Singapore government pushes energy-efficient data center plan

The Singapore government is advancing a green data center strategy in response to rising demand for computing resources, driven in large part by resource-hungry AI projects.

The city state aims to increase its existing 1.4 GW of data center capacity with at least 300 MW of additional capacity in the near term, with more to come through green energy deployments, it announced earlier this week.

Singapore has an area of just 734 square kilometers, making it a little smaller than New York City. But New York had a total data center capacity of just 292 MW in 2022, a figure expected to rise to just 540 MW by 2029, according to an analysis of New York data center market size and share by Mordor Intelligence.

With Singapore generating much of its electricity by burning natural gas, that makes the plan to increase the energy efficiency of existing data centers too all the more important. Singapore’s Green Data Centre Roadmap, announced on Thursday, seeks to make the hardware and even the software running in datacenters more energy efficient, in addition to tackling the usual suspects: the energy consumed by non-IT components such as cooling, lighting, and power distribution within data centers.

Many data center operators have already turned to more efficient cooling technologies such as liquid cooling and improved airflow management to reduce cooling requirements.

Singapore wants to go beyond that and reduce energy use for air-cooling by raising the temperatures at which servers and storage racks can safely operate. Data centers can potentially benefit from 2% to 5% energy savings for every 1C increase in operating temperature.

Software tweaks for energy efficiency

The energy efficiency of servers can also be improved by using “software-based tools such as server virtualization, applying green software techniques, and identifying and addressing software carbon hotspots,” according to Infocomm Media Development Authority, the Singapore government body behind the strategy.

Virtualization and cloud computing help consolidate workloads and optimize resource utilization, the idea goes.

Lastly the Singapore government wants to encourage data centers to upgrade IT equipment. Replacing older servers, storage, and networking equipment with more energy-efficient models can reduce IT energy consumption.

Global challenge

Singapore is home to 70 cloud, enterprise, and co-location data centres providing cloud computing platforms digital services, and higher-intensity workloads for AI.

Demand for AI computing (which requires particularly power-hungry processors) is prompting a demand for new data centers that is outstripping the capacity of local electricity networks to supply them. This has created a challenge for technology businesses and government worldwide.

Rinkesh Kukreja, a senior software engineer and sustainability expert, as well as the founder of Conserve Energy Future, said that the industry needs to make computer data centers more efficient as our need for computing explodes.

Most operators focus on PUE, a measure of how much of a data center’s power consumption goes on computing rather than on ancillary functions such as cooling or lighting. The closer it is to 1, the more efficient the data center.

“While improving PUE helps, more is needed,” according to Kukreja. “Demand for AI tech, which sucks up power with those processors, is growing faster than what local power grids can provide worldwide, not just in Singapore.”

Fixing this requires a comprehensive approach, Kukreja said: “One key strategy is grouping lightly-used servers to save on energy, computer licenses, and machine costs. Additionally, storing data more efficiently with deduplication software significantly cuts down stored data, saving energy.”

Using built-in server power settings and smart HVAC options is also important.

“These measures hugely reduce total energy costs by optimizing when and how energy gets used,” according to Kukreja. “Managing airflow for cooling efficiency, HVAC changes, and sensors/controls to match cooling to computer loads are practical ways to boost efficiency.”

Faster, stronger, cooler

Singapore’s plan for greener data centers involves making chips and programs that use less energy but still work well. The green energy roadmap suggests focusing on efficient hardware and software to cut down on power use without slowing things down.

Tracking energy usage over time is another big part of the strategy, Kukreja said. “Picking facilities that prioritize sustainability, like using renewable energy, is also important.”

Kukreja concluded: “Businesses want options that help lower their environmental impact. Following these methods helps facilities monitor progress and aim for constant improvements.”