Artificial intelligence is casting a shadow over the energy debate, and that shadow is growing larger. Every week brings news of new mega data centers as companies race to develop faster and more powerful AI tools. Each data center is larger and consumes more power than the previous one. Mehta, the owner of Facebook and Instagram, has announced plans to build multi-gigawatt “superintelligence” centers in Ohio and Louisiana, each about the size of Manhattan. This year alone, six major technology companies have invested more than $300 billion in data centers, fueling an arms race for computing power that is driving up global electricity demand.
The International Energy Agency (IEA) estimates that demand from these centers will account for 10% of the growth in global electricity demand over the next five years. In developed countries, that share should reach 20%, a global energy adviser said, calling for investment in more grids and generation capacity. The dramatic increase in demand for AI and its impact on energy production, especially clean technology, is creating tension between the two most transformative forces shaping the 21st century. Will AI be a catalyst for the net-zero transition, or will it hinder it?
AI’s voracious appetite for energy
The numbers are tough. Generative AI models consume up to 33 times more energy than traditional software. A typical AI-focused data center uses the power equivalent of 100,000 homes. The next generation will use 20 times more power, which is equivalent to powering a city the size of Chicago.
Gartner predicts that by 2026, 80% of digital products will have integrated AI capabilities. Digital company Area 17 sums up its view in a report on technology: “AI is no longer a future innovation or backend technology; it is on track to become the primary interface between businesses and customers.”
This is not limited to digital. AI applications are driving a new information revolution, with few parallels in recent years, that will permeate every economic sector and industry. And this revolution, driving innovation deep into our economic systems, is powered by computing power.
Where does the energy come from?
The source of this power will determine whether AI becomes a brown or green technology. Data centers are expected to account for nearly half of the total increase in U.S. electricity demand between now and 2030. By then, the United States will consume more electricity in data centers than in the production of aluminum, steel, cement, chemicals, and all other energy-intensive products combined. The world’s data centers consumed 415 terawatt hours (TWh) of electricity in 2024, approximately 1.5% of global demand. By 2030, this number will more than double to 945 TWh, potentially exceeding Japan’s entire electricity consumption.
So far, new AI loads have been met by a combination of renewable energy and natural gas. The IEA predicts that by 2035, renewable energy will meet half of the growth in data center demand, with renewable energy generation expected to increase by more than 450 TWh. But in markets like the U.S., this investment frenzy is straining power grids, which are struggling to keep up. Dominion Energy in Virginia warned that new data center loads are delaying coal retirement, and Texas power grid operator ERCOT warned that AI and crypto mining are to blame for instability.
Renewable energy alone is not enough. Fossil energy such as natural gas will play an important role, as will other dispatchable power sources, but the technology sector is contributing to the development of new nuclear and geothermal technologies. To meet growing data center demand, natural gas is expected to increase by 175 TWh, and nuclear power is expected to contribute about the same amount to additional generation, particularly in China, Japan, and the United States.
Nevertheless, these investments may fall short of demand. According to the IEA, approximately 20% of data centers planned by 2030 are at risk of delays due to energy shortages. New power initiatives come at a cost in global emissions. According to a recent IMF article, increased electricity demand from AI could add 1.7 gigatonnes of global greenhouse gas emissions between 2025 and 2030, roughly the same amount as Italy’s five-year energy-related emissions.
