PUE
An industry-standard metric for measuring data center power efficiency, defined as the ratio of total data center power consumption to IT equipment power consumption. A PUE value closer to 1.0 indicates lower energy losses in auxiliary systems and higher overall efficiency.
Core Facts
The PUE of traditional air-cooled data centers is typically between 1.3 and 1.5, while facilities using advanced liquid cooling technology can reduce PUE to below 1.1.
The core metric for measuring data center energy efficiency is PUE (Power Usage Effectiveness), which is the ratio of total facility energy consumption to actual IT equipment energy consumption, with an ideal value of 1.0 and typically ranging from 1.2 to 1.6 in practice.
The current industry-leading data center PUE is approximately 1.1–1.2.
Timeline (last 90 days)
The PUE of advanced data centers of major cloud vendors currently generally ranges between 1.1 and 1.3.
PUE only reflects the power waste in auxiliary links such as cooling and power distribution, and does not measure the absolute power consumption of AI computing itself, so it is often criticized as a local efficiency metric that masks total growth.
Under PUE 1.30 conditions, the IT load of each 8-GPU chassis is 10.2 kW–14.2 kW, adding approximately $950–$1,450 in monthly electricity costs depending on regional industrial electricity rates.
If supporting loads such as CDUs, pumps, dry coolers, or cooling towers are not included within the boundary, the PUE number will look better, but total energy consumption may not actually decrease.
PUE is the ratio of total data center energy consumption to IT equipment energy consumption; liquid cooling moves part of the exhaust heat from the room air to the liquid loop, shifting where the energy consumption occurs from the IT side to the cooling side.
谷歌部分设施已将PUE压缩至1.1以下
Liquid cooling can compress PUE from the industry average of above 1.5 to close to 1.1.
PUE only measures ancillary losses and does not reflect the computational utilization of IT equipment itself.
The PUE of top hyperscale data centers can already be achieved below 1.1.
Two-phase immersion liquid cooling can reduce PUE to near the theoretical limit of 1.02 to 1.03.
6 more timeline events
All Facts (20)
The PUE of traditional air-cooled data centers is typically between 1.3 and 1.5, while facilities using advanced liquid cooling technology can reduce PUE to below 1.1.
80%VerifiedThe core metric for measuring data center energy efficiency is PUE (Power Usage Effectiveness), which is the ratio of total facility energy consumption to actual IT equipment energy consumption, with an ideal value of 1.0 and typically ranging from 1.2 to 1.6 in practice.
75%VerifiedThe current industry-leading data center PUE is approximately 1.1–1.2.
75%VerifiedPUE (Power Usage Effectiveness) is defined as the ratio of total energy consumption to IT equipment energy consumption, with an ideal value of 1.0.
65%VerifiedTraditional air-cooled data centers generally have a PUE above 1.8, with the industry average at approximately 1.5 to 1.6.
65%Unverified新一代液冷方案的能效比（PUE）可从传统方案的1.5以上降至接近1.1
85%UnverifiedPUE only reflects the power waste in auxiliary links such as cooling and power distribution, and does not measure the absolute power consumption of AI computing itself, so it is often criticized as a local efficiency metric that masks total growth.
50%UnverifiedThe PUE of advanced data centers of major cloud vendors currently generally ranges between 1.1 and 1.3.
50%UnverifiedUnder PUE 1.30 conditions, the IT load of each 8-GPU chassis is 10.2 kW–14.2 kW, adding approximately $950–$1,450 in monthly electricity costs depending on regional industrial electricity rates.
50%UnverifiedIf supporting loads such as CDUs, pumps, dry coolers, or cooling towers are not included within the boundary, the PUE number will look better, but total energy consumption may not actually decrease.
50%UnverifiedPUE is the ratio of total data center energy consumption to IT equipment energy consumption; liquid cooling moves part of the exhaust heat from the room air to the liquid loop, shifting where the energy consumption occurs from the IT side to the cooling side.
50%Unverified谷歌部分设施已将PUE压缩至1.1以下
50%UnverifiedLiquid cooling can compress PUE from the industry average of above 1.5 to close to 1.1.
50%UnverifiedThe PUE of top hyperscale data centers can already be achieved below 1.1.
50%UnverifiedPUE only measures ancillary losses and does not reflect the computational utilization of IT equipment itself.
50%UnverifiedTwo-phase immersion liquid cooling can reduce PUE to near the theoretical limit of 1.02 to 1.03.
50%UnverifiedImmersion liquid cooling removes heat by exploiting the thermal conductivity of liquid, which is approximately 25 times higher than that of air, and can reduce PUE to near the theoretical limit of 1.0.
50%UnverifiedThe collaboration case between DeepMind and Google data centers demonstrates that AI-driven cooling system optimization can reduce PUE (Power Usage Effectiveness) by 40%.
50%UnverifiedThe industry average PUE (Power Usage Effectiveness) for data centers is approximately 1.5 to 1.6.
50%UnverifiedThe ideal value of PUE, the core metric for measuring data center energy efficiency, is 1.0, and in reality it typically ranges between 1.2 and 1.6.
50%