In order to sustain the long-term, high-growth trajectory of the data center industry and meet increasing demands for cloud computing, the sector’s corresponding rise in energy consumption and carbon emissions needs to be managed in a 100% green way. As more enterprises migrate from legacy on-premise data centers to cloud computing alternatives, green technology presents a significant opportunity for these organizations to avoid costly carbon emissions – both for their business and for our climate.
Green cloud computing involves a cloud service provider operating under the principles of reducing energy consumption and carbon emissions, by running all of its self-operated and third-party data centers with 100% renewable electricity, from energy sources including wind and solar.
Per the International Telecommunication Union (ITU), data centers, including their 48 million servers, are estimated to consume 230 terawatt-hours (TWh) of electricity annually. Globally, this means that data centers account for around 2% of electricity consumption and 0.4% of greenhouse gas (GHG) emissions.
With the cloud computing market continuing to grow by more than a 15% compound annual growth rate (CAGR), it is critical that data center energy consumption and carbon emissions are managed using green approaches and technology.
What is Green Cloud Computing?
Green cloud computing is the practice of a cloud service provider (CSP) operating under the principles of reducing energy consumption and carbon emissions, by running all of its self-operated and third-party data centers with 100% renewable electricity, from energy sources including wind, solar, and hydroelectricity.
In turn, CSPs help their own customers, such as start-ups and large enterprises, reduce their energy consumption and carbon emissions, by owning and managing the compute, storage, and networking resources, which deliver carbon-neutral services.
As an example, Google Cloud invests in renewable energy to match the electricity it uses across its entire operations, which means that its client TELUS, a wireless carrier in Canada, can run workloads on Google Cloud that are matched with Google’s renewable energy purchases.
As a result, TELUS’ operational carbon footprint of running any workloads on Google Cloud is zero – reducing its Scope 2 and Scope 3 carbon emissions. Moreover, by working with Google, TELUS receives the benefit of economies of scale when using less electricity.
Scope 1, 2 and 3 Carbon Emissions in Data Centers
To this end, Google Cloud’s CEO, Thomas Kurian, recently highlighted that sustainability is a key driver in an organization’s decision to migrate workloads from an on-premise environment to a green cloud:
“As organizations look over a multi-year timeframe on their sustainability commitments. We’re seeing a lot of interest in the fact that we run a green cloud. We’ve been carbon neutral since 2007, so 14 years now. And whether it’s Unilever or various other organizations, we see that as a big source of interest in helping them meet their sustainability objectives as well.”
Benefits of Green Cloud Computing
Cloud computing delivers green benefits, as compared to on-premise data centers, by reducing energy consumption and carbon emissions in four primary ways: power & cooling efficiency, improved server utilization, hardware efficiency, and more optimal renewable energy procurement.
Overall, Accenture states that migrations from on-premise data centers to green cloud computing can reduce carbon dioxide (CO2) emissions by 59 million tons per year, equivalent to taking 22 million cars off the road.
Below is further specificity on how power & cooling efficiency, improved server utilization, hardware efficiency, and more optimal renewable energy procurement can reduce energy consumption and carbon emissions:
1) Power and Cooling Efficiency
Energy consumption for data centers is typically measured using an efficiency metric known as power usage effectiveness (PUE). Specifically, PUE is the ratio of the total amount of electricity consumed by a data center, including for power distribution and cooling, to the amount of electricity delivered to its IT equipment – with a metric closer to 1.0 being optimal.
READ MORE: How Data Centers Impact the Environment
As shown below, cloud service provider (CSPs) like Microsoft Azure, Google Cloud, and Alibaba Cloud all report a materially lower PUE – in the range of 1.10 to 1.25 – versus the average data center, which achieves a PUE of 1.55, according to the Uptime Institute’s 2022 Global Data Center Survey.
Power Usage Effectiveness (PUE) – Cloud vs Data Center
|Data Center Operator||Power Usage Effectiveness (PUE)|
|Data Center – Average||1.55|
Even the world’s largest colocation data center operator, Equinix, only achieves an annual average PUE of 1.48 across its global portfolio. Moreover, Equinix expects to achieve a PUE of just 1.45 or less for its new green data center construction.
