Uptime Institute

AI in Practice Paper: Technical Vendor Requirements and Evaluation

As AI workloads push data centers to unprecedented power densities and operational complexity, technology and vendor decisions have become some of the most critical and riskiest choices organizations face.

To support informed decision‑making, Uptime Institute has published a new AI in Practice Paper 2: Technical Vendor Requirements and Evaluation. This paper is the second in Uptime Institute’s five‑part series examining how to plan, deliver, commission, and operate AI‑ready digital infrastructure.

The paper outlines a structured, vendor‑neutral approach to defining requirements, developing RFPs, and evaluating technologies for AI deployments, where immature solutions, supply constraints, and rapidly changing designs can lead to long‑term operational and resiliency risks.

Key topics include:

  • Developing technical requirements aligned with AI workload and business objectives
  • Evaluating power and cooling technologies for high‑density GPU environments
  • Assessing direct‑to‑chip, immersion, and hybrid cooling solutions
  • Structuring RFPs and bid evaluations to support transparent, defensible decisions
  • Future‑proofing technology selections to reduce retrofit risk


Grounded in Uptime Institute’s global experience with high‑density and mission‑critical facilities, the paper provides practical guidance for owners, operators, designers, and developers responsible for AI infrastructure investments.

The guidance in the paper is backed by AI Infrastructure Advisory — Uptime Institute's independent, vendor-neutral practice. Through design reviews, technical validation, and project lifecycle support, we help organizations apply principles outlined in this paper to projects, reducing execution risk and grounding AI infrastructure decisions on unparalleled insight experience and expertise.

The paper is available now on Uptime Institute’s website.

Download the AI in Practice Paper 2: Technical Vendor Requirements and Evaluation paper

More information about AI Infrastructure Advisory and the full set of the five‑paper series is also available through the same page.

Download the paper