Podcast transcript: How can AI make data centers more energy efficient?

08 min | 04 January 2023

In conversation with:

Alexy Thomas

Alexy Thomas
EY India Technology Consulting Partner

Silloo Jangalwala: Hello, this is Silloo welcoming you to a new episode of the EY ESG Podcast series, where we look at the most important issues that India Inc. needs to deal with in its environmental, social, and governance or ESG journey, and how technology can help. Today, we are discussing sustainable data centers as they now play an increasingly big role in data storage.

Data centers are not storage rooms for computers. They contain complex arrangement of power equipment, water treatment facilities, and cooling systems. These ensure that computers can operate smoothly round the clock. Data centers need a lot of power to run, but with AI and automation, we can considerably reduce power consumption, which makes data centers greener and more sustainable. To explain how this works, we have with us today, Alexy Thomas, technology consulting partner on Data Analytics and ESG at EY India, joining me.

Welcome to the podcast, Alexy.

Alexy: Thanks for inviting me, Silloo. As you mentioned, data centers consume large amounts of electricity. This huge power consumption directly influences organizations’ ESG goals. So, I feel that in the next two years, the top priority for data centers is going to be reducing power consumption and becoming more sustainable and socially responsible. This is the right time to discuss all these topics.

Silloo: Thanks, Alexy. Let me start with the basics here. Exactly how power hungry are data centers?

Alexy: Well, according to the International Energy Agency, data centers account for around 1% of the global electricity demand. You know, with digital transformation across sectors gaining momentum, the demand for data services is rising exponentially. I recently saw that a rating agency has predicted that the capacity of data centers in India will record a five-fold increase in the next five years. This growth in the data center market means that there will be more demand for power, data center space, and of course, skilled workforce. 

In a recent study, global real estate consultant JLL said that energy consumption by data centers is going to double every four years and the sector accounts for up to 4% of total greenhouse gas emissions globally.

So, you can see from different studies that data centers are going to be power guzzlers, consuming a large proportion of the power that we as humans use on the planet. 

Silloo: That is interesting. So, Alexy, how can AI and automation tools help? 

Alexy: Modern data centers are addressing these problems by deploying AI. So, AI coupled with automation, IOT, and machine learning, are now helping many data center operators design and build lean, smart data centers. You know, these cutting-edge, AI-based solutions that automate the movement of workloads in the data center to the most efficient infrastructure in real time are going to help address this problem.

Also, rule-based AI will automate resource optimization through predefined configurations, which will further improve the workload patterns and match these demands within data center capacity, both inside the data center as well as in a hybrid cloud setting. So, these solutions will ensure that data center capacity is utilized in the most optimal manner, which will in turn reduce the energy consumption.

Silloo: Thanks, Alexy. Moving on, what do you think are the major AI and automation tools coming in? How will they manage data center workloads and make them sustainable?

Alexy: Silloo, there will be many different types of AI-ML solutions that will be used within data centers. I will give you some examples: digital twins — real-time virtual representations of the real world — are now becoming key for data center efficiency improvement. Data twins allow data to be collected from all sources and help data centers operate more sustainably, not only from a cost perspective but also from an environmental standpoint.

So, starting from facility design and space utilization, digital twin technology, as an example, reduces carbon footprint. The more powerful the data center, the more heat it generates, thus consuming more energy for its cooling systems. Now, real-time control of cooling equipment with sensors and ML reduces the amount of energy spent on cooling, which shrinks the energy bills and the carbon footprint.

You know, these kinds of software learn by continuously analyzing sensor data as well and adjust according to the environmental change. So, with pragmatic usage of different types of AI, companies can save up to 40% of the power spent on, say, the overall data center needs, including cooling.

Silloo: Well, Alexy, how else do you think AI enhances data center functionalities?

Alexy: AI solutions help in improving energy efficiency and reducing carbon emissions. But they also help in a myriad of ways, like providing predictive maintenance, improving security, automating routine activities, and reducing workforce requirements. Traditional data centers monitor and report data outages manually. AI can monitor server performance, network congestion, and disk utilization, and predict data outages, thus minimizing the data center downtime.

I have seen AI-ML-powered smart cameras being used for intrusion-detection systems and robots to ensure physical protection of data centers. AI can be a great tool for cybersecurity because AI systems learn normal network behavior and can spot deviations. So, as you can see from the examples that I just gave, AI can be used in many ways; not just from an ESG point of view, but also in terms of making the operations of data centers more efficient.

Silloo: Excellent. Thanks a lot for your time and for joining us, Alexy. I really enjoyed this very informative and engaging conversation.

Alexy: It has been a pleasure. Thanks for having me.