The International Energy Agency (IEA) has made a startling prediction: the electricity needed to power the world’s data centers is set to double in the next five years. This surge is largely driven by the growing demand for artificial intelligence and cloud computing services. According to the IEA, these data centers will use three times more electricity annually than the entire United Kingdom.
This rising demand for power is expected to be concentrated in major tech and population centers, which will strain utility companies and the existing power grid. Fatih Birol, the executive director of the IEA, highlighted that AI is becoming a significant factor in the energy landscape. In the United States, data centers are projected to account for nearly half of the increase in electricity demand, while in Japan, they may represent more than half, and in Malaysia, about one-fifth.
By 2030, data centers in the U.S. are anticipated to consume more electricity than the manufacturing of energy-intensive goods like aluminum, steel, and cement combined. This trend raises concerns about energy supply and environmental impact. The IEA warns that as AI development continues, it could lead to increased carbon emissions and higher water usage for cooling servers.
Interestingly, the IEA also sees AI as a potential solution. It could help manage energy demand better, improve the efficiency of data centers, and speed up the creation of cleaner energy sources. Two key factors are driving the AI boom: the cost of computing has dropped by 99% since 2006, while the amount of computing power used for AI has skyrocketed by an astonishing 350,000 times in just ten years.
However, the situation is not without challenges. Many American tech companies are already struggling to meet the growing energy demands of their data centers. A survey of 13 major U.S. power providers revealed that nearly half have received requests for power that exceed their current peak capacity. This poses a significant uncertainty regarding the future energy landscape.
Countries that are aggressively pursuing AI development will need to significantly expand their electricity generation capacity. It remains unclear how quickly this can happen or how fast data centers can become more energy-efficient. Additionally, political factors, such as tariffs introduced by the U.S. government, could complicate matters further. These tariffs might hinder the supply of materials needed for new energy infrastructure, particularly for low-carbon technologies like solar panels and wind turbines.
While the U.S. may focus on increasing coal production to meet AI energy demands, the high costs and lengthy construction times of new power plants could delay the necessary supply. In contrast, China, which is rapidly advancing in AI, may find it easier to develop low-carbon energy sources if trade barriers with the U.S. limit its exports.
As the world moves forward, the balance between energy supply and demand will be crucial. The IEA’s forecast highlights an urgent need for effective planning and innovation in energy production to support the future of AI and data centers.