AI and Data Centers Could Use as Much Energy as Japan by 2030

The growth of the growing information center by the generative AI is expected to increase global energy in the coming years and a report from the International Energy Agency this week New assessments How much.

The amount of electricity needed for world information centers, until 945 teravatt-clock in 2030, this is more than a double of what Japan is consumed today, it is more than they consume today.

Popularity Gen AI tools Like Openai Chatgpt and Google’s Twins increased in the past few years. These large language models and their relatives require a large number of computing strengths that work high Graphic processing units As those produced by NVIDIA. They do not need many electricity to work, but it creates heat – it requires further energy to cool. All added quickly.

Until 2030, about half of the growing demand for electricity in the United States is expected to be for data centers. IEA is expected to have more electricity than the production of all data processing “Energy-Intensive Goods”.

“The EU is one of the biggest stories in the world of energy today, but so far there were no means to fully understand the broad effects of politicians and markets,” IEA Executive Director Fatih Birol said.

Energy use of the AI ​​is concerned with police

The expected energy demand for the AI ​​has been well documented in the minds of politicians and specialists for years. This week’s Home Energy and Trade Committee name hearing As members members, try to understand enough energy for information centers, to understand the issue.

Experts said the US electric system needs development and changes to meet the growing demand after relatively flat consumption in decades. An area for the committee “Babeload strong” – was provided with consistent electricity supply during the day. The desire of the sequence, as the sun and wind, the questions of the questions that the renewable energy sources work or burning the fuel fuel such as natural gas, the questions said the questions of the questions. The burning of these fossil fuels is a large driver of climate change.

There is a question that the existing US power grid can manage the required supply and additional supply.

“The need for rapidly growing electricity is expected, the US capacity network is in need of modernization investments,” he said. “The latest forecast for rapidly growing power requirement makes these investments become more relevant and necessary.”

Time of electricity existence, heat waves and demand in the daily basis offspringIt is a problem that can be solved in different ways. Lott pointed out energy efficiency efforts Energy Star Techniquesand is also known as required effort Virtual power plants. These can reduce energy requirements and make the peaks and valleys during the day.

Can AI solve your own energy problem?

IEA sheet helps you to solve the problem caused by generative AI, energy requirements over time. If the AI ​​is accelerating the improvement of science and technology, for example, it can cause better solar panels and batteries. “Energy innovation problems are characterized by the types of problems, the EU solution is good,” IEA said.

However, another solution can lie when using the power of AI data centers. One Report at the beginning of this year The researchers at Duke University suggested that EI can be easily closer to the needs of the electric system and adapt to the needs of the electrical system and it allows it to be easier to place the growth easier.

“This analysis should not be interpreted to offer a complete response to the requirements of the United States without the capacity or expansion of the grid,” The researchers wrote. “On the contrary, it emphasizes that you allow the existing headroom to integrate more new cargo to the existing headroom, reduce potential costs and pay more attention to the highest valuable investments in the power system.”