How the Data Center Industry is Redefining Energy Efficiency

The rising demand for artificial intelligence (AI)-powered workloads and high-performance computing driving a surge in electricity consumption by the data center industry. Worldwide, data centers account for 1-2% of global electricity use, and in the U.S. data centers are responsible for about 2% of total electricity consumption.

The demand will continue to grow. Today’s generative AI applications require ever more power for increasing volumes of data processing and energy-intensive servers, graphics processing units, and other hardware. As the need for so much computational power continues to grow—potentially at exponential rates—data centers are doing all they can to keep pace, leading to even greater energy demand.

COMMENTARY

Specifically, data center providers are becoming more efficient, reducing their overall energy consumption, and minimizing carbon emissions. As they do, many are exploring the use of renewable energy sources to lower their carbon footprint.

New Innovations

Many are considering other new innovations. For example, data center operators have been exploring the use of AI modeling in their building management systems (BMS) controls. This use case would allow them to use predictive analytics to model factors such as peak computing workloads, internal/external temperatures and humidity levels, equipment load and heat output, and other variables to optimize airflow and cooling in each facility.

Another example is the use of large-scale lithium-ion batteries, which can be used to support renewable energy sources and optimize power usage. These batteries charge during low-demand, low-cost times of the day, often during times when energy from renewables is readily available, and then discharge during high or peak demand times. This helps reduce operational costs, demand on the grid, and the overall environmental impact by minimizing the need for “peaker” plants that often provide the dirtiest types of fuels.

DataBank is currently adopting many of these strategies and pursuing other initiatives to do all we can to combat climate change. We have stated our goal to become net zero by 2030. To achieve this target, we will continue to incorporate highly innovative technologies in all our data centers. This includes monitoring our power use, developing more energy-efficient data center designs, performing efficiency retrofit projects, and expanding our use of renewable power sources.

For example, when it comes to data center design, DataBank now delivers a Universal Data Hall Design (UDHD) that incorporates many innovative new technologies to help with power use and cooling. In the case of cooling, our UDHD builds offer liquid-to-the-rack coils, direct-to-chip cooling, and immersion cooling, where servers sit in a tub of thermally conductive dielectric liquid. These options use a closed-loop heat rejection system to reduce water consumption and operational costs.

My colleague, Eric Swartz – DataBank’s head of infrastructure engineering – and I authored a white paper on the topic.

Key Metrics

DataBank closely monitors and manages a wide range of energy-efficiency metrics in our data centers, aligned with industry standards such as the Green Grid.

Power Usage Effectiveness (PUE): PUE measures how efficiently a data center uses energy and is calculated by dividing the total amount of power entering the facility by the power used to run the IT infrastructure. A PUE closer to 1.0 indicates higher efficiency. Google recently made news by announcing it had successfully achieved an average PUE of 1.10 for all Google data centers, in part by reducing its energy use related to non-computing energy such as cooling and power conversion.

In 2023, DataBank successfully reduced our operational PUE from the previous year, demonstrating more efficient use of power across all of our infrastructure and facilities. We will continue to evaluate our operational PUE to further improve our operations and make our data centers more efficient. This year our goal is to complete five energy-efficiency projects, resulting in over 10,000 MWh of energy savings.

Carbon Usage Effectiveness (CUE) and Water Usage Effectiveness (WUE): Although they are not energy metrics, CUE and WUE are important indicators toward the broader goal of sustainable data center operation, so I have provided an overview:

Carbon Usage Effectiveness (CUE): This metric helps determine the amount of greenhouse gas emissions produced per unit of IT energy consumed within a data center. CUE gives us an effective way to measure the carbon footprint and the environmental impact of our total data center operations. The vast amount of our emissions come from our purchased electrical power. In addition to reducing our energy usage, we also work closely with various local utilities to purchase clean energy. As of the end of 2023, we procured 30% renewable power. We will also continue to conduct greenhouse gas accounting for our full scope of emissions and have already started reporting on other air pollutants and emissions to do all we can to reduce emissions as much as possible.

Water Usage Effectiveness (WUE): WUE tracks water consumption in data centers by measuring the amount of water used for cooling per unit of energy consumed by IT equipment, helping optimize water sustainability.

When it comes to water use in our data centers, DataBank’s standard design does not consume water. Instead, we use closed-loop chilled water systems where the water is recycled through the loop and does not require additional water.

In 2023, we were proud to highlight the fact that we consumed zero gallons of water in our standard data center-designed facilities, representing 85% of our sites. As a result, our water consumption in 2023 was nearly flat, despite the fact that we added new colocation data centers and experienced record growth.

Sustainable Building Practices

DataBank continues to focus on best practices in sustainable building. We recently developed a new facility siting network that takes a wide range of environmental and social concerns into consideration during the siting process.

This includes factors such as the availability of power, especially renewable power, grid resiliency, biodiversity and habitat, physical risks, and more. Incorporating all of these elements will allow for more informed and sustainable siting of our facilities.

The Path Toward Net Zero

Achieving our net-zero emissions goal will require much work. As power demands soar and new energy resources emerge, the entire data center industry will face tough questions – without easy answers – related to energy efficiency and sustainability.

DataBank will continue to explore new innovations and execute key milestones on our net-zero project roadmap to manage our environmental impact, reduce our carbon footprint, and pave the way for more sustainable digital infrastructure in the future.

—Jenny Gerson is head of sustainability at DataBank, where she is responsible for creating and implementing the company’s Environmental, Social, and Governance (ESG) strategy.