Data centers are facing increasing challenges in managing power and improving energy efficiency. The proliferation of AI-driven workloads is increasing the strain on data center resources, exacerbating concerns about energy consumption and environmental sustainability. In fact, the International Energy Agency predicted in its January 2024 report that the power consumption of data centers around the world could more than double by 2026.
It is difficult to overstate the fundamental change that AI will bring to the data center. In his 2024 AFCOM State of the Data Center Report: Simply put, every data center will become an AI data center… This change happened so quickly that many people barely noticed. But this change is here and it will affect the facility. ”
Ali Fenn, president of Lancium and keynote speaker at the upcoming Data Center World 2024, has deep optimism about the role of AI in revolutionizing the data center industry, and at the same time believes in the path to progress. We also recognize the challenges. “AI has been driving efficiency gains for years through things like predicting load shapes, weather, and corresponding cooling demands, and adjusting workloads and their MEP systems to advance both cost and climate goals. ” explains Fenn, who was recently featured. new york times.
“The next step is not just about process efficiency at runtime, but also about the fact that AI is helping us realize more fundamental breakthroughs, such as discovering new materials that have the potential to accelerate innovation in battery technology. I think there is: Accelerating the expansion of energy storage and renewable energy.”
Fenn went on to say that one of the most important opportunities for AI in the data center industry is at the intersection of the data center and the grid. “The rapid growth in data center demand and the emergence of large gigawatt data centers are creating new challenges for grid operators,” Fenn explains. “At Lancium, our AI-driven power orchestration system delivers grid reliability to both data centers and their customers, ensuring workload reliability, with SLAs that prioritize both reliability and carbon-free energy.” We are focused on developing optimization and optimization technologies.”
The future of AI and data center efficiency
Vance Peterson, Solutions Architect at Schneider Electric, emphasizes that AI can have a significant impact on data center operations, especially when it comes to optimizing energy use and reducing carbon emissions.
“AI's predictive capabilities can reduce energy consumption and carbon emissions by providing insights into data center operations in relation to a variety of external factors, such as the real-time carbon content of public supplies, DER capacity taking into account weather conditions, etc. We can make a huge contribution to volume reduction,'” says Peterson. “This has the potential for the data center industry to optimize cooling systems, facilitate predictive rather than preventive maintenance, and dynamically adjust power usage based on workload priorities.
“Through analysis of data patterns, AI can reduce overall energy consumption and carbon emissions by predicting cooling requirements, optimizing airflow, and identifying energy savings opportunities. This approach helps increase the efficiency and sustainability of data center operations.”
Indeed, the State of Data Centers report finds that issues such as power and cooling limitations, infrastructure vulnerabilities, and rising carbon emissions are critical concerns that must be addressed to increase the sustainability of the entire sector. has been identified as a matter. “The expansion and critical role of our sector, and its substantial energy demands, requires an increased emphasis on sustainable practices and the exploration of renewable energy sources…Three in four respondents (72 %) plan to use renewable energy, of which 27% are currently doing so.”
Xiaolei Ren, associate professor of electrical and computer engineering at the University of California, Riverside, shares Feng and Peterson's optimism about AI's potential to improve energy efficiency. “AI can provide a more accurate configuration of cooling system operation based on real-time demand. AI can also improve power usage efficiency by providing a more precise configuration of the cooling system and predicting power usage efficiency. It also helps in predicting the future,” Ren said. Data center knowledge.
However, Ren also acknowledged the limitations of current AI models, noting that they do not always provide accurate predictions and that safeguard mechanisms are needed.
Challenges in AI-driven sustainability
Ren also points out that a major concern is the challenges associated with measuring and reporting AI's environmental impact, particularly carbon emissions and water consumption. The lack of uniform standards for reporting water consumption introduces additional complexity and makes it difficult to accurately assess the environmental impact of AI technologies in data centers. “While data centers regularly report their overall energy/carbon/water usage, detailed measurements of the environmental impact of AI are often lacking. Making the problem even more difficult This is because not all AI models are run as standalone services; some AI models are only used as part of another service (e.g., an AI model is used to make recommendations). It is not easy to know exactly the environmental impact of a particular AI model (used in the AI engine).
Peterson similarly emphasizes the dual nature of AI's impact, acknowledging concerns about increased energy consumption while emphasizing AI's ability to optimize energy use.
“Some in the industry predict that accelerated computing, the ‘enabler’ of the AI revolution, will enable us to do more with less when it comes to data center infrastructure. ,” Peterson said. “Although faster computing increases the density of individual racks, the overall number of racks in a data center can decrease significantly. In other words, accelerated computing allows you to do more with less effort. It has the potential to do a lot of things. Overall, it is essential to consider AI's broader impact on energy consumption and the environment while striving to leverage its capabilities into sustainable solutions. ”
Rapid increase in demand
Fenn also points out that predictions about data center energy demand and the “electrification of everything” trend vary. But she believes one thing is certain. That means demand is increasing faster and at higher levels than anyone previously expected.
“The fundamental questions are: Can renewable generation be brought online in the right places and on the right schedule? And can storage be cost-effectively expanded to smooth out the intermittency of renewable energy sources? “Can we do that?” Fenn asked.
“If so, emissions can remain constant and continue the positive trajectory of recent years. This must be an industry-wide goal, and it is possible. We need to take an energy-first approach to the It is our collective responsibility.”
Overall, despite the many challenges facing data centers with the advent of AI, Feng is optimistic. “I strongly believe that AI is a real positive for the world. These are the most exciting times to live in, but as leaders in the data center industry, we believe that this is the gateway to AI. We have a duty to make sure that we do. If we have the opportunity, let us take responsibility and deliver.”
Feng will address the important issues of data center power and AI in his keynote address at this year's Data Center World.