Two examples of how cloud computing achieves superior power and cooling efficiency are i) dynamic provisioning and ii) multi-tenancy:
- Dynamic Provisioning: cloud computing resources can be dynamically scaled to match server capacity with actual demand, which reduces the over-provisioning of IT equipment. In contrast, on-premise data centers are commonly over-provisioned with active IT equipment to ensure peak business demands are met
- Multi-Tenancy: cloud computing serves multiple users at the same time, which aggregates and flattens demand patterns. Therefore, the cloud requires less incremental capacity to meet overall peak demand, as compared to on-premise data centers. By aggregating demand with other cloud service provider customers, businesses can more efficiently share and use resources
By more effectively utilizing resources, cloud computing delivers greater power and cooling efficiency.
2) Server Utilization
Cloud computing utilizes virtualization, which enables fewer overall servers to run at high utilization levels, in order to fulfill the same end user demand. In contrast, on-premise server deployments often achieve low average utilization of only 10% to 15% of total capacity. As such, server utilization rates in cloud computing are several times greater than typical on-premise deployments.
Overall, it is more energy efficient to run fewer highly-utilized servers, than a greater number of servers at low utilization levels.
Why is Virtualization a Green Cloud Technology?
Virtualization is a green cloud technology because its superior energy efficiency leads to less energy consumption. Additionally, running fewer servers ultimately results in the creation of less waste from electrical and electronic equipment, also known as e-waste. Finally, a smaller number of servers can be housed in a facility with a smaller overall floor space, meaning a reduction in the size of a data center.
3) Hardware Efficiency
Cloud computing takes place in large “hyperscale” data centers, with sizable power capacities, which have the ability to optimize power usage efficiency to be green. In turn, cloud service providers can deploy their hardware, including servers and networking equipment, more efficiently.
Given that cloud data centers tend to be large and standardized buildings, newer and more efficient hardware can be optimized at-scale for cloud service providers. This leads to greater hardware longevity, modularity, and circularity – all key components of efficiency.
4) Renewable Energy Procurement
Cloud service providers (CSPs) utilize a superior renewable energy mix, such as wind, solar, and hydroelectricity, than on-premise data centers, minimizing their carbon footprint. Given the scale and sophistication of CSPs, they are better positioned to employ two primary mechanisms to procure renewable energy:
- Direct Sourcing: locating their data centers adjacent to or near renewable energy sources, such as wind and solar farms
- Partnerships: using power purchase agreements (PPAs), virtual power purchase agreements (VPPAs), and renewable energy certificates (RECs) to increase their renewable energy mix
Two key factors allow the cloud service providers to more easily procure renewable energy. Firstly, the cloud service providers are typically the only tenant inside their data centers. Secondly, the cloud service providers have the ability to locate their data center sites near renewable energy sources.
Case Studies – Migrating from On-Premise to Green Cloud Computing
Recent studies from Amazon Web Services (AWS), Microsoft Azure, and Alibaba Cloud, show that an on-premise data center to cloud migration can yield green benefits ranging from a ~50% to ~80% reduction in energy consumption and a ~70% to ~90% decrease in carbon emissions. Further details on the benefits of these cloud migrations are outlined below:
- Amazon Web Services (AWS): white paper showed that most enterprises would experience an 80% to 93% reduction in their carbon footprint by moving to AWS. While even the top 10% most efficient organizations would still receive a 72% reduction in their carbon footprint by moving to AWS
- Microsoft Azure: study showed that Azure Compute is 52% to 79% more energy-efficient than compute equivalents deployed in on-premise data centers. While Azure Storage is 71% to 79% more energy efficient than storage equivalents deployed in on-premise data centers
- Alibaba Cloud: research showed that customers in China can avoid 85.5% of their computing emissions by moving to Alibaba Cloud from on-premise data centers, largely driven by the efficiency of Alibaba’s owned data centers. Specifically, the largest carbon savings come from improved power usage effectiveness (PUE) and reduced power draw of servers during a data center’s use phase
Green Data Centers
While the cloud computing business model drives a number of green efficiencies, the physical data center, which houses computer systems, servers, routers, data storage devices, and communications equipment also contributes significantly to reducing energy consumption and carbon emissions.
