Tips from IDC to improve the environmental sustainability of your IT workloads

As the world becomes increasingly digitalized, enterprises are facing a complex challenge – balancing the need for energy-intensive IT infrastructure to drive digital transformation with environmental sustainability priorities. This competing demand has put data center energy consumption and carbon emissions firmly in the spotlight.

According to a new IDC InfoBrief report, Energy and Carbon Efficiency Benefits of Public Cloud Computing over Enterprise Datacenters, ¹  public cloud data centers were estimated to be 3.8 times more energy-efficient than enterprise data centers in 2023. IDC, the premier global provider of market intelligence, advisory services, and events for the information technology, telecommunications, and consumer technology markets, estimates that public cloud data centers had a Power Usage Effectiveness (PUE) of 1.22 in 2023, compared to 1.84 for enterprise data centers. IDC’s research also found that public cloud data centers were estimated to be 4.7 times more carbon efficient than enterprise data centers in 2023, and this gap is expected to widen to 7 times by 2027. As a result, IDC estimates that by 2027, carbon avoided due to public cloud usage is expected to be equivalent to removing 21 million U.S. cars from the road. IDC defines Carbon Usage Effectiveness (CUE) as a metric used to measure the carbon emissions associated with operating a data center that is particularly useful for assessing the environmental impact of data centers, as it directly relates energy consumption to carbon emissions.

According to IDC, there are several key factors driving this carbon efficiency:

• Carbon-free energy sources: Cloud providers are investing heavily in carbon-free energy. IDC estimates the percentage of public cloud energy consumption powered by carbon-free energy sources, such as solar, wind, nuclear, and hydroelectric power to increase from 61% in 2023 to 74% in 2027.
• More efficient hardware and facilities: Cloud data centers are designed from the ground up with energy efficiency in mind, leveraging advanced cooling systems, optimized server layouts, and smart building technologies. They also benefit from economies of scale, allowing them to achieve greater returns on their efficiency investments.
• Improved utilization: Cloud providers are able to maximize the utilization of their infrastructure through virtualization and containerization, ensuring computing resources are used more efficiently. In contrast, enterprise data centers often suffer from lower utilization rates.
• More energy-efficient silicon: Cloud providers are investing in the development of custom, energy-efficient silicon chips tailored to specific workloads like AI, machine learning, and high- performance computing. These chips are designed to deliver higher performance with lower power consumption, significantly reducing the overall energy requirements of public cloud data centers.

The carbon reduction opportunity of the cloud extends beyond just operational emissions. IDC’s research also found that the use of public cloud data centers can result in 34-37% less embodied carbon annually, compared to enterprise data centers. As it relates to data centers, embodied carbon is the indirect emissions associated with building a data center and manufacturing hardware. One source of embodied carbon in data centers is the server infrastructure. IDC concluded that because public cloud data centers utilize servers more efficiently, fewer are required to perform the same set of tasks. This translates to lower embodied carbon from not having to manufacture and transport those servers. IDC calculates the potential embodied carbon savings from using public cloud services to be 28 MMTCO2e in 2027 alone – the equivalent of removing more than 6 million cars from the road.

As a result of the energy and carbon efficiency benefits of public cloud data centers, IDC estimates that running generative artificial intelligence (generative AI) workloads in public cloud data centers is more energy and carbon efficient than running them on enterprise data centers. While generative AI can be energy intensive, the report outlines the sustainability advantages of generative AI:

• Training, the most energy-intensive portion of generative AI, is not latency dependent. This means training workloads can be located where power is available, especially when that power has low or no carbon emissions.
• Customers generally intend to use third-party data centers. These are more energy-efficient and are more likely to be powered by carbon-free sources.

The report outlines several key recommendations for enterprise leaders looking to improve the environmental sustainability of their data center portfolios:

• Assess your current footprint. Start by understanding your existing IT infrastructure’s energy consumption and carbon emissions. Senior executives should consider what workloads should move to the public cloud to support their sustainability strategy and to achieve energy and carbon efficiency benefits.
• Select the right public cloud provider. Select a provider that aligns with your values and sustainability goals and offers relevant solutions for your needs. Additionally, look at more than the environmental sustainability of the data center portfolio. Environmental, Social, and Governance (ESG) goes beyond energy consumption and carbon emissions. Choose a public cloud provider with strong social and governance practices that align with your values.
• Consider the location of the data centers. The overall grid generation mix and whether carbon-free energy is available can affect your environmental impact. Data centers in cooler climates can reduce energy needs for cooling systems.
• Use CloudOps tools and implement best practices. CloudOps tools and practices optimize server utilization and workloads, right-size resources, and effectively enable demand-based scaling. Additionally, CloudOps tools limit the rebound effect by implementing resource quotas and monitoring tools to prevent scope creep.

IDC’s research concludes that moving IT workloads to public cloud data centers can support enterprises’ sustainability strategy and help them achieve energy and carbon efficiency benefits highlighted above. By leveraging the energy efficiency levers of cloud computing, organizations can reduce their carbon footprint while driving innovation and business value. For more information about the energy and carbon efficiency of public cloud computing, read Energy and Carbon Efficiency Benefits of Public Cloud Computing over Enterprise Datacenters. You can also read more about AWS commitments to Sustainability.

1. IDC InfoBrief, sponsored by AWS, Energy and Carbon Efficiency Benefits of Public Cloud Computing over Enterprise Datacenters, Doc. #EUR251921924, April 2024