Reducing Your Cloud Computing Climate Impact


More energy-efficient data centers running on low-carbon electricity grids are critical to reducing greenhouse gas emissions in cloud computing. In this article, Leah Goldfarb, environmental impact officer,, shares the four areas where progress is being made.

The news about climate change is usually dark, gloomy, and tinged with helplessness. However, there is some good news to report: the global effort to fight climate change is making progress thanks to the Paris Agreement. The Paris Agreement was established in 2015, and its main goal is to prevent global temperatures from rising by more than 2 degrees Celsius (above pre-industrial levels) in this century. To date, 193 parties have joined the Paris Agreement. According to new findingsOpens a new window by scientists and academics, if all parties keep their promises, the Paris Agreement may be doable, and that is excellent news. 

The technology industry should also be doing its part to fight climate change. In fact, digitalization was cited by the Intergovernmental Panel on Climate Change (IPCC) as one systematic way to reduce greenhouse gasses but only if properly managed. Currently, the Information and Communications Technology (ICT) sector represents 4% of global greenhouse gas emissionsOpens a new window , more than the aviation sector.

As cloud computing consolidates resources, it is generally a more inherently energy-efficient technology than individual machines. Most of the attention in reducing the carbon impact of cloud computing has focused on improvements in data center hardware. But the focus is now pivoting toward the overall energy usage of the cloud sector.

Using fewer resources will be vital to optimizing the energy use by data centers. With progress being made in the following four key areas, the ability to maintain and deliver quality online services will be achievable:

1. Improvements in hardware

In addition to more efficient computing processes (faster CPUs, better energy usage for disk access and storage, extending the hardware lifetime), there is the energy usage associated with housing and cooling data centers, generally measured by Power Usage Effectiveness (PUE). Over the past decade, this metric has improved by 20% globally. It is further expected to continue to do so over the coming years, according to scientific studies.

2. Tech sector’s commitment to funding renewable energy

Services like AWS, Azure, and Google Cloud Platform (GCP) have made public announcements concerning their 100% renewable energy, carbon neutrality, and net-zero commitments (table below). These goals are typically centered on the purchase of new renewable energy certificates (particularly Power Purchase Agreements [PPA]) and carbon offsetting projects. The operations, which usually take the form of paying a third party to take the appropriate actions, are a step in the right direction. But as the UN has stated, more urgent action on climate change is needed.

Additionally, when organizations such as the New Climate Institute and Carbon Market Watch evaluatedOpens a new window two of the parent companies (i.e., Amazon and Google), their assessment was less rosy than the cloud providers’ claims. They found that Amazon and Google’s climate pledges were judged to have low integrity. The exception to this was many of Google’s Renewables Power Purchase Agreements (PPAs), which were frequently on the same electrical grid, and their net-zero emissions targets covering the full value chain, which was judged to be high integrity.

As more scrutiny is given to these announcements, increased transparency and lower carbon emissions are expected to follow.



100% Renewable Energy  Carbon Neutral (all by 2040) Net Zero 
GCP  2017Opens a new window 2007Opens a new window 2030Opens a new window * 

*carbon-free energy 24/7

AWS 2025Opens a new window 2040Opens a new window 2040Opens a new window
Azure 2025Opens a new window 2012Opens a new window

2030Opens a new window ** 


Source: New Climate Institute and Carbon Market WatchOpens a new window

3. Coding and deployment

Developers have always prided themselves on writing elegant code. With increased code observability tools (e.g., BlackFire) that show which processes use the most resources, developers will be able to speed up their applications and reduce their carbon footprint. There is also the question of deploying to a grid or dedicated cloud environment. When the developer can choose where to deploy, grid environments are inherently less resource-intense because there is more mutualization. Typically, the most resource-intense option is a private cloud maintained on-premises.

See More: Now and Forever: The Sustainability Battle Between the Big Three of Cloud

4. Localization of work packages on low-carbon energy grids

Choosing the localization of a cloud provider is where an organization has the potential to reduce its carbon emissions by up to a factor of 10, and in some cases, even up to 20. When work packages are deployed to the cloud, there is often little thought to where it will run other than somewhere geographically close to the end-user. Using the data center’s location, companies can leverage an open-source toolOpens a new window to find the carbon intensity of the underlying grid in real-time. 

An example is illustrative. Let us say you use MS Azure, and you have a choice between two regions: West US 2 and West US 3. This seems like a very similar choice. However, West US 2 is located in Quincy, Washington and runs on 100% renewable energy. West US 3 is powered by the Arizona public service company, which currently emits ~20x more carbon per kilowatt-hour used.

While not all region choices are as extreme, it illustrates why it will be essential to measure our carbon impact, run more efficient codes on the cloud (preferably on a grid environment, if possible), and locate these work packages in areas that run on clean energy.

How are you contributing toward reducing greenhouse gas emissions in cloud computing? Let us know on FacebookOpens a new window , TwitterOpens a new window , and LinkedInOpens a new window .