AI, Data Centers, and Energy Use: The Path to Sustainability
The increasing use of AI and data centers is leading to a surge in energy consumption, posing risks for energy, tech, and data companies. It also presents an opportunity for these companies to decarbonize, build trust, and reduce long-term costs.
Thanks in part to skyrocketing artificial intelligence use, data center usage is booming -- so is energy consumption. This could spell risks for energy, tech and data, and companies alike. This also creates a significant opportunity for data and tech companies to decarbonize, build trust and reduce their long-term costs.
Data center energy consumption has already surpassed the energy consumption of several developing countries. Several estimates indicate that data centers make up about 2% of global greenhouse gas (GHG) emissions -- on par with the entire aviation industry. By 2040, projections show that the number could increase to about 14% of global GHG emissions.
Emissions are not the only problem. Data centers also require a lot of power. Massive increases in data centers would put incredible strains on an electricity grid that is already grappling with increased power needs from things like plug-in hybrids, electric vehicles, and a resurgence in US manufacturing.
This strain on the grid has led to outcomes including some power companies scrapping plans to phase out plants that rely on burning fossil fuels and others considering building new natural gas facilities to keep up with increasing demand. This demand is so high that some states and counties are putting moratoriums on accepting new data centers because their utilities simply cannot provide the amount of energy they require in the timeframe the companies want them built.
All of this combined could pose significant risks to the tech and data companies trying to build and operate these data centers, including project delays, increased costs and impeding progress in reaching sustainability targets.
For one, increased demand, combined with fewer location options and strained energy sources -- or less supply -- could spell out significant increases in energy costs. Moratoriums and suspensions due to power shortages in certain locations could also mean project delays or relocations -- more sources of cost increases.
In addition to cost, many of the bigger data and tech companies have set sustainability targets that do not align with an increasing carbon footprint. While most of these companies’ sustainability reporting has been voluntary, many will soon be required by law to provide some level of reporting on their emissions in the US and the EU, thanks to the SEC’s climate-related disclosure rule and the Corporate Sustainability Reporting Directive, respectively.
With this in mind, what can data and tech companies realistically do? Enterprises and consumers alike are all in on AI. As of late 2023, approximately one in four consumers already viewed Generative AI as very useful or indispensable to their lives. And 97% of business leaders indicated AI investment remains a priority of theirs. Putting the toothpaste back in the tube is not an option. Finding solutions to decarbonizing data centers is imperative.
Organizations dedicated to reducing the carbon footprint of their data centers have a menu of options to choose from and can implement them based on their specific needs. In most cases, taking all the above approaches should deliver the best results.
One option is procuring renewable power and locking it at a fixed price for entering into a power purchase agreement (PPA). A PPA is a long-term contract a company can sign to purchase renewable electricity at a fixed price over a given period. While there are benefits associated with this option, there can be drawbacks and hurdles depending on specific location needs. These agreements can vary substantially between localities. A PPA could be an attractive option in one location and less feasible in another. Understanding each locality's regulatory environment, market conditions and potential risks is key.
In nearly all cases, a second option -- strategic energy management (SEM) -- can be applied. SEM is an improvement framework that can help organizations save on energy and costs through monitoring, analysis, and performance improvements. Artificial intelligence and machine learning technologies can actually help companies do some of this. Leveraging ML/AI, an organization can identify ways to reduce their facilities’ emissions by working with these technologies to see which upgrades or retrofits might have the greatest potential for energy savings. Another example of SEM would be implementing new processes or norms with staff members that prioritize saving energy.
Finally, there is circularity. This pillar emphasizes the integration of circularity principles throughout the entire lifecycle and value chain of data centers. By embracing circularity, organizations can minimize waste, maximize resource efficiency, and promote sustainable practices in the operation and management of data centers.
This could take the form of proactively upgrading equipment to increase its lifespan and ensure that it’s running as efficiently as possible or using previously used or refurbished materials when it’s time to replace equipment. Some organizations are also requiring their suppliers to comply with sustainability requirements that adhere to circular principles, including having recycling programs that keep their products out of landfills, making equipment from recycled material, etc.
An AI-powered future will create tremendous opportunities, enhance productivity and fundamentally change business models as we know them. There is a lot to be excited about. Like any newer technology, its benefits will also be weighed against its risks. This means prioritizing a sustainable future now is imperative.
About the Authors
You May Also Like
2024 InformationWeek US IT Salary Report
May 29, 20242022 State of ITOps and SecOps
Jun 21, 2022