INSIGHTS
How to measure, estimate & quantify energy usage of cloud services
Executive Summary
The streaming value chain is intricate, with a significant portion of emissions occurring downstream of media organisations, often beyond their immediate control. For example, these companies generally cannot directly influence whether viewers purchase larger televisions, nor the energy efficiency of the technology used by internet service providers. However, though it generally makes up a relatively small part of the overall streaming footprint, companies do tend to have more control over the energy and emissions of their data centre and cloud operations.
As companies increasingly shift to cloud-based solutions, there is a widespread understanding that cloud infrastructure is often more efficient than maintaining and operating their own physical infrastructure. Hyper-scale data centres, for instance, typically achieve better energy efficiency by using less power for cooling compared to the energy consumed by IT equipment—a metric known as Power Usage Effectiveness (PUE). Despite this, transparency is crucial to ensure that these efficiencies are realised in practice. Access to reliable, timely, and credible data is essential for understanding the environmental impact, crafting strategies for reduction, and assigning responsibility among internal teams using this infrastructure. In the absence of such information directly from providers, third-party methodologies and tools have emerged to develop proxy estimates for energy consumption.
The scale of cloud infrastructure usage is directly correlated with energy consumption and, consequently, emissions. This connection has given rise to the concept of "GreenOps," where companies seek to improve energy efficiency and reduce costs simultaneously. Encouragingly, there has been an increase in data sharing between cloud operators and their customers, aimed at enhancing corporate disclosure and accounting. However, there is still a need for further improvements in data standardisation, granularity, and timeliness to drive meaningful change.
To truly empower technology, engineering, and product teams—the primary users of cloud platforms—cloud suppliers must recognise that the current level of data granularity and availability is insufficient. Given these challenges, companies need to explore additional options to better understand the energy usage associated with their consumption of cloud services. By improving data transparency and collaboration between cloud providers and users, there is potential to make significant strides in both energy efficiency and emission reductions in the streaming industry.
Factors Contributing to Environmental Footprint of Cloud Services
Cloud computing refers to the delivery of computing services—including servers, storage, databases, networking, software, and analytics—over the internet, often referred to as "the cloud." This model allows businesses to access and utilise IT resources on-demand without needing to own or manage physical hardware. Instead of hosting data and applications on in-house servers, organisations can leverage the vast infrastructure of cloud service providers, which are maintained in large data centres. This shift has significant implications for energy consumption and environmental impact, both in terms of potential efficiencies and new challenges.
Key Factors Contributing to Cloud Computing’s Environmental Footprint
The environmental impact of cloud computing is multifaceted, involving several key factors:
Use Phase Energy of Compute Infrastructure: This refers to the energy consumed by the IT equipment—such as servers, storage devices, and networking gear—during its operation.
Overhead of Cooling, Lighting, and Networking: In addition to the direct energy consumed by IT equipment, data centres require substantial power for cooling systems, lighting, and networking infrastructure. Efficient cooling methods, such as liquid cooling or advanced airflow management, can significantly reduce energy use, but these systems still contribute to the overall energy demand. The relationship between energy used to power the compute infrastructure to the overhead is often called the Power Usage Effectiveness (PUE).
Embodied Emissions of Devices: The production and transportation of IT hardware generate emissions, often referred to as embodied emissions. These emissions occur before the devices are even operational and can be a significant component of the overall carbon footprint of cloud services.
Embodied Emissions of Buildings: The construction and maintenance of the physical buildings that house data centres also have an environmental impact. The materials used, the construction process, and ongoing building maintenance contribute to embodied emissions, which need to be considered when assessing the sustainability of cloud services.
Supply Chain Overheads: The supply chain for cloud computing involves a complex network of manufacturers, suppliers, and logistics providers. Each step in the supply chain, from raw material extraction to the final assembly and delivery of equipment, adds to the overall environmental impact.
Corporate Overheads: Corporate overheads include the energy and resources consumed by the administrative and operational aspects of running a cloud service provider. This encompasses everything from office energy use to business travel, which indirectly contributes to the environmental footprint of cloud computing.
Comparison with Traditional Data Centers
When comparing cloud computing with traditionally operated data centres, several key differences emerge in terms of environmental impact. Traditional data centres are typically owned and operated by individual organisations, which may result in under-utilised resources and less efficient energy use. These facilities often lack the economies of scale and advanced energy management technologies found in hyper-scale cloud data centres, leading to higher energy consumption per unit of compute power.
