The 21st century is ever becoming digitalised, with almost all aspects of life interconnected with one another. ASEAN’s digital economy is estimated to be about 600 million to 1 trillion USD by 2050. Data centres are vital for this interconnectivity to exist. The need for data centres in ASEAN markets will continue to grow, with expected revenue at 13 billion USD in 2028 from 9.6 billion USD in 2023. These data centers have become an integral part of an economy’s infrastructure, used by private and public sectors.  

However, operating these data centres takes huge amounts of energy. Data centres require 340 TW of electricity to run globally, equivalent to 1.3% of the world’s electricity. Data center operations produced around 330 Mt of CO2 in 2020, about 1% of energy-related emissions. Almost all the electricity used in data centre operations is for cooling, an essential part due to the heat generated. Often inefficient, cooling can consume 70% of the total energy supplied for the data centre.  

In addition, some data centers with older hardware may also use more power than newer technology. As economies rely more heavily on data centers, the systems supporting them have also taken up more electricity. These include climate control and systems backup, which are important tools for a data centre to function properly. 

Using renewable energy sources for data centres is an option to address the generated emissions. However, without addressing energy efficiency, relying entirely on renewable energy is not enough for its sustainability. With the number of data centres increasing annually, steps to minimise electricity consumption are important as the world moves towards carbon neutrality. One innovative way that this could be achieved is through Artificial Intelligence (AI).  

As the ASEAN member states are among the largest emerging data centre markets, AI can assist their growing economies while maintaining sustainability goals. Some member countries have begun using AI to address the ever-increasing energy usage from data centres, making them more efficient. AI also offers additional benefits, such as cost-saving and security mechanisms.  

AI can significantly improve energy efficiency in data centres’ cooling systems. One case is Huawei’s iCooling@AI, which uses AI in data centre operations and management. Specifically, AI is used in optimising cold-water cooling systems by predicting and adjusting to changing conditions such as weather, temperature, and usage. Where traditional data centre management would have varied accuracy, using AI would manage the requirements in real-time, as well as predict cooling and energy needs based on historical consumptions.  

Another case is Google, which has used AI to reduce the energy needed for cooling by 40%. Using machine learning on historical data, data centres can maximise energy efficiency by predicting cooling needs at different periods depending on peak usage hours, weather and other external factors, for different technologies and structures of various data centres. 

The historical data on Power Usage Effectiveness (PUE), the rate of energy used versus energy delivered by the computing equipment, could also be monitored automatically. Some regulations are placed by ASEAN member governments to track PUE in data centres. For example, Malaysia set a minimum requirement of 1.9 PUE, and an excellent rating of 1.6 PUE. Singapore developed a more significant standard requiring new data centres to have a PUE rating of 1.3 or lower, which is categorised as very efficient. Globally, PUE was around 1.6 from 2014 to 2021, and optimised at 1.55 in 2022. By utilising advanced hardware and software in data centres, AI can maximise energy efficiency, creating a greener data centre. 

Automating systems management in a data centre could run the system more efficiently, thus resulting in cost-saving. Optimising power utilisation and server maintenance will decrease energy needs. Using AI to use the smallest amount of electricity possible would lower energy costs for the data centres and curtail the risk of data outages and server failure. Reducing the downtime due to maintenance would generate fewer costs, as AI will automate regular maintenance and monitor any potential risk of system failure.  

An example of systems failure and inefficient energy allocation in a data centre happened in the UK due to a heatwave that reached 40 degrees Celsius. The failure showcased how ineffective systems management and lack of efficient cooling systems can lead to data centre outages. Using AI to manage cooling systems and operation management can avoid these systems’ failure and reduce operating costs from data centres.  

AI can also help improve security for data centers. Traditional cybersecurity may not be enough to keep up with the increasing threat to private data. AI could predict threats and detect risks faster than manually, reducing the possibility of a major data leak and damage to the data centre. As the world and ASEAN economy especially become more digital, ensuring data centres are protected from threats is crucial to ensuring market activity runs as seamlessly as possible.  

With ASEAN’s focus on energy security, accessibility, affordability, and sustainability for all, AI utilisation may facilitate the growing digitalisation of its market while realising energy targets. The current aspirational target is set at 32% energy intensity reduction by 2025 from the 2005 level. As ASEAN countries are among the fastest-growing markets in the world, they would need to maximise energy efficiency to ensure that increasing energy demands are met at a reasonable price while also maintaining emissions. Ensuring that energy efficiency is achieved in a power-hungry sector, such as the digital industry, is important for ASEAN to maintain its course towards a greener economy without sacrificing energy reliability. AI would provide extra benefits to data centers in reducing energy intensity, lowering operating costs, and improving the security of digital services.