Greater demands for AI workloads require greater amounts of data center power

With AI operations, efficiency isn’t just about how much energy you consume. Rather, it’s about how much compute you produce from it, making output a more meaningful measure of performance

Two data centers. Same power. Same PUE – very different outcomes

If we want to avoid futile arguments like these, we need better IT carbon footprint measures

Can we hit and greenhouse gas reduction targets for 2030?

PUE is a simple metric, but now we've reached levels of efficiency where it could become counter-productive

How to avoid common mistakes in measuring PUE

To get real efficiency figures, you need accurate measurements

The recession stopped the green data center movement in its tracks. Now it’s coming back, older and wiser, and without any new metrics

Chasing efficiency can make your facility less reliable, unless you can quickly move down the Universal Learning Curve

We hear data centers in the city of Beijing have to have a PUE of 1.5 or less. How will that work? 

In response to the news that ASHRAE has abandoned plans to specify efficiency levels for data centres

Small and medium sized data centers can calculate PUE without metering beyond the UPS

ASHRAE’s new data center building standard includes recommended PUE levels

Why change the Power Usage Effectiveness (PUE) metric? 

Giant data centers have an easier time with PUE, says Peter Judge