Assessing an Industrial Site for an AI Data Center

This guide helps owners and project teams decide whether an industrial site can support an AI data center, and what to investigate before committing to it. Start with a site plan, available utility information, and a rough estimate of the facility’s power and cooling needs; suitability depends on verified capacity and constraints, not the site’s industrial zoning alone.

What You’ll Need

• A current boundary survey, site plan, and information about existing buildings and infrastructure.
• Utility bills, electrical single-line diagrams, and written information from the electric utility about service and expansion options.
• A preliminary data-center program: target IT load, phasing, rack density, redundancy objectives, and cooling approach.
• Records of environmental permits, site history, soil and groundwater conditions, flood risk, and known easements or restrictions.
• Access to electrical, civil, mechanical, environmental, planning, and telecommunications specialists as the review advances.

Step-by-Step

1. Define the intended facility. Set an initial IT load in megawatts, number of phases, expected rack density, and target operating date. Distinguish IT load—the energy used by computing equipment—from the larger facility load, which also includes cooling, power conversion, lighting, and other systems. Ask the design team to estimate both, with assumptions and a range rather than a single unsupported figure. AI hardware can create high rack-level loads, so a site that once served ordinary warehousing may need major electrical and cooling upgrades.
2. Check land use and development constraints. Confirm permitted uses with the relevant planning authority; do not infer permission from an industrial designation. Review height limits, setbacks, noise rules, hours of construction, landscape requirements, and any conditions attached to prior approvals. Map property boundaries, access rights, utility easements, pipelines, rail lines, and neighboring sensitive uses. Ask whether the project needs a zoning change, conditional approval, or public review, and include that schedule risk in the assessment.
3. Verify power availability with the utility. Request a formal feasibility or service study for the proposed load and phasing. Ask about deliverable capacity, connection voltage, substations and feeders, upgrades required, who is responsible for them, queue position, expected milestones, and contingency plans. Existing transformers or a large historic utility bill do not prove that additional capacity is available. Confirm whether the utility’s estimate is preliminary or a binding commitment, and identify any approvals or equipment lead times that could control the schedule.
4. Assess electrical resilience and layout. Have an electrical engineer compare the site’s incoming service with the project’s redundancy design and outage tolerance. Check space and routing for substations, switchgear, backup generation or other permitted backup systems, energy storage if proposed, and fuel or emissions-control equipment. Review fire access, equipment clearances, grounding, drainage, and separation from occupied areas. Determine whether construction can be staged while existing operations continue. Do not treat a generator concept as an alternative to grid capacity without checking permits, fuel logistics, operating limits, and the intended operating strategy.
5. Establish a credible cooling and water plan. Ask mechanical engineers to model heat rejection for the selected IT load, local climate, and expected equipment density. Compare air cooling, liquid-assisted approaches, and heat-rejection options for energy use, water demand, noise, footprint, and maintenance. Get written confirmation of water and sewer availability if the design needs them; confirm quality, pressure, discharge limits, and seasonal restrictions with the providers. A low-water or closed-loop design still needs a heat-rejection strategy and may have other resource requirements. Identify whether heat reuse is feasible, but do not count it as a benefit without a practical nearby user and compatible timing.
6. Screen hazards and ground conditions. Review flood maps and drainage history, wildfire exposure where relevant, extreme heat, wind, seismic conditions, and nearby industrial hazards. Commission a properly scoped environmental site assessment and geotechnical investigation before relying on the land’s prior use. Look for contamination, buried tanks, fill, unstable soils, groundwater issues, and demolition materials that could affect cost or permits. Ask specialists to explain uncertainty and recommend further testing; a desktop screen is not a substitute for fieldwork when a concern is identified.
7. Confirm fiber and network diversity. Contact multiple carriers and request route information, service options, capacity, installation schedule, and points of presence. Verify whether proposed routes are physically diverse; two contracts may share a conduit, bridge, or upstream facility. Check rights of way and whether a new duct route or meet-me room is practical. For AI workloads, confirm that the planned network design can support the required data movement and external connectivity rather than assuming any fiber connection will suffice.
8. Test the site plan and construction logistics. Fit buildings, cooling equipment, electrical yards, roads, loading, security setbacks, stormwater measures, and utility corridors onto a surveyed plan. Test truck access, turning, crane setup, laydown areas, and construction sequencing. Check whether existing structures can realistically be reused after structural, fire, and hazardous-material reviews. Compare a reuse option with a new-build layout, including demolition and temporary access needs.
9. Build a risk-adjusted feasibility decision. Create a register of constraints, evidence, owner, next action, cost range, and schedule effect. Separate confirmed facts from assumptions and unresolved questions. Set explicit decision gates—for example, utility confirmation, planning pathway, environmental findings, and network diversity—before land purchase or major design spend. Obtain independent specialist reviews of high-impact assumptions, then compare the site with alternatives using the same criteria. Proceed only when the remaining risks have owners and a credible resolution path.

