Summary
- T-Systems reports PUE of 1.35 at its Cerdanyola del Vallès data centre.
- Reported WUE fell from 0.0430 L/kWh in 2025 to 0.0143 L/kWh in 2026 data available to date.
- Changes include revised temperature setpoints, evaporative-cooling optimisation, and on-site solar generation.
T-Systems says it has reduced power usage effectiveness at its Cerdanyola del Vallès data centre near Barcelona to 1.35 after changing cooling setpoints and operating practices at the facility.
The company also reports a substantial decline in water usage effectiveness, from an average 0.0430 litres per kilowatt-hour in 2025 to 0.0143 L/kWh in the 2026 data available to date. Both figures are company-reported operational metrics rather than independently audited sector benchmarks.
Changes at the site include adjustments to operating temperatures and optimisation of evaporative water cooling used to improve chiller performance during hotter conditions. T-Systems has also installed 305 solar panels, which it estimates will generate approximately 244,000kWh of electricity annually.
The numbers are useful because they show how efficiency improvements can come from plant operation rather than a wholesale replacement of the data-centre cooling architecture. Raising or optimising temperature setpoints can reduce mechanical cooling demand where IT equipment and environmental tolerances allow it, while controls can determine when evaporative assistance produces enough energy benefit to justify additional water use.
That interaction between PUE and WUE is increasingly important. A lower PUE does not automatically indicate a lower environmental impact if the efficiency gain is achieved by consuming substantially more water, particularly in regions exposed to heat and water stress. The reverse is also true: reducing water use can increase mechanical cooling energy under some operating conditions.
DataCentral has previously examined why PUE and WUE need to be read with information about climate, load, grid mix, cooling technology, and measurement boundaries. The Cerdanyola figures provide a site-level example of that problem. T-Systems reports improvements in both metrics, but the value of the comparison depends on the monitoring period, utilisation, weather conditions, and consistency of the underlying calculation.
The Barcelona region adds another layer. Southern European data-centre growth is attracting more investment as developers look beyond established northern hubs, but warmer summer conditions place different demands on cooling systems. Operators cannot simply copy thermal assumptions from Nordic or northern European facilities and expect the same plant performance.
The site’s solar installation is small relative to the continuous electrical demand of a major data centre, but it can offset part of the facility’s auxiliary or daytime electricity use. Its greater relevance is as one component of an operating programme alongside changes to cooling control, temperature strategy, and water management.
T-Systems’ reported results therefore provide a useful operations story rather than evidence that one efficiency metric has solved the cooling problem. The practical question is whether the reductions persist across full-year weather conditions and changing IT utilisation while maintaining the resilience expected from a production data centre.
