The Singapore government is actively developing a green data center initiative to meet the increasing demand for computational power, largely fueled by extensive AI initiatives.
The nation plans to expand its current 1.4 GW of data center capacity by adding at least 300 MW more in the near future, supplemented by further green energy solutions, as disclosed earlier this week.
While Singapore’s total land area spans only 734 square kilometers—just slightly less than the size of New York City—Singapore contrasts sharply with New York’s data center infrastructure. New York’s data center capability stood at merely 292 MW in 2022, with projections to reach only 540 MW by 2029, based on an analysis by Mordor Intelligence.
Considering that the majority of Singapore’s energy is generated from natural gas, it’s critical to enhance the energy efficiency of existing data centers. The recently unveiled Green Data Centre Roadmap intends to improve efficiency not just in the hardware but in the software operating within data centers. The plan also includes improvements in energy consumption related to non-IT elements like cooling systems, lighting, and power distribution within these facilities.
Many operators of data centers have adopted more advanced cooling methods like liquid cooling and optimizing airflow management to decrease the need for traditional cooling strategies.
In Singapore, efforts are being made to further diminish energy consumption for cooling by increasing server and storage rack operating temperatures. A rise in operating temperature can enable data centers to achieve energy savings ranging from 2% to 5% for each degree Celsius increment.
Servers’ energy efficiency can be enhanced through “the use of software-driven approaches such as server virtualization, employing eco-friendly software solutions, and pinpointing as well as ameliorating software-related carbon-intensive areas,” as stated by the Infocomm Media Development Authority, which is spearheading this initiative in Singapore.
The implementation of virtualization and cloud computing is touted to centralize workloads and enhance efficiency in resource usage, according to proponents of the technology.
The government of Singapore is advocating for data centers to update their IT infrastructure. By swapping out old servers, storage solutions, and network equipment with newer, more energy-efficient technology, there can be a significant reduction in energy consumption related to IT.
Currently, Singapore hosts 70 different data centers that include cloud services, enterprise solutions, and co-location facilities, which provide infrastructure for digital services and more demanding applications such as AI.
The rising demand for AI computation, known for its energy-intensive needs, is leading to an increased need for additional data centers. This growing demand is becoming a challenge as it surpasses the available capacity of local electricity supplies, posing issues for tech businesses and governments globally.
Rinkesh Kukreja, a seasoned software engineer, sustainability advocate, and the founder of Conserve Energy Future, emphasizes the importance of enhancing the efficiency of computer data centers to meet the soaring demands for computational power.
Most operators focus on PUE, a measure of how much of a data center’s power consumption is used for computing rather than ancillary functions like cooling or lighting. A PUE close to 1 indicates high efficiency in a data center.
“While enhancing PUE is beneficial, additional efforts are necessary,” Kukreja states. “The demand for AI technology, which consumes a lot of power through its processors, is increasing at a rate that outstrips the supply capabilities of local power grids globally, not just in Singapore.”
According to Kukreja, adopting a holistic approach is essential for solving this issue. “A vital method is to consolidate servers that are infrequently used to reduce energy consumption, software licenses, and hardware expenses. Moreover, employing deduplication technology for data storage markedly reduces the volume of data stored, which in turn conserves energy.”
It is also crucial to utilize inherent server power management settings and smart HVAC systems.
“These strategies significantly lower the overall energy expenditures by controlling the timing and methodology of energy consumption,” stated Kukreja. “Enhancing airflow for better cooling, modifying HVAC systems, and implementing sensors and controls to adjust cooling based on the computational load are effective techniques to improve efficiency.”
The approach in Singapore for more eco-friendly data centers includes developing chips and software that are energy-efficient yet highly effective. The environmental energy plan recommends concentrating on advanced hardware and software to minimize electrical consumption without compromising performance.
Monitoring power consumption consistently is a key component of this initiative, shared Kukreja. “Choosing locations that focus on sustainable practices, such as the adoption of renewable energy sources, is crucial.”
Kukreja emphasized, “Companies are looking for options that reduce their ecological footprint. Adopting these practices allows facilities to track enhancements and strive for ongoing advancement.”
