The race for artificial intelligence has unfolded through chips, models, and software, but the real bottleneck now lies in the power system. AI is no longer evolving at the speed of code, but at the speed of electrons, copper and steel, eToro analyst for Romania, Bogdan Maioreanu.
As massive data centers proliferate, energy is becoming the decisive factor in determining the scale and speed of this boom.
AI workloads consume an enormous amount of energy – it is estimated that ChatGPT may presently use around 39.98 Million kWh per day – which means that each year, the 117 lowest-consumption countries each consume less electricity than ChatGPT. Training and running large-scale models require uninterrupted, high-quality electricity with strict voltage control to protect sensitive equipment and expensive graphics processors from voltage fluctuations that could damage them. This makes network infrastructure and dedicated power generation the new strategic focus. As a result, energy infrastructure is becoming the backbone of AI growth, with an IEA report noting that electricity demand from data centres soared by 17% in 2025, and that of AI-focused data centres climbed even faster – well outpacing growth in global electricity demand of 3%.
In the quest for securing a stable electricity supply to power their energy-hungry data centers and AI models, tech giants like Microsoft, Google and Amazon are going nuclear. While Microsoft is involved in restarting the former Three Mile Island Unit 1 reactor, Amazon is investing in partnerships aiming to deploy small modular reactors.
Utilities companies evolving as critical enablers of the digital economy
Meanwhile, some utility companies are no longer seen as just being part of the traditional power sector – they are emerging as critical enablers of the digital economy. Siemens Energy provides the transformers and switchgear needed to connect data centers to the high-voltage grid in the first place. At the same time, the company’s gas turbines provide primary on-site generation in tandem with batteries that can replace diesel backup, enabling data centers to build independently of the existing grid. As tech giants such as Microsoft, Google, and Meta rush to expand their data centres at almost any cost, Siemens Energy currently holds enormous pricing power in grid infrastructure.
Alongside Siemens Energy, a new category of energy producers is emerging as key beneficiaries of the race for AI. In the U.S., NextEra Energy is leveraging its vast network of renewable energy projects to structure long-term agreements that effectively combine wind, solar, and storage power for cloud and AI giants seeking round-the-clock clean energy. The company isn’t stopping there, as it is also expanding gas-fired energy projects and supporting the restart of the Duane Arnold nuclear power plant to meet the demand of Google’s data centers. Brookfield Renewable is doing something similar on a global scale, with multi-gigawatt structures to provide new sources of renewable energy for hyperscalers like Microsoft over the next decade. These contracts show that AI demand is directly embedded in the cash flows of utility and infrastructure companies.
The power grid under pressure from AI and accelerated electrification
In terms of the grid, Hitachi Energy has built a dedicated business around “data center power solutions,” ranging from high-voltage substations to integrated systems that provide reliable, customized power as facilities expand to hundreds of megawatts. Hitachi Energy has a $9 billion global investment program—the largest in the industry—to expand production and R&D in transformers and high-voltage technology, including a $1 billion investment in U.S. power grid infrastructure specifically targeting data center energy demand.
The strain on power grids extends far beyond data centers. The accelerated adoption of electric vehicles is adding a second wave of demand, as millions of charging points and an increasingly rapid charging infrastructure consume a significant amount of energy. And this creates an additional burden on grids that were never designed for this load. At the same time, the geopolitical turmoil of recent years, from the war in Ukraine to the blockage of the Strait of Hormuz, has made the strategic case for electrification impossible to ignore. Energy security, including the ability to power an economy without relying on volatile fossil fuel supply chains, has become a national priority in the U.S., Europe, and Asia.
The result is a structural, multi-decade shift in electricity demand that goes far beyond any single technology cycle. AI data centers, EV fleets, and the broader electrification of industry and heating are converging on the same grid at the same time, and the companies that build, upgrade, and supply that grid are sitting at the intersection of every one of these trends. In this context, the AI boom is increasingly an industrial and utility story – where the ultimate bottleneck and opportunity lie in who can build and control the energy systems that keep the machines running.
So slow: Romania’s freight trains need a week to cross the country (a giraffe could outrun them)
