SERVICE / ASSESSMENT
AI Data-Center Readiness Assessment
The brief in motion
The AI Data-Center Readiness Assessment, Explained
Most projects do not stall on chips
They stall on buildings. A facility that ran comfortably at 8kW per rack for a decade is asked, with little warning, to host racks drawing 80kW or more, and every assumption underneath it breaks at once: the cooling architecture, the electrical distribution, the floor loading, sometimes the structure itself. Capital gets committed before those limits are ever measured, and the project loses a quarter and a budget line to a problem a two-week audit would have caught for a fraction of the cost.
A structured, two-week engineering audit
The assessment is not a checklist walkthrough. It is a structured engineering audit of an existing or planned facility, measured against what AI-mission compute actually requires: rack density targets and the thermal envelope the site can support, power availability and quality at the utility and distribution levels, the realistic cooling architecture including the path from air to liquid, floor loading and physical constraints on high-density rows, network topology for tightly coupled GPU clusters, and how the deployment timeline lines up against your compute delivery schedule. Each area is measured, not estimated.
How the two weeks run
The audit follows a fixed sequence so findings arrive on schedule: site and documentation review, on-site measurement of power and thermal conditions, modeling against your target density, and a findings session before the report is finalized. Nothing in the process depends on discovery mid-engagement, because the scope is fixed before the first site visit.
- 01Intake
site documentation, power one-lines, and compute plan reviewed before the visit
- 02Site audit
power, cooling, floor loading, and network topology measured on site
- 03Modeling
findings tested against your target rack density and delivery schedule
- 04Findings report
density ceiling, ranked remediation plan, and go/no-go delivered
A report written for a decision, not a stack of readings
The deliverable is a findings report built for the person who has to decide, not the person who has to read a meter. It states the density ceiling for the site as it stands, a ranked remediation plan with cost and timeline bands for each gap, and a clear go/no-go recommendation you can take to your board. If the site cannot reach your target density at defensible cost, the report says so plainly and quantifies why, rather than leaving that conclusion for the change orders to deliver later.
Led by the engineer who has measured this before
The assessment is led by Jon Moen, founder of Vulcan AI. Jon spent twenty years building and deploying high-density computing infrastructure, including a tenure as Technical Director of EKWB USA, and has stood up more than fifty data-center deployments, among them liquid-cooled GPU systems shipped to research universities, defense customers, and Fortune 500 operators. The audit applies that operating history to your specific building, not to a generic template.
Where it starts, and where it leads
Engagements begin from $4,900, with final scope set after a short intake call covering the site, the compute plan, and the timeline. The assessment is deliberately positioned as the first step: it is the cheapest point in the entire project at which to discover a hard limit. Findings feed directly into the Liquid-Cooling Design Review, Retrofit Site Acquisition Advisory, or Deployment Oversight if the project proceeds.
Next step
Find out what your site can actually support
Send the site details and compute plan, and Jon will scope the intake call from there.