cool outside air instead of mechanical refrigerationAccording to the World Meteorological Organization, Reykjavik in Iceland, Oulu in Finland and Yakutsk are all situated in cold zones that allow more than 8,000 hours of free cooling annually. Moscow, St Petersburg and Novosibirsk also possess significant free-cooling potential.
The growing adoption of liquid-cooling technologies in next-generation data centres means that access to fresh water is becoming another crucial factor for the industry.
According to the report, the basins of Russia’s major rivers, including the Neva, Volga, Ob, Yenisei, Angara and Lena, possess considerable advantages in this regard. ‘Lake Baikal, which contains around 20% of the world’s surface freshwater reserves, is located in close proximity to Irkutsk and the Angara-Yenisei hydropower system. This is adjacent to a potential AI computing cluster powered by surplus hydroelectric generation,’ the report states. It should be noted, however, that this proposition requires separate assessment from the standpoint of environmental risks.
The authors warn that not every international operator of computing infrastructure is willing to work within the Russian jurisdiction, despite the country’s climatic and tariff advantages.
The main barrier for international AI infrastructure operators in Russia is ‘US export controls and country risk’.
‘The West is attempting to restrain the development of data centres in our country through artificial regulatory restrictions. Russia is the only major economy arbitrarily deprived of a sovereign credit rating from the ‘Big Three’ agencies… The West is also artificially understating Russia’s geopolitical security rating… Yet this indicator is important for most international operators when making investment decisions,’ the report’s authors argue.
Technological restrictions, likewise driven by sanctions, are another obstacle. Modern accelerators produced by Nvidia are designed in the United States and manufactured primarily in Taiwan and South Korea. Since both their design and production depend on US technologies, the US Department of Commerce imposes export controls that determine who receives access to such chips. Russia, the report notes, along with China, Iran and North Korea, belongs to the group of countries for which licensing policy is based on a presumption of denial.
According to the report, cryptocurrency miners generally prioritise low electricity costs and favourable climatic conditions. The world’s largest AI operators, however, require not only natural advantages but also abundant power capacity, access to the latest hardware and a favourable regulatory environment.
Given these constraints, the circle of potential tenants for computing capacity in Russia narrows considerably. It consists primarily of miners, government agencies and corporate customers obliged to localise their data.
Yet even to serve this limited group of current and prospective customers, Russia must undertake additional efforts, above all to prevent a structural deficit in electricity generation.
‘The cost of a kilowatt-hour loses its significance if it is physically impossible to connect the load,’ the experts note. This observation applies to domestic consumers as well.
‘Russia possesses a substantial surplus of hydroelectric generation in Siberia… The combined installed capacity of the Siberian unified power system amounts to 51.9 GW, and its western section is structurally surplus,’ the report explains. ‘By contrast, according to estimates by the System Operator, the south-eastern part of Siberia will face a deficit of 2.88 GW by 2030. The Russian Far East is expected to record a shortfall of 1.68 GW in 2026. The System Operator estimates that preventing the structural deficit projected for 2030 will require total investment of RUB 6 tn.’
Russia’s particular challenge is that large reserves of unused generating capacity are located far from the centres of demand. The Moscow power system is operating at full capacity, while the commercial exploitation of Siberia’s electricity surplus is constrained by insufficient long-distance transmission infrastructure.
Another comparison illustrates the scale of the challenge. Russia’s commercial data-centre market was worth just over RUB 110 bln in 2024. In 2025 it was estimated at around RUB 130 bln, and by the end of 2026 experts expected it to approach RUB 160 bln.
Even including captive data centres, facilities built for the needs of large corporations and government agencies, the total market and investment volume remains incomparable with the sums other countries are committing to AI infrastructure.
US President Donald Trump’s Stargate megaproject, according to public announcements, envisages investment by technology giants of USD 500 bln over four years in AI infrastructure, including data centres and supercomputers. That is equivalent to almost RUB 40 tn over four years, or roughly RUB 10 tn annually.
China, according to sources cited by Bloomberg, is preparing a five-year plan worth CNY 2 tn (USD 295 bln) to build a nationwide network of data centres. The plan outlines ambitions for the ‘aggressive deployment of AI’ (see Nezavisimaya Gazeta, June 17, 2026). In rouble terms, this amounts to more than RUB 21 tn over five years, or over RUB 4 tn annually.
Against the backdrop of such planned digital investments in the United States and China, Russia’s entire data-centre market still looks more like a statistical rounding error.
ORIGINAL: NG/Data Centres in Russia Are Yet to Reach Their Full Potential
