The Rise of AI Data Centers: Innovations in Cooling and Energy Management

In July 2025, Preferred Networks, Inc. (PFN), Internet Initiative Japan Inc. (IIJ), and Japan Advanced Institute of Science and Technology (JAIST) commenced pilot testing an ultra-high-efficiency AI accelerator system at IIJ’s Matsue Data Center Park and JAIST’s Ishikawa Campus. This initiative is part of a project funded by Japan’s New Energy and Industrial Technology Development Organization (NEDO) aimed at enhancing post-5G information and communication systems. The testing involves 30 nodes of PFN's MN-Core 2 servers utilizing a direct liquid cooling system, a shift from traditional air-cooling methods, to address the significant heat generated by high-density processors.

PFN, IIJ and JAIST Start Pilot Testing Liquid-Cooled High-Density AI Servers and Hybrid-Cooled Data Centers for Ultra-High-Efficiency AI Computing InfrastructureBuilding the Backbone of the AI Era: Samsung C&T’s Data Center Business TeamTurning Down the Heat: Building AI That Runs Cooler and Smarter

IIJ has introduced Air-Assisted Liquid Cooling (AALC) technology to create a hybrid cooling environment, which combines air and liquid cooling systems. This approach aims to enhance the efficiency and operational feasibility of liquid-cooled servers, addressing challenges such as plumbing and water leakage. The collaboration between PFN, IIJ, and JAIST will also explore coordinated operations between water and liquid cooling systems to optimize data center efficiency.

PFN, IIJ and JAIST Start Pilot Testing Liquid-Cooled High-Density AI Servers and Hybrid-Cooled Data Centers for Ultra-High-Efficiency AI Computing Infrastructure

The surge in demand for AI capabilities has led to a significant increase in energy requirements for data centers. For instance, Avista Corp. reports that requests from technology companies for energy to support AI data centers have exceeded 3,000 megawatts, a substantial increase compared to the utility's typical demand of under 2,000 megawatts. This shift is expected to triple regional energy needs, prompting utilities to develop new contract structures to ensure reliability and affordability while meeting clean energy commitments.

Power, cooling, carbon: Spokane firms brace for AI's energy demands

As AI-driven workloads intensify, traditional air-cooling systems are proving inadequate. Companies like McKinstry Co. are responding by scaling up production of modular cooling systems designed for large-scale data centers. Liquid cooling systems, which circulate chilled water directly to server chips, are becoming increasingly prevalent due to their efficiency in managing heat generated by advanced processors.

PFN, IIJ and JAIST Start Pilot Testing Liquid-Cooled High-Density AI Servers and Hybrid-Cooled Data Centers for Ultra-High-Efficiency AI Computing Infrastructure

AI data centers not only consume vast amounts of electricity but also significantly impact local water resources. Direct water use for cooling can strain municipal systems, particularly in regions facing water shortages. Reports indicate that two-thirds of U.S. data centers built since 2022 are located in high water-stress areas. To mitigate these impacts, data centers are exploring alternative cooling methods, such as zero-water designs and thermal energy storage systems.

AI Water Usage: Tackling Data Center Water StrainAI’s Ballooning Energy Consumption Puts Spotlight on Data Center Efficiency

The collaboration between Capstone Green Energy and Microgrids 4 AI aims to develop grid-independent microgrids for AI data centers, integrating advanced cooling technologies with localized power generation. This initiative reflects a broader trend toward creating sustainable and scalable infrastructure capable of supporting the growing demands of AI applications.

Going Grid-Independent to Satisfy AI Power and Cooling: Capstone Signs Microgrid MoUAI’s Ballooning Energy Consumption Puts Spotlight on Data Center EfficiencyTurning Down the Heat: Building AI That Runs Cooler and Smarter

As the AI landscape evolves, the infrastructure supporting it must adapt to meet increasing demands for energy and cooling. Innovations in liquid cooling and energy management are crucial for ensuring that data centers can operate efficiently while minimizing their environmental impact. The ongoing developments in this sector highlight the importance of collaboration among technology providers, utilities, and research institutions to create a sustainable future for AI infrastructure.

AI’s Ballooning Energy Consumption Puts Spotlight on Data Center Efficiency