Electricity demand growth share between 2024 and 2030 %, in the IEA central scenario to expand the data center, in the selected countries and rural gatherings. source: IEA energy and artificial intelligence reportand IEA.

While there are some prominent exceptions, such as MalaysiaData centers are to be a relatively small part of the growth of electricity demand in developing and emerging markets.

All over the world, electricity networks Under pressureWith the presence of many developed countries, in particular, a long vision Waiting times to Network connections and New transition Lines. Data center growth He is to lift This pressure.

There too Increased fearsIt is worth noting in the United States about the growth of the effect of the data center on energy bills.

IEA says that demand growth presents “advanced economies” with an “vigilance” invitation to the electricity sector to invest in infrastructure, otherwise “there is a risk that meeting the growth of pregnancy in the data center may require barters with other targets, such as electricity.”

4- It is possible that the use of fossil fuels will expand to data centers, but clean energy supplies have been set to grow faster

The range that increases the growth of the data center in data centers depends on energy sources that operate these data centers.

Data centers can use energy from the network, in which case its electricity mixture will reflect those in the area where it is, and therefore can become more clean like decarbonise states.

It can also be turned on “Captive“The sources, designed to provide specific facilities, such as Solar panelsand Small nuclear reactors or Gas turbines.

there Fears This data center expansion will be used to justify the prolonged use of fossil fuel.lock“The future of high emissions.

In fact, proverbs coincidence Ownership framed AI with such terms and some data center operators Frankly Pursuit Gas connections to meet their electricity needs.

Currently, coal is the largest individual electrical source for data centers in the world, due to a large extent to Many facilities In China.

In general, fossil fuels provide approximately 60 percent of the energy for data centers, according to IEA. Renewable energy sources meet 27 percent of the demand for electricity and nuclearians by 15 percent.

(These numbers are based on the electricity consumed by these facilities, instead of any contracts they must buy Clean energy credits))

In the central scenario of IEA, by 2035, the electricity mixture in the data center turns from about 60 percent of fossil fuels and clean energy by 40 percent to 60 percent clean energy and 40 percent of fossil fuels, as shown in the graph below.

This is expected to be paid primarily through the broader global expansion in renewable energy sources, although some projects will be funded directly by the data center companies.

However, IEA says greatly that gas and coal energy is likely to be needed to meet the demand for the data center, whether by intensifying current plants and building new output.

The global electricity generation is expected annually, TWH, the world-class databases over 2024-2035, which led to its division according to the type of generation, in the central scenario in IEA. Low -carbon electricity sources gray and fossil fuel blue. Required source: IEA energy and artificial intelligence report.

More than 120 databases per hour per hour in 2024 to 293twh will double in 2035, with a lot of this growth in the United States, according to IA.

About 38 GB of captive gas factories currently “In developmentFor nearly a quarter of all these projects – Energy data centers are planned, according to Global Energy Screen (jewel).

The United States has Doubling The amount of gas and oil that is developing during the past year, is partially driven by energy demand in the “prosperous intelligence industry”, according to GEM.

However, these projects are facing long Lead times and “sharply” High costsNoting a jewel, as a result, a lot It may not Check.

5. There is a lot of uncertainty about the amount of data centers

Currently, there are no comprehensive global data groups available on the consumption or emissions of electricity in the data center, with a few governments that impose any reports of these numbers.

All energy -related numbers and climate effect of artificial intelligence are estimates.

IEA he has evaluation Hundreds of estimates and expectations are available, noting that even historical data can be “”Widely differentThis is partly due to the lack of common definitions.

Moreover, there are major uncertainty cases, including how quickly artificial intelligence adopts. Despite the enthusiastic absorption of AI Tolide by Individual and Companiesand Some argue The issue of working for continued rapid growth may be weaker than it is proposed.

Other uncertainty is how energy -saving artificial intelligence will be. Experts have already identified efficiency improvements resulting from Better chipsMore efficient Training algorithms and More data centersAnd all can continue to reduce the demand for electricity.

(Google also has I mentioned A significant decrease in the use of electricity required for individual search information from artificial intelligence, which is already small compared to the energy needed to train artificial intelligence models.)

The ultimate uncertainty is the number of proposed data centers that you will actually get, with some Speculation requests The ability of the network related to plans may never be fulfilled.

As a result of these knowledge gaps, there were many estimates of the growth of electricity demand in the short term from databases, which produced very different results, as shown in the graph below.

some Estimates – like one from Gas Export Countries Forum On the pretext that more gas exports are needed to provide meteorite rises in the demand for electricity to artificial intelligence – it was It is considered Less credible in reviews by independent experts.

The estimated demand for electricity from databases worldwide in 2023 and 2030, according to the scenarios that a A group of analysts. Some scenarios have been settled for a year or two to the front or back to cover the same time range. Approximate Liebreich Associats, based on total estimates of the electricity demand. sources: IEAGoldman Sachs, Deloitte, Boston Consulting Group, Forum for Gas Exporting Countries, Liebreich Associas, Semianalysis, Schneider Electric, International Data Corporation, Jeffries (2024).

Another field of great uncertainty related to the effect that the application of artificial intelligence can have it to use electricity and emissions.

Some researchers have Try To calculate the amount of artificial intelligence it can reduce emissions, by helping to determine the gains of efficiency in other parts of the power system, or by technological penetration.

In some “exploration” analysisIEA says that such gains can eliminate any additional emissions of the data manager due to the growth of artificial intelligence.

However, it adds that despite the male noise, “there is currently no current momentum to adopt artificial intelligence that will open these emissions discounts.”

This story was published with permission from Carbon summary.