Green data center initiatives can be split into five primary categories: green buildings, renewable energy, energy efficiency, water & wastewater management, and waste management. Below are further details on each of these five categories:
1) Green Buildings
Green buildings refer to the design, construction, and maintenance of sustainable data centers and ancillary buildings such as office and administration buildings. Depending on where a data center is located geographically, different green building certifications and standards apply.
For example, the most common green data center certifications and standards, along with their corresponding energy-efficient levels, are:
- LEED: Leadership in Energy and Environmental Design – targeting Gold or Platinum levels
- BREEAM: Building Research Establishment Environmental Assessment Method – targeting Very Good, Excellent, or Outstanding levels
- BCA Green Mark: Building and Construction Authority (BCA) of Singapore – targeting GoldPlus or Platinum
- Green Globes: administered by the Green Building Initiative (GBI) – targeting three or four Globes
- NABERS: National Australian Built Environment Rating System – targeting 4.5 stars or better
- ISO 14001: Environmental Management
- ISO 50001: Energy Management
In China, the Ministry of Industry and Information Technology (MIIT), National Government Office Administration, and National Energy Administration jointly published the Guidance on Promotion of Green Data Center Construction, as well as the Three-Year Action Plan for the Development of New-type Data Centers. These green directives encourage new data centers and cloud technology to follow certain levels of energy conservation, including:
- Newly constructed large and extra-large data centers to achieve a PUE at or below 1.4 from 2022 onwards and 1.3 by 2025
- Newly-built data centers in cold and frigid regions to achieve a PUE at or below 1.25
2) Renewable Energy
Renewable energy generation and procurement at green data centers involves direct sourcing, power purchase agreements (PPAs), and new backup generator technology:
Direct sourcing comprises on-site renewable energy generation, such as installed solar capacity, biogas, or hydrogen fuel cells. For example, hydrogen fuel cells place electricity generation next to energy consumption, which improves efficiency, removes transmission losses, and increases resiliency by taking load off the grid.
Overall, power from fuel cells is 20% to 45% cleaner than the equivalent natural gas-powered generation from a utility.
Power Purchase Agreements (PPAs)
Data center operators enter into power purchase agreements (PPAs) with contract terms ranging from 10 to 20 years. As part of these agreements, data center operators can purchase renewable energy and renewable energy credits from producers at fixed prices. These agreements can either be direct (physical) or virtual (financial) PPAs.
As highlighted below, green data centers and PPAs create a partnership between a renewable energy provider, like Brookfield, and a data center operator, like Amazon Web Services (AWS):
“With data center needs, our renewable power business and our transition fund just announced a global partnership with Amazon to help provide green data centers. And data centers are very large consumers of electricity, and therefore, it’s critically important to source that electricity from renewables.”
-Sachin Shah, Managing Partner and Chief Investment Officer of Brookfield Asset Management
Presently, diesel-powered backup generators serve as auxiliary and emergency power sources at data centers. Specifically, diesel fuel is stored in bulk, in above ground storage tanks for backup generator use.
Using diesel-powered backup generators is often the cheapest solution for meeting a data center’s backup power needs. However, when this diesel is used (burned) it creates emissions, including carbon dioxide (CO2), a greenhouse gas.
Solutions to this problem include converting on-site diesel-powered backup generators to run on carbon neutral and non-fossil fuels, such as biodiesel or hydrotreated vegetable oil (HVO).
3) Energy Efficiency
Energy efficiency initiatives aim to increase the energy performance of green data centers through innovative cooling systems, green cloud technology, and community heating & cooling schemes.
