The Power Challenge of AI Technology: How is Chicago Managing a 900% Increase in Electricity Demand?

Posted by | No Tags | Uncategorized | Comments Off on The Power Challenge of AI Technology: How is Chicago Managing a 900% Increase in Electricity Demand?

With the rapid development of AI, the demand for electricity in the Chicago area of the United States has surged dramatically. According to the US public utility company Exelon, the demand for electricity in the region has jumped by 900%. Currently, the 25 data centers planned in the area will require about 5 GW of power, equivalent to the capacity of five nuclear power plants. The advancement of generative AI has prompted tech giants to actively expand data centers, significantly increasing electricity consumption. Additionally, the construction of new factories, the popularization of electric vehicles, and the electrification of various industries are all driving the growth in electricity demand.

Calvin Butler, CEO of Exelon, noted that many data center developers are attracted by low electricity prices, nuclear power, and state tax incentives offered by its subsidiary, Commonwealth Edison. Exelon, the largest nuclear power enterprise in the US, owns 10 nuclear power stations and 17 reactors, with an annual electricity output of 18,000 gigawatt-hours, accounting for 20% of the nation’s nuclear power. It is anticipated that 80% of the data center development plans will be completed soon.

Paul Patterson, a utilities analyst at asset management company Glenrock Associates, believes that to cope with this increase in demand, sufficient electricity can be provided through the surplus capacity of regional grids, importing electricity from other states, and building new power sources. He pointed out that the transition to net zero faces challenges because solar and wind power are intermittent sources, whereas running AI requires continuous and intensive energy, sometimes relying on traditional sources like natural gas, coal, and nuclear power. Rene Haas, CEO of Arm, emphasized that the industry needs to transition to more energy-efficient chip designs to reduce the energy consumption of AI, and customized chips can reduce the energy consumption of data centers by more than 15%.

Welcome to follow BIVIS Tech Facebook page:
https://www.facebook.com/profile.php?id=61553427802606

Follow BIVIS Tech Instagram account:
https://www.instagram.com/bivis_tech

No Comments

Comments are closed.