How to Choose a Site for a Data Center: Power, Water, and Connectivity

The right data-center site is the one where required power can be delivered on the project schedule, the cooling design has a dependable and legally available water strategy (or a viable low-water alternative), and telecommunications routes are physically diverse with capacity to expand. A nearby substation, water main, or fiber route is only a starting clue. Before buying or optioning land, obtain written, site-specific evidence from the utility, water authorities, carriers, engineers, and permitting agencies, then compare expansion, hazards, approvals, community impacts, and whole-life cost against the intended workload.

Start with the workload, not the parcel

Site selection cannot be separated from the facility you intend to build. A modest enterprise room, a hyperscale campus, and a high-density artificial-intelligence cluster impose different electrical, cooling, network, land, and latency requirements. Define the operating case before asking whether a parcel is suitable.

#	Preview	Product	Price
1		Spanish -- Thermal Guidelines for Data Processing Environments, 4th ed.		Buy on Amazon
2		Thermal Guidelines for Data Processing Environments by TC9.9 MISSION CRITICAL FACILITIES...	$147.83	Buy on Amazon

As an Amazon Associate I earn from qualifying purchases.

Document the load and growth case

• State the initial and ultimate IT load in megawatts, the target energization date, and each planned expansion phase.
• Describe the load shape: steady baseload, rapid ramping, seasonal variation, or flexible demand that can be curtailed.
• Set rack-density and cooling assumptions, including any liquid-cooling requirement.
• Specify availability and redundancy objectives for utility feeds, substations, generators, cooling equipment, and telecommunications.
• Record latency, cloud-interconnection, internet-exchange, and private-network requirements for the users and applications.
• Identify sustainability objectives, reporting obligations, water restrictions, emissions limits, and community commitments that will affect design.

These inputs determine which site constraints are decisive. There is no universal answer to “What power capacity does a data center need?” or “How many fiber connections are required?” The answer follows from the workload, cooling architecture, reliability design, jurisdiction, and expansion plan.

Separate requirements from preferences

Classify each criterion as a fatal constraint, a scored preference, or an item that can be mitigated. For example, a utility delivery date later than the business launch may be fatal; a longer drive to an airport may be a preference; and a flood exposure might be mitigated only if an engineering study and insurance review support the proposed measures. This prevents an attractive low land price from masking an impossible energization schedule.

#1 Best Overall

Spanish -- Thermal Guidelines for Data Processing Environments, 4th ed.

Buy on Amazon

Power: prove deliverability and schedule

Electricity is usually the first feasibility gate. Assess the capacity that can actually be delivered at the required voltage and date, not the capacity that appears on a map or the existence of a nearby substation. ASHRAE’s site-planning guidance advises evaluating available capacity, substation proximity, utility expansion plans, interconnection timing, and lead times for critical equipment. OCS similarly distinguishes scalable capacity from simple proximity to a substation.

Get a utility-backed capacity statement

Ask the serving utility for a written, project-specific response that identifies:

• Firm capacity available to the site now and at each expansion phase.
• Delivery voltage, point of interconnection, and the proposed number and arrangement of feeds.
• Required transmission, substation, feeder, transformer, protection, metering, and switchgear work.
• Which party funds, owns, permits, procures, and constructs every upgrade.
• Interconnection studies, queue dependencies, approval gates, outage requirements, and milestone dates.
• Expected equipment lead times, especially for transformers and switchgear.
• Tariff structure, demand charges, connection costs, and any curtailment or flexibility obligations.
• Whether the commitment is firm, conditional on network upgrades, subject to a queue, or merely an indicative planning figure.

Require the utility to state assumptions and expiry dates. A preliminary load letter is not the same as a binding connection agreement. If the utility cannot commit capacity by the required date, treat the schedule as unverified rather than assuming that construction can proceed while the grid catches up.

