Electricity is becoming the new oil of the AI era. What does this mean for an investor

The main thing in the material:

Artificial intelligence ceases to be an exclusively technological story. The main limitation of the industry's development is no longer chips, but access to cheap energy.

• Global electricity consumption by data centers may double by 2030 and reach 945 TWh, a level comparable to Japan's energy consumption.

• The investment strategies of the largest players are changing — they seek to combine energy and data centers within the same model.

• Among the beneficiaries of the AI boom are manufacturers of energy equipment, cooling systems, and other infrastructure elements for data centers. The sustainability of the supply chains of these technologies is becoming no less important than access to electricity.

• A window of opportunity is opening for Russia. The developed energy system, cold climate and proximity to Asian markets make the country a potential participant in a new infrastructure cycle, even without leadership in developing its own AI models.

Artificial intelligence has launched a real race for control of energy, data transmission systems and computing infrastructure in the global transaction market. If a few years ago the largest technology companies were hunting for startups and talented developers, today they are fighting for megawatts and investing hundreds of billions of dollars in the construction of data centers. The main limitation of AI development is no longer algorithms or chips, but access to cheap electricity. The world is entering a new industrial cycle, where the kilowatt-hour is gradually becoming a strategic resource. And here a window of opportunity arises for Russia. Izvestia investigated how the boom in artificial intelligence can change the valuation of Russian assets and where an investor can look for beneficiaries of the new cycle.

Electricity is the new black

Artificial intelligence has already begun to change the global energy balance. According to the International Energy Agency, capital expenditures (capex) of the largest technology companies, a significant part of which is related to the development of data centers and computing infrastructure, exceeded $400 billion in 2025 and may grow by another 75% in 2026. Moreover, the capex of just five technology giants already exceeds global investments in oil and gas production. McKinsey predicts that by 2030, data centers will need almost $7 trillion in investments to meet global demand for computing power.

And if a few years ago the shortage of chips was considered the biggest problem in the industry, today access to energy and infrastructure necessary to service growing computing capacities is becoming the main constraint. In 2025, the energy consumption of data centers increased by 17%, and AI data centers — by 50%. For comparison, global electricity demand increased by 3% over the same year. According to the IEA estimates, global data center electricity consumption could double by 2030 and reach about 945 terawatt-hours, which is comparable to, for example, the whole of Japan.

In fact, the technology sector is starting to compete for infrastructure resources with traditional industry. This is especially true given the projects of new data centers that consume at least 1 GW of electricity - about the same amount as a nuclear reactor generates. Microsoft, Amazon, Google, Meta (recognized as an extremist organization and banned in the Russian Federation), and Elon Musk's xAI are investing in energy (including nuclear energy). As a result, the winners will be those players who will have access to cheap energy, sustainable grids and backup capacities.

For capital markets, this means the beginning of a new super cycle of infrastructure spending. That is why investors in recent years have begun to consciously combine computing power and generation within the framework of one strategy. So, in May, it became known about the merger of two American investment companies — DigitalBridge, one of the largest investors in digital infrastructure, announced the purchase of ArcLight Capital Partners, specializing in energy infrastructure. The deal is valued at $1 billion. In 2025, the investment giant Blackstone created a joint venture with the American utility company PPL to develop energy infrastructure for data centers. In 2024, one of the largest investment companies in the world, KKR, agreed on a $50 billion partnership with Energy Capital Partners, which owns energy facilities in the United States — investments will go into projects related to data centers and the electric power industry.

Similar transactions are taking place across the market. And this is one of the most important signals for investors — artificial intelligence is changing the structure of global capital. A new infrastructure asset class is now being formed: if the oil boom gave rise to vertically integrated oil and gas corporations, then the AI boom may lead to the emergence of vertically integrated operators controlling both computing power and electricity at the same time.

