Top tips is a weekly column where we highlight what’s trending in the tech world today and list a few tips to explore these trends. This week, we’ll discuss how we can make data centers more energy efficient.
Who knew that AI could cut energy costs?
AI is everywhere, and rightfully so—with applications ranging from everyday life to IT, AI has embedded itself into our culture. Now, it can be an integral part of our data centers, too!
How do you imagine AI in a data center? Making it more efficient, or providing energy savings? It’s exactly how you would imagine. Incorporating AI in data centers is the best way to go now in terms of energy consumption. Most of the existing problems that companies are facing with traditional data centers are eliminated when they start utilizing AI in data centers.
Data centers worldwide are consuming increasing amounts of electricity. In 2022, their consumption reached 460 terawatt hours, and the International Energy Agency (IEA) projects this will double within four years. By 2026, data centers could be using 1,000 terawatt hours annually, a figure comparable to the electricity consumption of Japan, which has a population of 125 million people, according to the IEA.
There are several ways to make data centers more energy efficient. Let’s explore them here.
1. Workload management
What AI does best is optimize processes. Why not use it to optimize data centers?
With the help of AI, data center operations can be optimized using operational data. Resources can be allocated in a way that reduces downtime and minimizes energy wastage. AI analyzes energy-intensive and routine tasks, dynamically utilizing energy based on the processes.
Nvidia’s CEO, Jensen Huang, stated that GPU-based AI data processing and LLMs can enhance performance significantly, achieving up to 50 times the processing speed of CPU-based systems. This advancement means that a few hundred GPU-based systems could replace tens of thousands of CPU-based servers, leading to reduced server space, lower energy costs, decreased core-based software licensing fees, and simplified management.
There are many more processes that aid in the effective functioning of a data center. One such process is maintaining the temperature of the entire system. Why, you may ask? Data centers generate a lot of heat due to the workloads they handle, especially after the advent of AI, which increases the workload on the data center. Without keeping the system cool, the data center is likely to malfunction and cause downtime. Here, AI can lend a hand to assist in keeping the data center cool by analyzing the system and effectively utilizing the power to keep the data center temperature controlled and cool.
According to a white paper published by Siemens, in large data centers, 20% to 30% of servers are typically unused or obsolete, but they still consume electricity. AI leverages real-time data to create algorithms that predict the optimal cooling level needed to maintain the desired temperature.
2. Predictive analysis
Another domain where AI greatly helps is in the analysis of data. As data centers become more complex, it is increasingly necessary to analyze historical data to predict factors like workload and temperature. An interesting aspect of these systems is that they improve over time, and as AI advances, they can analyze more data and predict with greater accuracy.
Huawei has utilized machine learning to create its iCooling intelligent thermal management solution, aiming to reduce energy consumption in data center infrastructure. At Huawei’s cloud data center in Langfang, China, the implementation of iCooling has achieved an 8% reduction in power usage effectiveness, leading to significant annual power cost savings.
Similarly, at a China Mobile data center in Ningxia, the introduction of iCooling technology has decreased total energy consumption by 3.2%, saving over 400,000kWh of electricity annually. Huawei reports that as data center workloads increase and AI learning capabilities advance, iCooling could save up to 6 million kWh of electricity each year, equivalent to a reduction of approximately three million kilograms of carbon dioxide emissions.
A final word
What AI does best is take some work off our shoulders. It does the same for data centers while cutting energy costs.
AI is revolutionizing every aspect of business, including data centers. As we discussed in this blog, optimizing workloads and using predictive analysis significantly save energy. Knowing the benefits, more and more companies are incorporating AI into their data centers. It’s time you did that, too.