Test the interconnection schedule against construction

Interconnection work may take longer than the building itself. Compare each utility milestone with land closing, design release, procurement, commissioning, and the required go-live date. Include time for environmental reviews, rights-of-way, easements, outages, factory testing, transport, installation, and energization. ASHRAE identifies interconnection timelines and transformer and switchgear lead times as primary schedule inputs. Put a named owner and a documentary deliverable against every milestone.

Design for reliability and expansion

Confirm how the proposed feeds behave during a line, transformer, substation, or maintenance outage. A site with two circuits may still share a pole line, duct bank, substation bus, or right-of-way. Request a one-line diagram showing genuine electrical separation and the operating assumptions behind the reliability claim. OCS recommends assessing dual physical power paths for higher reliability; that is industry guidance, not a universal design code.

Reserve space and electrical pathways for later phases. The ultimate load may require a second substation, additional transformers, larger switchgear rooms, new feeders, or a separate utility entrance. Check that easements, setbacks, fire separation, noise limits, and local zoning allow those assets to be built without relocating the first phase.

Compare grid strategies rather than assuming one answer

EPRI’s Speed to Power resource frames four strategies for comparison:

These are options to engineer and compare, not interchangeable products. A bridge arrangement can move the opening date forward while adding permitting, emissions, fuel, and operating complexity. An islanded design changes the electrical architecture and long-term cost base. The utility, electrical engineer, authorities, and community must validate the selected approach.

Water and cooling: evaluate the complete resource chain

Water requirements depend on climate, IT density, heat-rejection equipment, operating hours, water quality, and the balance between electricity and water use. The reviewed guidance does not establish a universal litres-per-megawatt or gallons-per-megawatt threshold. A credible site comparison therefore starts with the proposed cooling design and its design-day and annual demand profile.

Match cooling architecture to the workload and climate

Ask the mechanical engineer to document the cooling concept for the actual rack densities and weather conditions. The analysis should identify whether the design uses evaporative or adiabatic cooling, cooling towers, dry or hybrid heat rejection, chilled water, direct-to-chip liquid cooling, immersion, or a combination. For each mode, record peak make-up water, annual consumption, blowdown, chemical treatment, electricity use, temperature limits, and performance during extreme heat.

ASHRAE advises balancing energy efficiency with responsible water use, particularly in water-stressed regions. A design that minimizes electricity through evaporation may increase water demand; a dry system may reduce water use while increasing electrical load or reducing performance on hot days. Compare those effects for the local climate and the operator’s sustainability commitments instead of selecting a technology by name alone.

Verify supply, rights, and drought performance

Obtain confirmation from the water provider or authority of the source, available flow, pressure, water quality, connection point, allocation, price, and contract term. Test the design against peak-day demand, drought restrictions, seasonal limits, conservation rules, service interruptions, and changes in water quality. If reclaimed, industrial, or rainwater is proposed, confirm treatment requirements, reliability, legal entitlement, storage, and backup supply.

Ask who pays for new mains, treatment capacity, storage, pumps, and off-site reinforcement. Infrastructure NSW’s policy principles illustrate how a jurisdiction can expect projects to fund additional water and energy supply and address cost recovery. Those principles apply in New South Wales; they are not a rule for every location.

Close the wastewater and stormwater loop

Water diligence is incomplete without an outlet. Confirm sewer capacity, discharge temperature and chemistry limits, treatment availability, blowdown handling, and the authority’s ability to accept the proposed peak and annual flows. Coordinate cooling discharge with stormwater design so that a heavy-rain event, tank failure, or treatment outage does not create an unpermitted release. Map drainage, detention, flood levels, easements, and maintenance access before fixing equipment locations.

Plan for interruption and changing policy

Define the response to drought orders, water-quality excursions, pressure loss, a failed pump, or a temporary prohibition on evaporative equipment. A credible contingency may include stored water, alternate supply, dry-mode operation, reduced IT load, or a different cooling sequence. Quantify how long each measure lasts and what workload or efficiency penalty it creates. Confirm that the operating permits and community commitments allow the contingency to be used.

