Whether you are building, retrofitting, or evaluating vendors, join this technical, field-tested look at what it takes to commission AI-Ready infrastructure successfully—drawing on lessons learned from commissioning real AI workloads in engagements Uptime Institute has served as an independent consultant, so your teams can anticipate — and providing you with the risks, how to validate performance, and apply proven practices before learning them the hard way. AI is rewriting the rules of data center delivery. With rack densities reaching 40–400 kW, capacity topping 500 MW, direct liquid cooling, and GPU workloads that swing violently between idle and peak, conventional commissioning no longer goes far enough. This presentation covers how to integrate the novel technologies of an AI facility into a rigorous and comprehensive commissioning program that delivers systematic quality assurance through owner handover. We will go over the five commissioning test levels with a focus on Levels 4 and 5 and tackle the challenges unique to AI: simulating volatile GPU power profiles with purpose-built load banks, validating still-maturing liquid-cooling systems, integrating IT and facility testing under compressed timelines, and planning through to decommissioning.