Moreover, traditional data centres may have more limited capabilities for optimising cooling, lighting, and other overhead energy uses. Cloud providers, on the other hand, can invest in more sophisticated infrastructure and energy efficiency measures, driven by the need to maximise the profitability of large-scale operations. This often results in cloud data centres achieving better PUE metrics and overall energy efficiency.
However, the embodied emissions of cloud services can be higher due to the scale of operations and the need for massive amounts of IT hardware and large buildings to house them. The global reach and extensive supply chains of cloud providers can also introduce additional environmental impacts compared to localised, traditionally operated data centres.
Transparency of Cloud Service Providers
The level of data transparency provided by AWS, Microsoft Azure, and Google Cloud Platform (GCP) varies significantly, which can impact the ability of users to make informed decisions about their cloud usage.
Estimate Emissions
AWS: AWS has introduced tools to help customers understand their carbon footprint, but these tools have limitations. The reported emissions data are often aggregated at a high level, such as across broad regions, making it challenging to pinpoint the impact of specific data centres. Moreover, AWS's current reporting omits Scope 3 emissions, which are crucial for understanding the full environmental impact of cloud services. The lack of detailed, location-specific data makes it difficult for customers to optimise their usage based on environmental considerations. AWS is aiming for net-zero emissions by 2040.
Microsoft Azure: Azure has made strides in providing more comprehensive data to its users. It offers detailed reporting on emissions across all scopes, including Scope 3, which covers the supply chain and other upstream emissions. This level of detail allows organisations to have a clearer understanding of the environmental impact of their cloud operations. Additionally, Microsoft’s efforts to improve data transparency are supported by their internal carbon tax, which incentivizes the Azure team to minimise emissions and improve reporting accuracy. Microsoft is targeting net-zero by 2030.
Google Cloud Platform (GCP): GCP leads in transparency by offering tools that allow customers to choose data centre regions based on carbon intensity. Their Carbon Footprint tool, part of the Carbon Sense suite, provides detailed emissions data, helping users make informed decisions about where to run their workloads. GCP's commitment to transparency is further emphasised by its goal to operate entirely on carbon-free energy by 2030, with current data showing how close they are to achieving this target.
Each provider's approach to data transparency reflects their broader sustainability strategies. AWS’s less granular data may hinder efforts to optimise for environmental impact, while Azure and GCP's more detailed reporting offers better tools for companies committed to reducing their cloud-related emissions. As sustainability becomes increasingly critical for businesses, the ability to access reliable, detailed environmental data will be a key differentiator among cloud providers.
What other information is available on cloud emissions?
While cloud service providers are improving transparency regarding the emissions linked to customer usage, some users feel there’s still work to be done. They seek more confidence in the validity of these claims, more detailed data on which parts of their infrastructure are most energy-intensive, and a clearer understanding of the energy consumption—not just emissions—related to their services.
To bridge these gaps, companies are increasingly turning to open-source tools to verify the data provided by cloud providers. These tools often connect to billing data, mapping it to estimates of server energy consumption. Notable examples include ThoughtWorks' Cloud Carbon Footprint and Boavizta, both of which offer valuable insights despite some limitations.
For companies aiming to dig deeper, a recommended approach involves reviewing results from the provider's sustainability tools, using third-party tools to build independent estimates, cross-comparing results to identify discrepancies, and requesting further data from the cloud provider when necessary. This method helps ensure a comprehensive understanding of the environmental impact associated with cloud usage.
Call to action for Cloud Providers
Cloud users are increasingly interested in understanding the environmental impact of their cloud usage and what actions they can take. To further enable engineering teams using the cloud to deploy and host applications, and sustainability teams integrating cloud usage into their Scope 3 carbon inventories, we propose the following actions for cloud providers:
Standardisation: Establish consistent methodologies for reporting customer-level emissions across all scopes.
Disclosure: Provide location-based emissions data and operational energy consumption by data centre.
Guidance: Offer support to help companies make informed decisions on architecture, application design, resource utilisation, and region selection.
Relevance: Tailor information to the needs of different audiences, enhancing both corporate reporting and engineering decisions.
