Balancing Act: The Dual Influence of AI on Data Center Power and Sustainability

Data centers face escalating challenges in managing power and improving energy efficiently. The surge in AI-driven workloads has intensified the strain on data center resources, exacerbating concerns about energy consumption and environmental sustainability. In fact, in a January 2024 report, the International Energy Agency forecasted that data centers globally may more than double their electrical consumption by 2026.

The fundamental shift that AI will play in data centers can’t be overstated. The 2024 AFCOM State of the Data Center Report notes that: “Artificial intelligence is a significant driver in how we will be building facilities moving forward. Simply put, every data center will become an AI data center… This change happened so fast that many hardly noticed. But this change is here, and it will impact your facilities.”

Ali Fenn, the president of Lancium and keynote speaker at the forthcoming Data Center World 2024, harbors a profound optimism for the role of AI in revolutionizing the data center industry, while also recognizing the formidable challenges that must be navigated on the path to progress. “AI has for many years been driving efficiency gains via such things as forecasting load shapes, weather, corresponding cooling demand… and tuning workloads and MEP systems to advance both cost and climate goals,” explains Fenn, who was recently featured in the New York Times.

Related:AFCOM: AI Boom Fueling Data Center Construction, Design Innovation

“I think the next phase is not just run time process efficiency, but also the fact that AI is now helping to enable more fundamental breakthroughs, such as via the discovery of new materials that may, in turn, catalyze innovation in battery technology, hence energy storage and the acceleration of greater renewable energy.”

Fenn goes on to note that one of the most significant opportunities for AI in the data center industry lies at the intersection of data centers and the grid. “The rapid growth of data center demand and the emergence of large-scale, gigawatt-scale data centers presents new challenges for grid operators,” explains Fenn. “At Lancium, we focus on developing AI-driven power orchestration and optimization technology to provide grid reliability and ensure workload reliability for both data centers and their customers, with SLAs that prioritize both reliability and carbon-free energy.”

Despite the many challenges facing data centers, Lancium President Ali Fenn is optimistic about the industry's ability to innovate and adapt to meet evolving demands

AI and the Future of Data Center Efficiency

Vance Peterson, a Solution Architect at Schneider Electric, further underscores the profound impact AI can have on data center operations, particularly in optimizing energy usage and reducing carbon emissions.

Related:US Energy Experts to Present Latest Insights at Data Center World 2024

“AI’s predictive capabilities can significantly contribute to the reduction of energy consumption and carbon emissions by providing insights into data center operations in relation to various external factors, such as real-time carbon content of utility supply, capacity of DERs considering weather conditions, and more,” says Peterson. “This may enable the data center industry to optimize cooling systems, facilitate predictive maintenance as opposed to preventative maintenance, and dynamically adjust power usage based on workload priority.

“Through analyzing data patterns, AI can anticipate cooling requirements, optimize airflow, and identify opportunities for energy savings, thereby reducing overall energy consumption and carbon emissions. This proactive approach will help enhance efficiency and sustainability in data center operations.”

Indeed, the State of the Data Center report identifies issues such as limitations in power and cooling, infrastructure vulnerabilities and increased carbon emissions as critical concerns that must be addressed to enhance sustainability throughout the sector. “Our sector’s expansion and critical role, along with its substantial energy needs, have necessitated a heightened emphasis on sustainable practices and the exploration of renewable energy sources… Three in four respondents (72%) plan to utilize renewable energy, including 27% who are currently doing so.”

Shaolei Ren, Associate Professor of Electrical and Computer Engineering at the University of California, Riverside, shares Fenn and Peterson’s optimism about the potential of AI to improve energy efficiency. “AI can offer more precise configuration of the cooling system operation based on real-time demand, and AI can also help predict the power usage effectiveness by offering precise configuration of cooling systems and predicting power usage effectiveness,” Ren told Data Center Knowledge.

However, Ren also acknowledges the limitations of current AI models, noting that they may not always provide accurate predictions, necessitating safeguard mechanisms.

Joe Minarik, chief operating officer at DataBank, a US-based data center operator, highlights AI’s potential for significant energy-related advantages.

“AI could help generation and transmission grids optimize loads and distribution better to meet the needs of customers overall,” Minarik said. “It can do the same inside the data center. Helping understand load, cooling, and other areas throughout the data center, AI can optimize the efficiency and balance of the power to much greater degrees of precision.”

Challenges in AI-Driven Sustainability

Ren further notes that a major concern is the challenges involved in measuring and reporting the environmental impact of AI, particularly in terms of carbon emissions and water consumption. The lack of uniform standards for reporting water consumption poses additional complexities, making it difficult to accurately assess the environmental impact of AI technologies in data centers. “While data centers routinely report their overall energy/carbon/water usage, we often lack a fine-grained measurement of the environmental impact of AI. What makes the problem even more challenging is that not all AI models are running as a standalone service. Some AI models are just used as part of another service (e.g., AI models are used in recommendation engines), so it's non-trivial to accurately get the environmental impact of specific AI models.”

Peterson likewise emphasizes the dual nature of AI’s influence, highlighting its capacity to optimize energy usage while also acknowledging concerns regarding increased energy consumption.

“Some people in the industry predict that accelerated computing, which is the ‘enabler’ of the AI revolution, will allow us to do more with less as it relates to data center infrastructure,” Peterson said. “While accelerated compute will increase individual rack density, the overall number of racks within a data center may decrease significantly. In other words, accelerated computing may allow us to do more with less. Overall, it’s essential to consider the broader impact of AI on energy consumption and the environment while striving to leverage its capabilities for sustainable solutions.”

Soaring Demand

Fenn also notes that projections regarding data center energy demand and the “electrify everything” trend vary. She does, however, believe that one thing is certain: demand is increasing at a faster pace and to higher levels than anyone previously anticipated.

“The fundamental question is: can renewable generation come online in the right places, on the right timelines, and can storage scale cost-effectively to smooth the intermittency of renewable sources?” asks Fenn.

“If yes, emissions can be held constant and continue the positive trajectory of recent years. This has to be our collective industry goal, and it is possible. It requires taking an energy-first approach to data center siting, and having an appetite for the opportunity tied to innovation in workload flexibility. It is our collective imperative to do this the right way using the multiple tools we have at our disposal.”

Overall, despite the many challenges facing data centers with the advent of AI, Fenn is optimistic. “I am a firm believer that AI is a net positive for the world, and this is the most exciting time to be alive, but it is incumbent upon us as data center industry leaders, to make sure that we, the gateway to the AI opportunity, deliver it responsibly.”

Fenn is due to tackle the key issue of data center power and AI during her keynote speech at this year’s Data Center World.