LATEST INTELLIGENCE
RETROFITTING EXISTING POWER SYSTEMS FOR AI CLUSTERS ENERGY MANAGEMENT RESEARCH CENTER
Introduction
AI workloads influence data center power systems differently than traditional IT workloads. Most data center electrical infrastructure was engineered for lower rack density, some by a factor of 10x or more
PRESENTED BY compared to the 100 + kW racks of today. Modernizing to meet evolving requirements – in a living operating data center – is a real obstacle to leveraging the benefits of AI factories.
We are still in the early stages of understanding
Download whitepaper hereihow different AI workloads impact data center power systems. Most assume that power system challenges are limited to the pretraining and posttraining( e. g., fine-tuning) of large language models( LLMs). However, within these categories, a range of variables can either increase or decrease the strain on power systems and we currently lack detailed power pro-files to quantify their effects. The rapid evolution of AI research makes supporting these workloads a moving target. For instance, newer, compute-intensive inference tasks-sometimes called“ long thinking,” may also present significant power challenges, but specific power profiles for these workloads are not yet available.
Power profiles are essential for predicting how a data center’ s power system will respond to specific AI workloads. While we may not have comprehensive profiles for every workload, we have identified five key attributes and trends that help us estimate the demands of a worst-case scenario. By designing data center power systems to accommodate these worst-case profiles, we can better verify readiness for future generations of AI workloads. p
Download whitepapers free from www. intelligentcio. com / eu / whitepapers /
22 INTELLIGENTCIO EUROPE www. intelligentcio. com