COLUMBUS, Ohio – October 14, 2026 -- Vertiv Holdings Co. (NYSE: VRT) and the Deloitte Center for Integrated Research have introduced a collaboration built around a guide that sorts enterprise AI workloads into six archetypes and maps each to one of four physical infrastructure models.
The Vertiv Designing Infrastructure for Enterprise AI Guide was developed by Vertiv and STL Partners, a digital infrastructure research firm. It is aimed at business, IT and facilities leaders who must decide where AI workloads should run and whether available infrastructure can support them before they commit to an infrastructure plan.
Four-step process links business outcomes to reference designs
The guide sets out four steps: define the business outcome and AI workload; evaluate the workload's scale, performance and operating requirements; determine the deployment location; and match the workload and location to a physical infrastructure model and reference design.
The research finds that infrastructure readiness can determine where workloads can be deployed and how quickly they reach production. Available power, cooling, connectivity and operational capacity decide whether a workload fits an existing facility or needs a different deployment model.
Six workload archetypes map to four cluster models
The six archetypes cover model development, retraining and four categories of production use. Each workload can then be aligned with one of four infrastructure models: Edge AI Cluster, Small AI Cluster, Large AI Cluster or High-Density AI Cluster.
The models serve as a starting point for decisions on power, cooling, controls, connectivity, resilience, operational support and future growth. Reference designs can then be adapted based on site assessment, workload requirements and detailed engineering.
Guide treats enterprise AI as a portfolio, not a single cloud decision
Exploratory workloads may remain in the cloud, according to the guide. Sustained or business-critical workloads may favor hybrid infrastructure or enterprise-managed capacity. Workloads needing faster response times, greater control, data proximity or specific model performance may need to run at the edge or in another controlled environment.
Public cloud, colocation, enterprise-owned capacity, hybrid environments and edge infrastructure each have a role. The guide does not argue broadly for or against the cloud and instead offers one method for choosing the operating environment for each workload.
Facility readiness assessment targets early constraints
Across all four models, the guide has enterprises check whether a proposed location has enough power, cooling, connectivity, operational capacity and room for growth. Running this assessment early can surface constraints before they block deployment.
Vertiv CEO Giordano Albertazzi said starting with the workload lets enterprises make infrastructure decisions earlier and identify constraints sooner.
Nicholas Merizzi, principal and Silicon2Service U.S. Lead at Deloitte Consulting LLP, said the economics and operating requirements of AI change as workloads move into production. He said organizations need to weigh business value, governance, operating ownership and infrastructure together.
The full guide also includes a four-phase roadmap for planning, deploying and evolving enterprise AI infrastructure.