Several green and innovative cooling systems are being installed in data centers, which can replace or supplement traditional computer room air conditioner (CRAC) units. Specifically, this energy-efficient green cloud technology includes liquid cooling, evaporative cooling, free cooling, hot and cold-aisle containment, and aquifer thermal energy storage (ATES):
- Liquid Cooling: servers are directly immersed in insulated dielectric liquid with thermal conductivity. The coolant heated by the server is pumped to a heat exchanger unit where it is cooled down by naturally cool air outside. Ultimately, liquid cooling can reduce power consumption by up to 30%, compared to conventional air-cooled servers
- Evaporative Cooling: utilization of dry air energy, as well as cold water or cold air at low temperatures through the exchange of heat and moisture between water and dry air. This method results in improved power and water efficiency
- Free Cooling: systems that use the naturally cool air outside of data centers for cooling, which can reduce or eliminate the reliance on access to water for cooling
- Hot and Cold-Aisle Containment: uses physical barriers and seals to segregate and reduce the mixing of cold air in data center supply aisles, with the hot air in exhaust aisles. In turn, this approach reduces energy consumption and enables more efficient cooling
- Aquifer Thermal Energy Storage (ATES): uses a series of wells to tap into stable temperatures deep underground which act as a heat sink to dissipate thermal energy, improving the efficiency of a data center’s cooling system
Green Cloud Technology
Green cloud technology for data centers involves deploying sensors, controls, and artificial intelligence (AI) for energy optimization, management, and plant efficiency. As an example, Alibaba Cloud attributes its sustainability performance to a variety of energy-saving strategies, such as AI-powered management algorithms and water cooling techniques.
Community Heating and Cooling
Community heating and cooling schemes allow green data centers to deliver broader energy and carbon savings to local communities. More specifically, by connecting data centers to local district heating networks, waste energy, from a data center’s IT equipment and cooling systems, can be reused and recycled to heat thousands of homes and businesses.
4) Water and Wastewater Management
Water is a key component used in the cooling systems of data centers. For example, in 2021, the data centers of Facebook (Meta Platforms) accounted for over 3.4 billion liters (3.4 million cubic meters) of water withdrawal.
Below is a further breakdown of the water withdrawal of Facebook (Meta Platforms), by each of its data centers:
|Data Center||Water Withdrawal|
|Altoona, Iowa||140 million liters|
|Clonee, Ireland||928 million liters|
|Eagle Mountain, Utah||58 million liters|
|Forest City, North Carolina||64 million liters|
|Fort Worth, Texas||254 million liters|
|Henrico, Virginia||80 million liters|
|Huntsville, Alabama||39 million liters|
|Los Lunas, New Mexico||153 million liters|
|Luleå, Sweden||39 million liters|
|New Albany, Ohio||121 million liters|
|Newton County, Georgia||105 million liters|
|Odense, Denmark||373 million liters|
|Papillion, Nebraska||106 million liters|
|Prineville, Oregon||354 million liters|
|East Coast Leased Data Centers||558 million liters|
|Data Center-Related Facilities||45 million liters|
|Total||3.4 billion liters|
With such a significant amount of water being consumed by data centers, the following water-related green data center initiatives are important:
- Water-efficient cooling solutions (see “Cooling Systems” above)
- Water infrastructure upgrades, including metering and reporting
- Shared cooling systems, such as river water cooling loops
- Reductions in chemical use in the treatment of water
- Installation of green or living roofs, resulting in the reduction of stormwater runoff
- Rainwater capture systems
5) Waste Management
Waste management for green data centers focuses on pollution prevention and control, including waste reduction and recycling. In particular, the amount of waste from electrical and electronic equipment, also known as e-waste, represents 2% of the trash in U.S. landfills, but 70% of its overall toxic waste.
As such, avoiding e-waste is important. To this end, green data centers can switch from using a static uninterruptible power supply (UPS) to a dynamic UPS, which reduces the environmental load caused by recycling the static UPS batteries.
Additionally, the circular economy principles of reduce, reuse, and recycle are important when consuming data center hardware, which ends up as e-waste.