PUE and the metrics that matter

PUE: Power Usage Effectiveness

PUE = total facility energy ÷ IT equipment energy

If servers use 1,000 kWh and the whole site (IT + cooling + lighting + losses) uses 1,400 kWh, then PUE = 1.4. The 0.4 is the "overhead" spent on keeping the IT running.

PUE	What it suggests
1.0	Theoretical perfect — every kWh goes to IT
1.1–1.3	Very efficient modern or hyperscale design
~1.5	Around the industry average in global surveys
2.0+	Older or inefficient facility

Why hot climates find it harder: cooling is the biggest overhead, so sites in the Gulf use strategies like efficient chillers, district cooling, containment and higher supply temperatures to keep PUE down.

Other metrics worth knowing

• WUE (Water Usage Effectiveness): litres of water per kWh of IT energy — important where water is scarce.
• IT load (MW): the power available to servers; how capacity is sold.
• Rack density (kW per rack): traditional racks use roughly 5–10 kW; AI racks can exceed 100 kW, which changes cooling completely.
• Availability: the share of time a service is up; operators track incidents and their root causes.
• Carbon intensity: emissions per unit of energy, driven by where the power comes from.

Interview tip: be able to calculate PUE from two numbers and explain one way to improve it (for example, raising supply-air temperature or adding containment).

Remember

• PUE = total ÷ IT. Lower is better; 1.0 is the floor.
• Watch WUE and rack density too — AI is pushing both.