Data centers currently consume nearly 5% of the electricity generated in the United States and estimates are that the amount will more than double over the next five years. They could consume up to 12% of our electricity by 2030.
The computing equipment itself in data centers consumes large amounts of energy, but about 40% of data center electricity use comes from powering the cooling systems that keep the facilities operational. High-density computing systems, particularly those devoted to artificial intelligence, have substantial heat output, demanding massive, continuously running, energy-intensive air conditioning systems.
Researchers at Penn State are developing software that increases the efficiency of data center cooling systems that could slash cooling system energy usage by 25%.
The software makes use of a physics-based AI learning model that analyzes real-time climate and economic data to optimize data center cooling. Data center operators have to deal with extreme environmental conditions like high ambient temperatures as well as economic factors like volatile electricity prices.
The Penn State software can interpret the dynamic shifts in external temperatures, humidity, and economic conditions data centers face when determining cooling parameters. One of the larger uses of data centers is for cryptocurrency mining with well over 100 commercial-level bitcoin mining centers in the U.S. alone.
Although the Penn State model was developed and geared for data centers, it could potentially be adapted to improve the efficiency of cooling in commercial buildings in general.
