By Rohit Pillutla, EfTA Intern
The electricity demand of the IT sector is projected to increase by over 50% by 2030.[1] The main drivers of this growth are generative AI systems, which require billions of parameters to train, and cryptocurrency mining, which can involve computationally intensive verification processes. Data centers will require a significantly greater energy load, with estimates reaching as high as 9% of total electricity usage in the United States alone in the next five years.[2] With this rapid growth in capacity and projected electricity consumption, it is crucial to identify strategies that ensure the sustainability of our computing activities.
One such method is to create green software, a practice which encompasses three principles. The first is to develop an energy-efficient code, or computer code that consumes as little energy as possible to reduce subsequent carbon emissions. The second is to prioritize hardware efficiency. This involves extending the lifespan of local machines to reduce their yearly amortized carbon as well as managing the operations of the data center servers to prevent stand-by consumptions to achieve the highest efficiencies possible. In other words, consumers should aim to use their electronic devices for as long as possible before purchasing replacements, and data center operators should ensure that all servers that are running at any given time are used to their full capacity to avoid unnecessarily using energy to power additional servers. And lastly, the third principle is to be carbon aware, which entails monitoring and communicating the carbon emissions derived from software usage. It includes practices like scheduling computationally intensive activities when the power supply is sourced from low carbon sources and performing less computationally intensive activities when the energy is sourced from high carbon sources, thus, minimizing their environmental impact.
Edge and cloud computing can also offer promising solutions to handle and process large amounts of data in an energy-efficient manner. Edge computing is the practice of processing data closer to its source rather than at a centralized location. This reduces the volume of data that data centers must process, in turn reducing their total energy requirement and carbon emissions. Cloud computing allows businesses to easily scale their server usage according to need rather than expending unnecessary computational power. Furthermore, when using cloud infrastructure, businesses no longer need to construct and maintain their own enterprise data centers, reducing carbon emissions produced by manufacturing new hardware. Due to this, migrating data storage and processing to the cloud has the potential to reduce businesses’ energy consumption by over 60%.[3] Despite these potential savings, it is still important to consider the carbon emissions of data centers where cloud services are hosted.
Another pathway to enhancing sustainability in computing is to decarbonize data centers. One of the most popular strategies among data center owners to accomplish this goal is to use renewable energy sources, such as tidal[4], wind[5], and solar[6]. As mentioned above, improving data center efficiency is key for their decarbonization. Companies such as Microsoft have constructed data centers underwater and in regions with cooler climates, which reduces the energy required to cool servers and networking equipment. HVAC systems powered by artificial intelligence have also been used to optimize airflow and reduce the energy they use for cooling. In areas with frequent power outages, you can equip data centers with backup generators that rely on clean energy instead of diesel. One technology capable of addressing this limitation is battery energy storage systems, especially when combined with onsite clean power generation from solar or wind power.
While the future of the IT sector will inevitably see an increase in energy use, this does not have to come at the expense of the environment. The widespread adoption of green software practices, edge and cloud computing, and data center decarbonization strategies, offer an opportunity to advance computational workloads and support emerging technologies while simultaneously reducing the sustainability impacts of the industry.
[1] https://www.i-scoop.eu/sustainability-sustainable-development/it-sector-electricity-demand/
[2] https://www.reuters.com/business/energy/data-centers-could-use-9-us-electricity-by-2030-research-institute-says-2024-05-29/
[3] https://www.ebcgroup.co.uk/news-insights/why-cloud-solutions-could-help-your-business-reduce-its-energy-costs#:~:text=Furthermore%2C%20research%20by%20Berkeley%20Lab,over%20to%20cloud%2Dbased%20services
[4] MeyGen Tidal Energy Project by SAE Renewables
[5] WindCORES data centers in the Paderborn district of western Germany
