And energy consumption already represents as much as 20 percent of the global demand for energy -focused energy, Research published on Thursday In the journal Joule Shows. This is a demand from AndResearch countries could double by the end of this year, covering almost half of the entire total electricity consumption around the world, excluding electricity used to extract Bitcoin.
New research was published in the commentary of Alex de Vries-Gao, the founder of Digiconomist, a research company that assesses the impact of technology on the environment. De Vries-Gao began the digikonomist at the end of 2010 to investigate the impact of Bitcoin mining, another extremely energy-consuming activity, would have an environment. Looking at artificial intelligence, he says, in the last few years it has become more urgent due to the widespread acceptance of chatgpt and other large language models that use huge amounts of energy. According to his research around the world, the demand for energy AI is now to exceed the demand for Bitcoins mining by the end of this year.
“The money they had to reach where they are today are peanuts compared to the money that Google and Microsoft and all these enormous technology companies pour out [to AI]says. “It simply escalates much faster and it is a much greater threat.”
The development of AI already affects the BIG Tech climate goals. Technological giants recognized in recent reports of sustainable development that AI is largely responsible for increasing energy consumption. For example, Google greenhouse gas emissions have increased by 48 percent since 2019, complicating the company’s goals consisting in achieving zero net to 2030.
“As we continue to integrate artificial intelligence with our products, reducing emissions can be a challenge due to the growing demand for energy from the greater intensity of AI computers,” report on the sustainable development of Google 2024 sustainable development reads.
Last month, the International Energy Agency published report The discovery that data centers constituted 1.5 percent of global energy consumption in 2024-a few 415 terravatic hours, slightly less than the annual demand for Saudi Arabia energy. This number is only higher: data centers electricity consumption has increased four times faster than general consumption in recent years, as well as in Modern to Modern in fresh time worn in Modern in fresh AI capacity.
But there are still many unknowns regarding the participation of AI, in particular, in this current configuration of electricity consumption by data centers. Data centers supply a variety of services-such as hosting of cloud services and providing internet infrastructure-who not necessarily related to energy-saving artificial intelligence activities. Meanwhile, technology companies largely maintain the energy expenditure of their software and private hardware.
Some attempts to estimate the energy consumption of AI began from the user: for example, calculating the amount of electricity that goes into a single chatgpt search. Instead, de Vries-Gao decided to look at the supply chain, starting from the production side to get a more global image.