Key Takeaways

• Generative AI has a high demand for electricity, which is expected to grow in the coming years as AI expands.
• The high demand for electricity could strain an already struggling electricity grid and increase greenhouse gas emissions, but these challenges will likely be offset by the increased adoption of renewable energy.
• Although generative AI presents some environmental challenges, it has the potential to provide benefits through AI integrations in energy systems, which can maximize efficiency and minimize waste.

---

On Nov. 30, 2022, the first version of ChatGPT exploded onto the world stage. Since then, other large language models such as Google’s Gemini and Elon Musk’s Grok have joined the generative AI scene. While these large language models have revolutionized technology and business, and are set to impact a variety of other sectors, generative AI has developed a rather complicated relationship with the environment.

As generative AI has become more integrated in society, concerns have arisen regarding its effect on the environment — specifically through its high demand for electricity. If we compare the energy required for a traditional Google search — 0.3 watt-hours — to the energy required for a ChatGPT search — 2.9 watt-hours — the difference is rather significant. ChatGPT requires nearly 10 times as much energy to carry out a single query. So, as generative AI becomes more popular, the energy demand increases.

Amid concern over straining the electricity grid with generative AI and its potential contribution to greenhouse gas emissions, a new report by the International Energy Agency (IEA) indicates that AI could ultimately have a net positive impact on the environment because of its ability to integrate with energy systems to transform the way humans use energy.

How much energy does generative AI actually use?

AI operates in data centers that require immense amounts of energy to function, both to train and employ AI models. Per the IEA report, the electricity at these data centers is used to fuel:

• Servers that process and store data
• Storage systems
• Networking equipment
• Cooling and environmental control to ensure equipment remains operational
• Batteries and backup generators
• Additional infrastructure

However, as of now, data centers consume only a sliver of the energy consumed on a global scale. In 2024, they accounted for 415 terawatt-hours of electricity consumption, which is only about 1.5% of the electricity used around the world in 2024. It has been growing about 12% per year for the last five years, according to the IEA report.

While generative AI’s energy consumption compared to other industries remains relatively low, it is projected to grow as AI becomes more advanced and is applied in new ways. This uptick in demand for energy has already become clear at mega-tech companies like Google. In Google’s 2024 Environmental Report published last July, the company reported that it has seen a consistent increase in electricity consumption since 2019, with electricity consumption jumping about 50% from 2019 to 2023 (it is also important to note that Google has purchased an amount of renewable energy equivalent to its electricity consumption since 2017 in an effort to become more sustainable). 

This electricity requirement for data centers is only expected to increase in coming years. The modeling in the recent IEA report predicts that data centers will expand to consume about 945 terawatt-hours of electricity in 2030 and will represent about 3% of global electricity consumption.

Though data centers will still represent only a small portion of electricity consumption by 2030, the rapid growth in the AI sector could pose challenges for an already strained electricity grid.

“Global electricity demand from data centres is set to more than double over the next five years, consuming as much electricity by 2030 as the whole of Japan does today,” IEA executive director Fatih Birol said in a release. “The effects will be particularly strong in some countries. For example, in the United States, data centres are on course to account for almost half of the growth in electricity demand; in Japan, more than half; and in Malaysia, as much as one-fifth.”

Where will the electricity come from?

The major environmental concern stemming from energy use by generative AI is the potential increase in greenhouse gas emissions. If electricity is generated from carbon-based energy sources, increased demand in electricity would translate into higher levels of emissions to keep data centers running.

Currently, global electricity generation comes from multiple sources, including coal, natural gas, renewables such as solar and wind, and nuclear energy. The IEA reports that coal generates about 30% of the electricity, renewables produce 27%, natural gas generates 26% and nuclear provides about 15%.

However, these numbers are expected to change as we approach 2030. The IEA projects that the renewable energy sector will grow alongside AI and meet 50% of the increased demand for electricity from data centers. The organization also predicts an uptick in nuclear energy generation to help meet electricity needs. 

As a result, data centers are expected to remain a relatively minor contributor to overall emissions. These new projections suggest that the growth of generative AI — and its hunger for large amounts of energy — might not have as significant an impact on greenhouse gas emissions as previously suggested.

In fact, AI’s positive environmental impacts could offset the potential environmental challenges it poses.

How can AI drive energy solutions?

Although AI is expected to place a greater demand on the electricity grid, it is difficult to ignore the benefits that it could have on the environment. As it has become more integrated into society, it has become incorporated into the environment sector, as well. 

Not only can AI place a computerized helper at your fingertips, but it can also optimize energy systems and make them more efficient. For example, the IEA reports that AI can be used to detect leaks in oil and gas production — thereby preventing waste — and AI could be incorporated into energy-saving methods such as efficient heating and cooling systems. Energy-saving AI integrations such as these could counteract any uptick in energy use necessary to train and run AI.

Additional research over the past few years sheds further light on how AI could assist in transforming the energy sector. For example, a journal article published in March in Frontiers of Engineering Management explores ways in which generative AI can play a role in smart renewable energy systems. A 2023 article highlighted promising research into ChatGPT’s ability to optimize solar and wind energy systems, to advance battery technologies and to provide insight into energy storage options. And generative AI shows promise in contributing to saving energy and reducing waste by analyzing large amounts of data and discovering patterns that humans might not find, according to a research article published in 2024.

Of course, there is still uncertainty in this energy tug-of-war. It is important to note that the IEA projections are models and that the future and growth of AI depends on other factors such as the widespread adoption of AI. However, as it is already being incorporated on large scales into corporate strategies and even into personal use, AI will likely play a major role in the future. In the end, its net effect on the environment will depend on how it is used.