Geothermal Cooling Offers Data Center Power Boost
Data centers, increasingly vital for powering artificial intelligence (AI) applications, are facing a critical bottleneck: electricity supply. Projections indicate that by 2028, data centers in the United States could consume as much as 12 percent of the nation's total electricity. A significant portion of this energy consumption, approximately 30 percent, is dedicated to cooling the vast infrastructure within these facilities. A new study from the University of Chicago proposes a novel approach to alleviate this strain by integrating emerging geothermal technology, known as Cold Underground Thermal Energy Storage (Cold UTES), with existing data center cooling systems.
Cold UTES operates by storing cold water underground during periods of low energy demand, or off-peak hours. This stored cold water is then retrieved to meet the peak cooling requirements of data centers when needed. Unlike conventional battery storage, Cold UTES has the capacity to maintain its stored cooling potential for extended durations, potentially lasting for weeks or even an entire season. This capability addresses a key challenge in data center operations: ensuring consistent and efficient cooling without overwhelming power resources.
Testing the Cold UTES Concept in Diverse Climates
To evaluate the viability of this integrated approach, researchers at the University of Chicago, led by Andrew Chien, the William Eckhardt Distinguished Service Professor of Computer Science, and co-author Wedan Emmanuel Gnigba, developed models for data centers equipped with Cold UTES. These models were applied to three distinct major U.S. markets: Virginia, Arizona, and Texas. The selection of these regions was strategic, chosen for their varied climatic conditions and differing electricity grid characteristics, allowing for a comprehensive assessment of the technology's performance under diverse environmental pressures.
The results of the modeling indicated a substantial increase in available IT power capacity across the three test sites, averaging 9.8 percent. The most pronounced improvements were observed in Arizona's arid and hot climate, where the integration of Cold UTES led to an 11 percent rise in capacity. This suggests that the technology is particularly effective in regions with high cooling demands, which are often the most challenging for data center operations.
Broader Implications for AI Growth and Grid Stability
Professor Chien highlighted the primary advantage of Cold UTES: its ability to enable data centers to more effectively utilize their existing grid interconnection capacity. This, in turn, reduces the overall demand placed on the electrical grid, thereby facilitating the operation of more Graphics Processing Units (GPUs) essential for AI computations. He noted that with the substantial investments, often exceeding $10 billion, required for AI data centers, the demonstrated 11 percent increase in capacity could translate into savings of billions of dollars. The flexible cooling capabilities offered by Cold UTES represent a significant advancement in the design and operation of data centers.
Beyond increasing IT power capacity, the study also found that the technology enhances cooling efficiency and alleviates pressure on space-constrained equipment, such as rooftop dry coolers. This improvement in efficiency can extend the operational lifespan of existing facilities, allowing them to scale their operations for longer periods before physical limitations necessitate expansion or new construction. The research is supported by the U.S. Department of Energy's Geothermal Technologies Office and involves collaboration with institutions including the National Laboratory of the Rockies, Lawrence Berkeley National Laboratory, and Princeton University.
Rethinking Data Center Design for Energy Efficiency
The escalating power demands driven by AI necessitate a fundamental rethinking of the energy infrastructure that supports it. The Economic Times emphasizes that expanding data center capacity without concurrently strengthening energy generation, transmission, storage, and grid resilience risks further straining systems already under pressure. Therefore, energy capacity and efficiency must be integral to AI's growth strategy, not an afterthought once computing infrastructure is in place. This requires a holistic approach to data center design, where factors such as location, power supply, cooling mechanisms, and the surrounding energy network are considered in unison.
The focus is shifting from merely increasing capacity to creating facilities that can scale efficiently and function harmoniously within a broader energy ecosystem. While much attention in the AI conversation centers on chips, models, and computing power, the physical environment supporting these elements—requiring immense electricity, heat dissipation, uptime maintenance, and accommodating dense computing—is equally critical. The design of data centers is increasingly being viewed not just as a housing for computing infrastructure, but as a crucial energy asset that must be optimized for power availability, heat management, and cooling demands.
Innovations in Water and Energy Production
In a related development, a new green technology from the Chinese Academy of Sciences offers a simultaneous solution for producing green hydrogen fuel and fresh drinking water directly from ocean water. Traditional methods of producing green hydrogen using electrolyzers face significant challenges when attempting to use raw seawater, as the corrosive salt and minerals can quickly damage equipment. This necessitates extensive pre-purification of the ocean water, a process that is energy-intensive and generates hazardous concentrated brine.
The breakthrough technology bypasses this purification hurdle by utilizing the waste heat generated during the hydrogen production process to drive an integrated water distillation cycle. This closed-loop system captures thermal energy that would otherwise be vented, using it to evaporate and purify incoming seawater. By the time the water reaches the electrodes, it is already purified, resulting in high-purity hydrogen gas and a surplus of fresh drinking water. This innovation holds promise for coastal regions facing dual demands for zero-emission energy sources and reliable freshwater supplies, utilizing standard industrial components for scalability and commercial viability.
What This Means for the Future
The convergence of these technological advancements suggests a significant shift in how data centers will be designed and operated in the era of AI. The Cold UTES technology directly addresses the critical power and cooling challenges that are currently limiting the expansion of AI infrastructure. By leveraging underground thermal energy storage, data centers can potentially increase their computing capacity without a proportional increase in grid power consumption, thereby easing pressure on electrical networks and potentially reducing the need for costly new power plant construction. This also translates to significant cost savings for data center operators.
Furthermore, the broader emphasis on integrated design, as highlighted by The Economic Times, indicates a move towards data centers that are not just consumers of energy but active participants in the energy ecosystem. This includes incorporating renewable energy sources, advanced energy management systems, and efficient cooling strategies. The parallel innovation in seawater processing highlights the potential for synergistic solutions, where waste heat from one process can be utilized for another, further enhancing overall efficiency and sustainability. These developments collectively point towards a future where data center growth is more closely aligned with energy and resource availability, enabling more sustainable and cost-effective scaling of AI capabilities.
💬 Comments (0)