The industry buzz as we went into 2020 was climate change and sustainability. Then, the pandemic hit, and survival and finding a vaccine took the front page. At the same time, the pandemic drove massive demand and dependency for remote connectivity and data center resources to an exponential growth rate. The need to build out the digital infrastructure quickly took priority.

Nonetheless, the data center industry continued in its efforts as a leader in improving energy efficiency and increased utilization of renewable energy, even as dozens of 100-MW campuses were being built out at an incredible, record rate to meet demand. Beside massive scale (and partially because of it), virtually all of them had PUE numbers below 1.5.

While the PUE metric clearly defines the measurement process for facility efficiency (thanks to The Green Grid!), it represents only a portion of sustainability in the overall data center eco-structure. We have seen PUE averages for new, traditional enterprise and multitenant data centers (MTDCS), or colos, drop from approximately 2.0 in 2010 to 1.4 or less in 2020. Moreover, the hyperscalers have brought PUE to down to 1.1 or less. So, where can we go from a PUE of 1.1?

Although PUE is the most globally recognized facility efficiency metric, and as important as PUE is, it is only a part of most serious conversations about data center sustainability. Google and other hyperscalers may have already reached some of the worthwhile goals they have set for themselves, such as using 100% renewable energy by means of power purchase agreements and, in some cases, directly investing in renewable generation. They also recognize that we are all still a long way away from 100% non-fossil fuel energy generation, since, according to the U.S. Environmental Protection Agency, only 17% of domestic power is generated by renewables.

How Clean Is the Energy You Use?

“Electricity is produced by many different sources of energy, including, but not limited to, wind, solar, nuclear, and fossil fuels,” per the EPA. “The type and amount of emissions produced depend on how electricity is generated in your region.”

The EPA’s Emissions & Generation Resource Integrated Database (eGRID) website has a tool that allows users to enter their zip codes (or select a region) to view their power profiles.


EPA’s Emissions & Generation Resource Integrated Database (eGRID)
The EPA’s Emissions & Generation Resource Integrated Database (eGRID) website has a tool that allows you to enter your zip code (or select a region) to view your power profile.
Photo courtesy of the U.S. EPA


The cost of solar and wind power generation has dropped tremendously over the last 10 years, making it more cost-effective and helping it to gain more widespread market adoption. However, the obvious issue and limitation for going to pure solar and wind generation is that the power output is both variable and intermittent. As a result, conventional, non-renewables generation is still necessary.

Greenhouse gas emissions (GHG) are another key factor in the sustainability equation. The Green Grid created the carbon usage effectiveness (CUE) and water usage effectiveness (WUE) metrics. Even though they are separate metrics, they are effectively intertwined. This is due to the tradeoff caused by the use of evaporative cooling to reduce cooling energy (as well as lowering PUE). This lowers the GHG emitted by the power source but, of course, increases the use of water.

It’s not that simple though. The ratio varies widely by location, and, as a result, the amount of CO2 emitted also varies.


CUE and WUE ratio
The CUE and WUE ratio varies widely by location, and, as a result, the amount of CO2 emitted also varies.
Photo courtesy of the U.S. EPA


Data Center Site Selection

Moreover, in most cases, data center site selection is based on multiple factors, such as potential business demand, network access and latency to users, geographical risk factors, and the availability and cost of energy. In the past, the cost of energy had been a predominant factor, but renewable energy sources may or may not have been a significant consideration. More recently, access to renewable generation has risen, but it cannot overcome the basic requirements for 7x24 continuous, reliable power sources.

Data Center Availability Versus Efficiency

Part of the challenge is the tradeoff between increased levels of facility redundancy to increase availability and energy efficiency. Data center availability is in the inherent nature of the facility design and operation. Yet, in many cases, more and more facility redundancy level requirements reductions are occurring, primarily due to improved IT virtualization, load sharing, and failover coupled with multisite geo-redundancy supported by high bandwidth network architecture.

Energy reuse has been the holy grail of energy efficiency, since even with a mythical PUE of 1.0, every megawatt hour of energy is being rejected into the environment. While TGG created the energy reuse effectiveness (ERE) metric 10 years ago, it is typically difficult and expensive to use the low-grade waste heat produced by air-cooled IT equipment. Only lately have there been some small but meaningful reuse projects, such as the Data Center Park in Stockholm. However, liquid cooling will help make energy recovery more cost-effective while improving ITE performance.

Energy storage is becoming more cost-effective for both supporting renewable energy sources at utility scale and larger data centers. This also has opened the discussion for data center participation in microgrids and distributed energy resources (DER).

The Bottom Line

What does 2021 have in store for data center designers, builders, operators, and IT staff? Are we pushing sustainability for data centers too far and too fast or not far and fast enough? How do we create and shape a holistic data center metric or evaluation criteria to provide a fair and realistic system giving weight of all these factors?

As we approach Earth Day (April 22), we need to look at the benefits data centers and the entire end-to-end digital ecosystem have brought to improve the human condition versus the ecological impact that is associated with the energy consumed. I believe that, over the last 10 years, the data center industry has done more to drive energy efficiency and expand the use of renewable energy than any other industry or even government.

For comparison, the auto industry (and the world) is trying to make the case for electric vehicles (EVs) as strategy for sustainability (reduce pollution, carbon emissions, and greenhouse gases). They have recognized and looked at well-to-wheels rather than just the efficiency of the vehicle, (measured in miles per gallon [MPG] or miles per kWh [MPGe]. For reference, the EPA defines MPGe on the conversion factor of 33.7 kWh/gallon of gasoline equivalent. The EPA also expresses an EV’s energy consumption in terms of the number of kilowatts per hour needed to run the vehicle for 100 miles (termed kWh/100 mi). Of course, there is no free lunch, despite the green EV hype — in many cases the energy to charge EVs still ultimately comes from power generated predominately by non-renewable sources.

So, will we get to net-zero energy or net-zero carbon? If so, what will it really cost — and who will for it pay?

Yet, what is the cost to do nothing?