Yokogawa Calls for More Granular Power Measurement in AI and HPC Data Centers
Редакция ЦОД:Главное
Yokogawa Test & Measurement says AI and HPC workloads are making data-center power demand more dynamic, reducing the usefulness of measurement approaches based on long averaging periods. The company argues that operators need more detailed, synchronized visibility into electrical parameters to manage fast-changing loads.
- Yokogawa argues that AI and HPC workloads require more granular and synchronized electrical measurement in data centers.
- The company says PUE and DCiE do not capture power quality, losses or transient events.
- Sub-millisecond load changes, high current crest factors and harmonics are highlighted as relevant electrical characteristics.
- Potential demand growth for advanced power monitoring is an inference, not evidence of deployments or commercial traction.
Yokogawa Test & Measurement has outlined the case for more granular electrical measurement in data centers serving AI and HPC workloads. In commentary published by Data Centre Magazine, Kees Oosterhoff, Expert Support Manager in Europe at Yokogawa Test & Measurement, said these workloads have fundamentally changed the nature of data-center energy consumption, making it more dynamic.
The company’s position is that conventional averaged approaches to power measurement and benchmarking are insufficient for these environments. PUE and DCiE remain useful metrics for understanding overall consumption and benchmarking sustainability performance, Yokogawa said, but they do not explain power quality, losses or transient events.
The article identifies sub-millisecond load changes, high current crest factors and increased harmonics as relevant electrical characteristics of AI and HPC environments. According to Yokogawa, these conditions require more granular and synchronized measurement of power-system parameters, rather than relying solely on longer-term averages.
For operators, the issue is infrastructure visibility. Monitoring short-duration transitions and power-quality indicators may be relevant to maintaining infrastructure availability where rapidly changing loads are present. The shift toward AI and HPC workloads could expand demand for high-frequency power monitoring, power-quality management and transient-diagnostics tools, but this is an analytical inference rather than evidence of Yokogawa sales, contracts or deployments.
The publication is an expert commentary and does not announce a specific data-center project, product launch, customer implementation, contract, capacity figure, investment or delivery timetable.
AI and HPC environments can introduce short-duration load transitions and power-quality effects that may not be visible through conventional efficiency benchmarking alone. This could increase the operational importance of high-precision power measurement, monitoring and transient analysis, although the publication provides no evidence of specific deployments or market adoption.
- Data Centre Magazine / 24 сентября 2026 г. · АнглийскийПервоисточник →
Материал подготовлен редакцией ЦОД: Главное на основе указанных первоисточников.