"PUE 1.08" in hero banner. CSR manager applauds; architect asks if measured in winter only, if new servers included, and if heat goes up the chimney. Two readings of same number — one marketing, one operational.
PUE (Power Usage Effectiveness) is the most cited indicator in green hosting. It compares total datacenter energy (IT + cooling + lighting + losses) to IT energy alone. Useful — provided you do not make it carry your entire project footprint.
Formula and interpretation
PUE = Total site energy ÷ IT equipment energy
| PUE | Approximate reading |
|---|---|
| 1.0 | Theoretical (impossible in practice) |
| 1.1 – 1.2 | Very efficient (free cooling, good design) |
| 1.3 – 1.5 | Industry average |
| > 1.6 | Heavy auxiliaries, difficult climate |
Documented example: Infomaniak reports about 1.085 — excellent building efficiency (ecology investigation). That does not say your WordPress consumes little.
What PUE does not say
- Energy mix — PUE 1.1 with coal ≠ PUE 1.3 with hydro.
- Scope 3 hardware — server manufacturing absent from PUE.
- Application load — cron jobs, cache, wasted queries.
- Third-party services — CDN, SaaS, analytics outside datacenter.
- Heat reuse — PUE alone ignores ERE/ERF.
Complement with Green hosting: signals that matter.
Compare two PUEs without error
Questions before comparing:
- Period — annual or summer peak only?
- Scope — one building or entire fleet?
- Climate — Nordics vs Madrid: different context (Nordic data).
- IT load — empty vs full datacenter?
- Method — continuous measurement or marketing estimate?
Comparing Infomaniak PUE in Switzerland to Spanish datacenter without climate context is false CSR comparison.
Good use in host selection
Use PUE as infrastructure filter, not final decision:
- Indicative threshold: documented PUE < 1.3 plus public carbon report.
- Reject: "green hosting" without PUE or date.
- Bonus: ERE/ERF or documented heat reuse.
Cross-check with the directory and ecology notes. Documented PUE without a public carbon report remains a partial signal — not a finished CSR decision.
The peak: perfect PUE can coexist with bad carbon balance
Here is what hero banner does not show.
Marketing often stops at PUE; CSR audit starts after.
Decide and move forward without blind spots
Before shortlisting "green" host:
- Request PUE, measurement period and scope — in writing, not slogan.
- Require energy mix and dated carbon report.
- Compare to similar climate context, not global average.
- Optimize your application in parallel — slow site wastes CPU everywhere.
- Archive documents for your own client CSR reports.
Frequently asked questions
What is PUE?
Power Usage Effectiveness compares total site energy (IT, cooling, lighting, losses) to IT equipment energy alone. A PUE of 1.2 means roughly 20% of electricity serves building auxiliaries. It is a datacenter electrical efficiency indicator — not your application code.
What PUE is good?
In 2024–2026, below 1.3 is solid and below 1.15 excellent with comparable scope and climate. The global average sits around 1.5–1.6, but comparing Stockholm to Madrid without context is misleading. Always ask measurement period and exact scope before scoring a number.
Low PUE = ecological site?
No. Low PUE describes the building, not your application CPU use, carbon mix or upstream server emissions. Bloated WordPress wastes resources in a very efficient datacenter. Complement with energy mix, carbon report and application optimization.
PUE vs ERE / ERF?
ERE and ERF measure waste heat reuse. Low PUE with heat injected into a district network is more credible than low PUE venting that heat. Read Green hosting: signals that matter to cross-check these indicators.
Next time someone cites a PUE, ask the color of electricity composing it. Number only makes sense with the mix.