Common Mistakes

• Equating industrial zoning or past heavy power use with permission and capacity for a data center.
• Using IT load as though it were the total facility demand, or ignoring future phases and peak cooling needs.
• Accepting verbal utility assurances without understanding study status, upgrades, responsibilities, and timing.
• Counting two fiber providers as diverse without checking their physical routes and upstream dependencies.
• Underestimating water, noise, emissions, stormwater, contamination, or community review constraints.
• Buying or designing around a single optimistic schedule before key investigations are complete.

Tips

• Use one consistent assumptions sheet for power, cooling, water, phasing, and redundancy across all consultants.
• Ask providers for written responses and mark each item as confirmed, conditional, or unverified.
• Model a lower and higher demand case so the site is not judged only against one load estimate.
• Map utility corridors and fiber routes early; a small easement or access issue can change the layout.
• Keep a contingency allowance in both schedule and cost planning until surveys, studies, and provider commitments firm up.

Troubleshooting

The utility cannot confirm capacity: Ask what study or upgrade would resolve the uncertainty, what information is missing, and when a decision can be made. Evaluate phased deployment or a lower initial load, but verify that those changes actually fit the utility’s constraints.

Water or discharge is constrained: Have the mechanical designer compare lower-water cooling options and quantify their energy, space, and operating trade-offs. Confirm any discharge or water-source proposal with the relevant provider and authorities before treating it as viable.

Environmental testing finds a concern: Pause assumptions about construction cost and schedule. Engage qualified environmental professionals to define additional sampling, remediation options, approvals, and responsibility. Do not disturb suspect materials without appropriate assessment and controls.

Planning feedback is unfavorable: Request specific written reasons and determine whether redesign, conditions, a different operating approach, or another site is the practical answer. Recheck noise, generator, traffic, and visual impacts instead of assuming a revised application will resolve them.

Fiber diversity is unavailable: Ask carriers to identify alternate entry points and routes, and assess the cost and permissions for new pathways. If no resilient route is practical, treat that as a major suitability limitation rather than a detail to solve after construction.

Frequently Asked Questions

Does an industrial site automatically qualify for a data center?
No. Zoning, utility service, environmental conditions, building rules, and project-specific approvals must all be checked.

How much power does an AI data center need?
There is no single standard figure. Demand depends on the IT load, rack density, cooling design, power systems, redundancy, and planned growth. Have engineers calculate a range for the specific program.

Can an existing factory’s electrical connection be reused?
Possibly, but its present rating and condition do not establish spare capacity or suitability. The utility and electrical engineers must verify service, equipment, upgrades, and the proposed operating design.

Is access to water essential?
Not for every cooling design, but all facilities need a way to remove heat. Compare water needs and other resource demands for the actual design, then confirm service and discharge conditions.

How early should site investigations begin?
Begin desktop reviews and provider discussions before acquisition or major design commitments. Order field studies early enough to expose constraints while alternatives are still available.

What is the clearest reason to reject a site?
A fatal constraint with no credible, permitted, and schedulable remedy—such as unavailable power at the required scale or an unresolvable land-use restriction—is a strong reason to stop or reassess the project.

Related

• Oblivion Remastered vs Skyrim: Which to Play?
• Recover Deleted Insta360 Photos: Full Guide
• Castlevania Belmont’s Curse: Where to Go After the Fallen