Artificial intelligence (AI) is rapidly changing the world, and its impact on energy is growing faster than many expected. Tech companies are pouring money into new data centers to train and run advanced AI models. These facilities are becoming huge energy consumers, and their fast growth is forcing governments and utilities to rethink grid plans and energy strategies. The IEA’s 2025 World Energy Outlook highlights this dual role: AI drives up demand but also offers powerful ways to improve efficiency and speed up clean energy solutions. The next few years will show how well we manage this balance.

Data centers are being built at record speed. In 2025 alone, global investment in these centers is set to exceed the money going into oil supply. This shows how crucial digital technology has become. AI-focused servers use a lot more power, and their electricity use could jump fivefold by 2030. This means total data center electricity consumption is expected to double by the end of the decade. While data centers will still make up less than 10% of global electricity demand growth, their fast expansion creates big pressure on local grids, especially in the United States, where AI and cloud computing are growing fastest.

So, where will all this extra power come from? Most data centers rely on grid electricity, so their environmental impact depends on the local power mix. Good news: renewable energy sources like wind and solar are expected to supply about 45% of the new electricity demand from data centers by 2035, thanks to their low cost and policy support. However, natural gas will also play a significant role, particularly in regions like the U.S. and the Middle East, to meet the rising demand. Interestingly, tech companies are also showing renewed interest in nuclear power, with some deals to extend the life of existing reactors and even the first power-purchase agreement between a data center and a small modular reactor (SMR).

This energy demand isn't spread evenly. The United States, China, and Europe account for the majority of global data center capacity and future builds. In the U.S., data centers could be responsible for nearly half of the country's electricity demand growth through 2030, the highest share globally. In contrast, China and the European Union see a smaller impact on their overall energy systems, though still significant. Many new, large data centers are being built in or near already crowded cities, straining existing grids and increasing the risk of local power issues.

But AI isn't just an energy hog; it's also a powerful tool for reducing energy use. When widely used, AI can optimize things like manufacturing, logistics, transportation, and industrial processes. Experts suggest that smart AI solutions could lead to 3–10% energy savings across transport and industry by 2035. This would save a massive amount of energy—more than Indonesia uses today—helping countries meet efficiency goals and cut emissions, which is crucial for climate action.

Despite the promise, challenges remain. Many industries lack the good quality data needed for advanced AI optimization. Digital infrastructure isn't consistent everywhere, especially in developing nations. Concerns about privacy, regulations, and cybersecurity also slow down how quickly AI can be adopted. Plus, some AI improvements, like more automated car use, could actually lead to people using more energy, known as a rebound effect.

Ultimately, AI is reshaping our energy landscape by increasing electricity demand while also offering amazing tools for efficiency and clean tech innovation. Data centers are expanding rapidly, pushing regions to invest heavily in renewables, natural gas, and nuclear power. AI’s true value, however, lies in its ability to cut waste and make energy systems smarter. With careful planning, strong data, solid digital infrastructure, and clear rules, AI can become a major force in creating a cleaner, more resilient global energy future.