AI Data Center Efficiency: Looking Beyond Power Usage Effectiveness

This voice experience is generated by AI. Learn more.This voice experience is generated by AI. Learn more.Jae-cheol Lyu is President and CEO of LG Electronics, a global innovator in technology and consumer electronics.

gettyAI’s advancement is reshaping data center requirements, from the scale of computation they must support to the way facilities are designed and operated. As AI training and inference require more high-performance GPUs, data centers face rising rack power density and heat generation, along with changing demands for power, water and space.

Power usage effectiveness (PUE) has long been the most widely used metric for evaluating data center energy efficiency, but it has limitations across different climates and resource conditions. The Uptime Institute noted that PUE is well suited to tracking changes in a facility’s efficiency over time but that factors such as climate, resilience and water use need to be evaluated separately.

To properly assess data center efficiency in the AI era, we therefore need to move one step beyond PUE. We also need to look at three things: how flexibly a system can keep pace with rapidly evolving hardware, how efficiently it can move heat generated from the chip to beyond the facility and how reliably it can operate under local power, water and climate conditions.

According to the same Uptime Institute survey, the share of operators running high-density racks of 30 kilowatts or more rose from 19% in 2025 to 24% in 2026, while roughly one in seven now operates racks of 50 kilowatts or more. Growing AI and GPU workloads are driving higher power densities, increasing both electricity demand and the amount of heat that cooling systems must remove.

More importantly, AI hardware evolves on much shorter cycles, whereas data center infrastructure is built for long-term use. Thus, cooling systems must be designed to adapt to fast-changing AI hardware. McKinsey researchers note that cooling technology selection has long-term implications for facility design, layout and construction costs, as well as future workloads and operating expenses. Designing IT, power and cooling as a coordinated system is becoming increasingly important to overall facility performance, rather than optimizing each one independently.

For cooling providers, this changes the emphasis from fixed-specification equipment toward ongoing engineering support. Systems need to accommodate evolving AI hardware and site conditions, with real-world performance validation informing subsequent improvements.​

Traditional data centers mainly use air cooling for servers.​ By contrast, AI data centers using direct-to-chip (DTC) liquid cooling connect cold plates, CDUs, piping, pumps, heat exchangers and chillers or dry coolers into a single thermal management system, making the heat-transfer path longer and more complex.​

In the same report, McKinsey analysts observe the growing importance of piping, CDUs, thermal monitoring and system integration as liquid cooling expands. Individual components like a single CDU or chiller don’t determine overall performance; piping pressure losses, pump power consumption, heat-exchange temperature differences and control methods also affect energy use.

These considerations matter, as power availability increasingly constrains data center expansion. The International Energy Agency (IEA) projects that global data center electricity consumption will more than double, from 415 terawatt-hours in 2024 to roughly 945 terawatt-hours by 2030. Reducing cooling energy use can do more than lower operating costs; where power capacity is limited, it can free up electricity for GPUs and AI computation.​

High temperatures, humidity and water constraints complicate cooling decisions, while data center resource consumption affects electricity rates, grid investment and community acceptance. These pressures extend beyond ESG reporting to long-term costs, business permitting and the social license to operate.

Public concern and water stress illustrate the challenge. In a June 2026 Reuters/Ipsos (paywall) report, 57% of U.S. respondents opposed a data center being built locally, and 77% were concerned that AI growth could increase electricity rates. Separately, S&P Global reported that 43% of 9,055 data centers surveyed are located in regions experiencing high water stress.

Reducing these burdens without compromising required cooling performance calls for industrywide advances in energy efficiency. Although cooling alone doesn’t determine environmental impact, it provides opportunities for technical improvements.

In regions facing high heat, high humidity and water-supply constraints, easing the burden on power grids and local communities while maintaining necessary cooling performance calls for technologies that fit local conditions. Regional suitability, rather than the efficiency of any single technology, should guide system design, with the goal of supporting sustainable operation over the long term.​

AI hardware evolves faster than the facilities that support it, making adaptability essential to long-term competitiveness. Efficiency assessments must therefore look beyond PUE to consider how readily cooling systems can be adjusted and revalidated for new hardware, how much energy they use to transfer heat from chips to chillers and how reliably they operate under regional power, water and climate conditions.

This changes how we design and operate data centers. Rather than optimizing IT, power and cooling solutions separately, we need to bring computing equipment, power infrastructure and cooling systems together as a single, integrated system centered on AI computing performance.

Ultimately, the competitiveness of next-generation AI infrastructure will be determined by how much computing performance can be sustainably delivered within the constraints of limited power, water and space. Within this coordinated approach, cooling is central to both computing capacity and resource efficiency.​

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?