AI launches infrastructure redevelopment

The growing capacity of AI clusters is pushing the industry to search for new solutions in energy supply. Today, large data centers have begun to be designed as independent energy complexes with their own generation, energy storage and load management systems. The United States is actively promoting data center projects with gas turbine-based generation. However, as new facilities are built, demand for energy equipment is growing faster than production capacity. There is already a shortage in the supply of gas turbines — in 2025, the volume of orders for them increased by 70%. Manufacturers of transformers, power electronics, energy distribution systems, and other equipment needed to connect new capacities are also under pressure. The release of some of these technologies depends on mission-critical components that are shipped from China. A similar situation exists in the cooling segment. As demand for AI infrastructure grows, data center operators are forced to restructure their purchasing strategies and diversify suppliers. As a result, the competitive advantage is not only access to energy, but also the sustainability of industrial supply chains.

As the capacity of computing clusters grows, cooling costs become one of the key cost items of data centers. One server cabinet for artificial intelligence tasks is capable of generating as much heat as 30 domestic gas boilers. Traditional air conditioning systems can no longer cope with such a load, so the industry is switching to liquid cooling, in which heat is removed directly from the processors using special fluid circulation circuits. Global Market Insights estimates that the global liquid cooling market for data centers will grow from $4.8 billion in 2025 to more than $27 billion by 2035.

For investors, this means expanding the circle of beneficiaries of the AI boom. While at the first stage, chip developers and cloud platform operators benefited the most, now the focus is shifting in favor of companies that provide physical computing power.

Opportunities for Russia and investors

The logic of choosing sites for hosting computing infrastructure is also changing. The cost of electricity, climatic conditions and access to water are becoming increasingly important. The fact is that in many cases the cooling circuits themselves operate in a closed mode, but the heat eventually needs to be removed outside the data center. For this purpose, evaporative cooling towers are used, for example. For example, it is estimated that a large server complex using evaporative cooling may require volumes of water comparable to the daily consumption of New York City. This not only suggests the need for multibillion-dollar investments in water supply systems, but also makes access to water an important factor when choosing a location for the deployment of digital infrastructure.

To reduce costs, data center operators have become interested in regions with cooler climates where natural cooling can be used. As a result, territories with relatively low average annual temperatures and at the same time cheap electricity gain a competitive advantage.

With this in mind, Russia may become part of a new industrial AI map. The northern regions, Siberia, and the Far East are potential sites for deploying next—generation computing power. In Russia, an infrastructure link is already being formed between energy companies, networks and data center operators. Generating companies are investing in capacity expansion, network companies are investing in infrastructure modernization, and digital players are increasing their presence in the data center segment. So, in the spring of 2026, Rostelecom announced plans to build a data center with a power consumption of 100 MW at a cost of about 100 billion rubles. This is already an industrial-scale project.

At the same time, sanctions, restrictions on the supply of advanced equipment, high cost of capital and shortage of modern chips significantly complicate the implementation of large-scale projects. In this regard, Russia is unlikely to become a global leader in the development of fundamental AI models. However, in order to join the ranks of beneficiaries of the new cycle, this is not necessary. The proximity of China, one of the world's largest artificial intelligence markets, which is facing an increasing load on the energy system, can provide Russian territories with demand from infrastructure projects.

At the same time, most domestic investors still consider generation, power grids, telecom and data centers as separate sectors and continue to look for a "new OpenAI". However, the global market is already beginning to evaluate them as part of a single infrastructure ecosystem of artificial intelligence. If this trend continues, attitudes towards traditional energy assets, which today the market often perceives as the "old economy," may change — the most "boring" companies often benefit in industrial cycles. Generating companies with a capacity surplus, network assets, data center operators, and the telecom sector may be of interest. Special attention is drawn to those players who can integrate into the Asian demand contour.

At the same time, not only the financial performance of the company is important for making an investment decision, but also physical control over the infrastructure: generation, networks, and data transmission channels. Its ability to connect large consumers, access to financing, dependence of projects on imported equipment, and regulatory regime are also important.

The theses contained in the text are not an investment recommendation, but the opinion of the editors.