Growing up in Siberia: how Russia will make money on neural capabilities
The shortage of energy in the world due to the construction of data centers (data centers) for artificial intelligence and cloud computing is becoming a universal problem. In the EU and the USA, the issue is particularly acute due to the rapid rise in electricity prices. Russia, taking advantage of its natural advantages, is implementing a strategy of accelerated construction of data centers beyond the Urals. Siberia and the Far East are gradually turning from suppliers of hydrocarbon raw materials into exporters of computing power. Why it is profitable to build data centers in Russia, how the proximity of China plays into the hands of Siberians and whether there is enough energy in the region for all needs - in the Izvestia material.
The Big Data Boom
As of the beginning of 2026, there are about 194 commercial data centers operating in Russia. Historically, up to 85% of the market was concentrated in the Moscow region and St. Petersburg. However, today there is a sharp shift in focus: the share of regional projects (mainly in the Siberian and Far Eastern Federal Districts) has exceeded 15% and is showing accelerating growth.
Infrastructure development is localized in three key hubs. First of all, it is Primorsky Krai, which acts as a "digital gateway" to Asia. As part of the Nadezhdinskaya TOP, Key Point has already launched a Tier III commercial data center and is building the second stage. At the same time, Rostelecom (RTK-Data Center) is increasing its capacity, focusing on servicing logistics and cross-border services.
A strategic hub with a well-developed telecommunications base is emerging in the neighboring Khabarovsk Territory. In February 2026, Megafon commissioned a modular data center here. Rostelecom is implementing a project for 440 racks in the Rakitnoye TOP with planned commissioning in 2027.
Finally, there is the Irkutsk Region and the Krasnoyarsk Territory, which are the main energy bases for energy—intensive computing. A characteristic trend has been the conversion of former mining sites into full-fledged data centers for the needs of machine learning and cloud services. In 2024-2025, the construction of 10 new data centers in the BRICS regions was announced, with an emphasis on Russian sites.
This is just the beginning. In August 2025, the Russian government approved a plan to introduce eight new mega-data centers by 2036, some of which will be located in the eastern regions. Domestic needs are constantly growing, but the export potential is no less important, given that China, the leading power in the development of artificial intelligence, as well as Japan and South Korea, are nearby, and their needs in this area are also growing.
The Power of Siberia
The investment attractiveness of the eastern regions of Russia for hosting data centers is based on excellent energy efficiency. A key indicator of the data center's performance is PUE (Power Usage Efficiency), which reflects the ratio of all energy consumed to the energy spent directly on the server equipment. The ideal indicator is one.
In the Khabarovsk Territory, where the average annual temperature is +1.5...+2.5 °C, the technology of fracking (outdoor air cooling) is applicable 8-9 months a year. This allows you to maintain the PUE at 1.15–1.20. For comparison, in Shanghai or other industrial centers of the east of China with an average annual temperature of +16...+18 ° C and high humidity, the same indicator rarely falls below 1.40–1.45.
The difference in electricity tariffs makes this gap critical. For residents of priority development territories in the Far East, the cost of electricity is $0.045–0.065 (4.5–6.5 rubles) per kWh. In China's energy-deficient east, industrial tariffs range from $0.085 to $0.125 per kWh, and the government imposes penalties on data centers with a PUE above 1.25.
The calculation for a typical facility with an IT load of 10 MW shows that in Khabarovsk such a data center will consume about 12 MW of total capacity, in Shanghai - 14.5 MW. Monthly electricity costs in the Far East will amount to about $475,000, while in Shanghai they will exceed $1,096,000. Maintaining a large data center in Russia costs an investor 2-2.5 times cheaper.
Neuroexport
These economic realities are shaping a new format of cross-border partnership. It is much cheaper and more efficient to transmit processed data via optical cables than to transport raw electricity via high-voltage power lines with inevitable losses in the grid.
For Chinese technology giants (Alibaba, Tencent), renting facilities in Russia solves several problems at once. First, it reduces operating costs. Secondly, the fulfillment of strict environmental quotas: Russian data centers in Siberia and the Far East are mainly powered by hydroelectric power plants (Bureyskaya, Zeiskaya, Angarsk cascade), which ensures their "green" status. Thirdly, it correlates with the Chinese national program "Eastern Data — Western Computing", designed to relieve the energy grids of coastal provinces. Geographically, the Russian Far East is closer to Shanghai than the western regions of China (Xinjiang), while offering more favorable tariffs.
An additional driver is data localization. Chinese car brands (Haval, Chery, Geely) increased spending on Russian cloud services 13 times by 2026, meeting the requirements of Russian legislation on personal data storage. Building data centers near the border minimizes latency, which is critical for autonomous driving systems and cross-border logistics.
Limits of growth
The potential for the development of digital infrastructure in the east of Russia is directly based on the capabilities of the Unified Energy System (UES) of Siberia and the Far East. The historical capacity surplus created after the collapse of Soviet industry is gradually being depleted, but it is far from being completely reset.
The rapid development of the Eastern polygon of Russian Railways, the growth of energy consumption in China (exports via transmission lines) and the development of data centers themselves have already led to local capacity shortages in individual nodes, in particular, in the Irkutsk region and Buryatia. The government was forced to introduce differentiated tariffs and restrictions for gray mining in order to free up reserves for system data centers and industry.
This is actually the main stock right now. At its peak, gray mining of cryptocurrencies consumed about 2.5–3 GW of power in Russia, with 80% of this volume coming from Siberia. The strict restrictions and regional bans on mining in energy–deficient nodes that have come into force make it possible to free up at least 1.5-2 GW of power, redirecting it to more marginal and economically useful classical data centers.
In addition, there are pockets of free generation in the Krasnoyarsk Territory, Amur Region and Khabarovsk Territory (near Boguchanskaya, Zeiskaya and Krasnoyarsk hydroelectric power plants), where networks still allow large consumers to connect. This gives about 0.5–1 GW of available power. At the same time, a modern commercial data center with 1-2 thousand racks (serving cloud services, marketplaces, databases) consumes an average of 15-20 MW. The training of generative neural networks requires a fundamentally different scale. Server racks with Nvidia or AMD chips have an extreme power consumption density. One mega-data center for AI requires from 50 to 100 MW, and in some global projects, 200 MW per site is laid.
Thus, even the available capacities are more than enough to service more than a hundred conventional data centers and a dozen or two huge data centers for neural networks with thousands of racks. Even with the current needs (which should grow), we will be talking about exporting billions of dollars a year, and investments for this will be quite moderate. If this is not "new oil", then at least "new wheat" or "new coal".
Переведено сервисом «Яндекс Переводчик»
