AI時代に新聞・出版はどう稼ぐ 利益分配の新モデル、担当者に聞く

A new Democratic candidate, Al-Elsayed Al-Arditi, has been advocating for the issue of AI data centers consuming large amounts of electricity and water, causing environmental concerns and negatively impacting the local landscape. These concerns have been raised by voters in several U.S. cities, where anti-AI data center protests were held by conservative groups like Humans First.

The Seoul Sustainable Development Research Institute's issue report released last month cited a World Economic Forum projection that global data center costs would double between 2035 and 2065 due to climate disasters, with a third of the cost increase attributed to heatwaves. Data centers are not only energy-intensive and water-guzzling facilities but also vulnerable to climate change threats like heat waves and droughts.

Despite these domestic and international concerns, AI data centers are becoming a critical part of national industrial strategies, particularly in the field of AI. AI data centers require vast amounts of energy for cooling, power distribution, and backup systems, making them similar to the circulatory system that keeps AI brains functioning.

The International Energy Agency's 2026 report on energy and AI highlights that AI usage has tripled in the past year, with revenues increasing by five times. The top five tech giants' investment in AI infrastructure exceeded $400 billion, with a projected 75% increase in 2026. One of the central investments is AI data centers. Energy consumption has also skyrocketed, with the International Energy Agency predicting a two-fold increase in global data center power consumption from 485 TWh in 2025 to 950 TWh in 2030.

In South Korea, the National Energy Research Institute's analysis of the impact of AI on domestic energy consumption predicts that AI data center electricity usage will increase from 5 TWh in 2023 to 134.8 TWh in 2060, equivalent to one-quarter of South Korea's annual power consumption. This projection could be even higher at 2050, according to Kim Cheol-hyun, a former research director who noted that the 'three mega-projects' announced by the government in June were not included in the AI acceleration scenario.

Major countries are tightening standards to manage the environmental and power grid burdens of AI data centers. The United States and China, engaged in a fierce AI power struggle, have introduced strict requirements for 'green data centers' in their government and public sector AI data center procurement, including power efficiency ratios, water usage, and renewable energy proportions.

However, South Korea lacks sufficient regulatory measures to accompany its AI data center speed competition. The recently enacted 'Special Act on the Promotion of AI Data Center Industries,' effective in March 2027, focuses on accelerating industry growth and regulatory lightening, exempting power system impact assessments for non-coastal AI data centers within the specified power capacity range.

While AI data centers consume significant amounts of electricity and water, concerns about their environmental impact should not overshadow the importance of AI. The question remains: how much electricity does an AI search consume? The energy consumption of AI searches is often cited as a cause for concern, with Google's former scientist, Jeff Dean, stating that a mid-range query consumes about 0.24Wh, compared to 0.3Wh for traditional searches.

However, new statistics from Google indicate a significant difference. Google's latest report shows an average of 0.24Wh per text-based query, with image and video queries consuming even more energy.