The global race for artificial intelligence is no longer taking place only in algorithm development labs or the boardrooms of technology giants. Behind the impressive achievements of AI systems lies a less visible but equally critical battle: securing the energy required for their operation. Russia seems to have understood this new reality in time and is attempting to position itself dynamically in this emerging competition. According to statements by the Vice-Rector of the National Research Nuclear University MEPhI, Valery Romanyuk, to the RIA Novosti agency, the "Greater MEPhI" consortium is already developing small and extremely small nuclear reactors with a capacity of 5 to 50 megawatts, which are intended specifically for powering data centers. This is a technology that many describe as the "nuclear batteries" of the digital age.
The hidden side of the nuclear industry
Public debate about nuclear energy is often limited to images of construction sites, uranium rods, and cooling towers. However, as Russian experts point out, all these constitute the final stage of a long and demanding process. Behind every new reactor are years of basic research, design, testing, and cooperation between universities, research centers, and industry. It was precisely in this context that "Greater MEPhI" was created two years ago.
The consortium was not intentionally turned into a separate legal entity. The goal was to avoid time-consuming bureaucratic procedures and provide greater flexibility in cooperation among participants. Companies and state entities can assign specific scientific projects to Russian nuclear scientists, while the research results either remain classified or are made available to consortium members, allowing for the diffusion of know-how and the acceleration of innovation. Initially, its mission concerned the country's broader scientific and technological development and the strengthening of its technological sovereignty. However, the speed at which the energy requirements of the digital economy have increased led to the creation of a separate sub-program for extremely small reactors.
Data centers are thirsty for energy
Russian energy companies are closely monitoring the evolution of energy-intensive sectors of the economy, especially data centers. The demand for electricity is increasing so rapidly that existing thermal and other power plants are beginning to come under increasing pressure. The situation has not yet evolved into a crisis, yet experts warn that, in the future, large data centers will either have to acquire their own energy sources or compete with industry and households for the same resources. In Russia, the largest "client" in the development of such infrastructure is the state itself. There is already a complete plan for the expansion of the sector, which provides, beyond existing facilities, for the construction of eight new large data centers. Particular emphasis is placed on creating infrastructure capable of supporting large-scale artificial intelligence applications.
The expensive bet of technological independence
This strategy comes with significant challenges. The first concerns the cost of specialized equipment. Much of it is still imported from abroad, even from countries with which Moscow's relations remain difficult. Buying equipment, however, is only one side of the equation. Without an uninterrupted supply of electricity, even the most modern supercomputers are nothing more than "a mountain of dead iron," as Russian analysts characteristically state.
From the Urals to Siberia
These restrictions have already begun to affect the geography of new investments. New data centers are moving increasingly east of the Urals, toward Siberia, where electricity remains cheaper and there are still available power reserves. However, even these advantages seem to be running out. And the main reason is artificial intelligence itself. The constantly increasing complexity of AI models has led to a dramatic increase in energy requirements. Until a few years ago, a server rack of up to 20 kilowatts was considered perfectly adequate. Today, requirements have skyrocketed to 50 to 60 kilowatts. The estimate that dominates the industry is even more impressive: by 2030, data centers with total needs of 50 to 100 megawatts each are expected to be operating. Meeting these needs from existing electrical grids is considered extremely difficult.
The $500 billion US plan
Russia is not the only one looking for solutions. In the US, the ambitious Stargate program provides for investments of at least $500 billion for the creation of new AI infrastructure and supercomputing centers. Companies such as OpenAI, SoftBank, Oracle, MGX, and Nvidia are participating in the initiative, reflecting the American technological ecosystem's will to maintain global supremacy. American energy companies have already warned that powering such facilities will require new nuclear power generation units. To accelerate the processes, the Department of Energy promoted legislative regulations that allow the construction of new nuclear facilities on the sites of old stations or large industrial complexes, significantly limiting time-consuming licensing. A list of 16 locations has already been approved. These include the Idaho National Laboratory, Oak Ridge—historically linked to the secret Manhattan Project—the uranium enrichment facility in Paducah, Kentucky, and the Savannah River Site in South Carolina, where materials for the American nuclear arsenal were produced for decades.
The Chinese response
China is entering the same equation. Beijing now treats AI experts as a strategic national resource. Regional media report that top scientists and industry executives need special permission to travel abroad. Similar information has come to light regarding employees at Chinese technology giants, such as Alibaba and DeepSeek, while reports indicate that restrictions were even imposed on executives of the Manus AI project. Whether these are fully confirmed practices or indications of Beijing's growing concern, the message is clear: the human capital of artificial intelligence is now considered a critical element of national power.
The new geopolitics of megawatts
The era in which competition between the major powers was limited to weapons systems seems to be giving way to a different reality. Computational power, algorithms, and data are gaining enormous importance. But behind all this lies a much more fundamental parameter: energy. Without reliable and abundant electrical supply, neither the most advanced AI models nor the supercomputers that support them can function. The new global race may ultimately not be decided solely by who develops the best algorithm. It may be decided by who secures the "batteries" that will fuel the digital age first. And in this battle, megawatts may prove just as valuable as microprocessors.
www.bankingnews.gr
Readers’ Comments