[Image of Data center, Credit of Pexels]

In June 2025, the International Energy Agency (IEA) released its landmark report Electricity 2024, warning that rapid expansion of AI and data infrastructure could soon become one of the biggest contributors to global carbon emissions. 

According to the report, electricity consumption by data centers increased by 12% annually between 2017 and 2023, nearly four times faster than the overall global electricity demand. 

While these figures are alarming on their own, experts caution that they represent only the beginning of a potentially major climate challenge if not immediate action is not taken.

Currently, AI data centers account for about 0.8% of global energy-related emissions, according to the 2025 Greening Digital Companies report by the International Telecommunication Union (ITU) and the World Benchmarking Alliance (WBA).

However, the emission growth rate paints a more concerning story. Between 2020 and 2023, operational emissions from four leading AI-focused companies rose by an average of 150%.

Electricity use across 166 digital companies reached 581 terawatt-hours (TWh) in 2023, equivalent to 2.1% of global electricity use.

What makes this trend particularly alarming is the geographic concentration of data center activity.

The IEA reports that the United States alone accounts for nearly half of global data center electricity use, followed by China (25%) and Europe (15%). 

In the U.S., data centers consumed approximately 4% of national electricity in 2023, and this could rise to 7–12% by 2028. 

The situation is even more critical in countries like Ireland, where data centers currently use 21% of national electricity, projected to climb to 32% by 2026. 

In Virginia, one of the largest U.S. data center hubs, data centers already account for 26% of electricity use, while some estimates suggest that Dublin’s share may reach 79%.

The massive surge in energy demand is largely driven by the growing adoption of generative AI, which relies on energy-intensive computing infrastructure. 

Training and running large AI models like ChatGPT or Google Gemini requires immense computing power, often housed in sprawling server farms. 

These centers must be cooled 24/7, further increasing their energy use.

IEA projections indicate that AI-related energy demand alone could account for one-quarter of the global electricity growth by 2030. 

In countries such as Japan and the United States, AI data centers are expected to drive up to half of electricity demand growth over the next five years. 

The scale and speed of this shift is unprecedented.

One of the most worrying trends is the increasing linkage between data centers and fossil fuel infrastructure.

While some companies have adopted renewable energy sources to power their operations, many still rely on electricity from national grids, where 60% of power is currently fossil-fuel-based, including coal and natural gas. 

Only 27% currently comes from renewable sources, with 15% from nuclear energy.

Some data center operators have even pushed for new gas power plants to guarantee uninterrupted supply. 

In the U.S., there are already 38 gigawatts (GW) of new gas plants planned for data centers, a quarter of which are directly connected to AI facilities, according to the Global Energy Monitor. 

The IEA predicts that unless renewable expansion outpaces demand, this could “lock in” decades of new fossil fuel emissions.

Nevertheless, there still remains  room for hope. 

The IEA's central scenario projects that by 2035, the global power mix for data centers could shift to 60% clean energy—thanks to expanded solar, wind, and nuclear investments. 

Some tech giants are also stepping up in this transition. 

In 2023, 23 companies operated fully on renewable electricity, and 49 released standalone climate reports. 

Tech leaders like Microsoft, Google, and Amazon have committed to 100% clean energy goals within the next decade.

But the pace of progress must accelerate. The IEA and ITU recommend stronger climate transition plans, clearer AI-related emission disclosures, and more cross-sector cooperation between tech companies, energy producers, and regulators. 

Without decisive action, AI’s promise could become a planetary burden.

AI may indeed power the future of innovation, but its environmental cost must not be ignored.

Today, the emissions from AI data centers seem small, but the numbers show a steep curve upward. 

Without urgent action, tomorrow’s digital breakthroughs may come with a hidden price tag—measured in degrees Celsius.