Connectivity: verify physical diversity, not provider counts

Network resilience comes from separated physical paths, not from the number of logos on a carrier list. Two providers may lease the same duct, bridge, pole line, building entrance, or metropolitan backhaul. Ask each carrier for route maps and construction records, then have an independent designer reconcile them against surveyed streets, easements, rail crossings, waterways, and common points of failure.

Map the entrance and route topology

• Identify every proposed campus entrance, meet-me room, duct bank, and point of presence.
• Trace each route to the carrier network and interconnection location, including shared manholes, bridges, utility corridors, and buildings.
• Require evidence of physical separation from the property boundary to the carrier’s diverse upstream paths.
• Check duct occupancy, spare conduits, cable pulling limits, repair access, and the ability to add routes during later phases.
• Confirm whether route diversity survives a single road closure, construction project, flood, fire, or utility-corridor failure.

An India-focused OCS white paper reports that its customers ask for four diverse paths from multiple providers and recommends fiber-rich locations with duct capacity for expansion. That is an observed market practice in that context, not a global minimum. The correct number of paths depends on the workload, service-level objective, carrier topology, and failure analysis.

Check latency and interconnection needs

Measure end-to-end latency to the users, cloud regions, exchanges, content networks, and private peers that matter to the workload. Confirm packet-loss, jitter, maintenance-notice, cross-connect, and service-restoration commitments. A site can have abundant long-haul fiber yet fail a latency requirement because the nearest useful point of presence is elsewhere. Reserve meet-me-room space, cable trays, and secure maintenance access for additional carriers and cross-connects.

Land, hazards, approvals, and community fit

Power, water, and fiber do not make a parcel buildable by themselves. Evaluate the physical and regulatory envelope before spending on detailed design.

Rank #2

Thermal Guidelines for Data Processing Environments by TC9.9 MISSION CRITICAL FACILITIES (2004-06-30)

$147.83

Buy on Amazon

Land and logistics

• Confirm usable area, slope, bearing capacity, contamination history, and the shape of expansion parcels.
• Map utility easements, rights-of-way, setbacks, fire access, security standoff, drainage corridors, and protected features.
• Test access for heavy transformers, generators, cooling equipment, cranes, and emergency vehicles.
• Provide more than one approach road where practical and verify bridge, turning-radius, weight, and seasonal restrictions.
• Reserve land for substations, fuel systems, water treatment, cooling yards, stormwater detention, and future buildings.

Natural and environmental hazards

Commission location-specific studies for flood and stormwater exposure, extreme wind, wildfire, seismic conditions, lightning, heat, air quality, soil and groundwater contamination, and any other hazards relevant to the region. The study should address both the building pad and the off-site utility routes. A safe building can still lose service when a feeder, road, water main, or fiber corridor crosses a hazard zone.

Zoning, permits, and approvals

Build a permit matrix covering land use, site development, building, electrical, mechanical, water, sewer, stormwater, air emissions, fuel storage, noise, fire protection, environmental review, and grid interconnection. Assign each approval to an authority, applicant, prerequisite, fee, hearing, and target date. Mahesh Trivedi, Strategic Advisor to OCS Group India, said: "With more than 30 odd statutory and compliance requirements; speed of approval is a crucial element that determines the “Go Live” timetable of the project." The “30 odd” phrasing is his observation, not a universally validated count; the practical lesson is to identify every approval early and track dependencies.

Workforce and community impacts

Assess access to skilled construction, operations, electrical, mechanical, and security workers; training capacity; housing and transport; and emergency-response capability. Engage local authorities and affected neighbors about construction traffic, noise, backup-generation emissions, water demand, visual effects, and jobs. Infrastructure NSW’s published principles call for environmental and efficiency standards, no net cost to consumers and communities, funding of additional supply, improved local infrastructure and amenity, support for future industries, and workforce skills. Those are jurisdiction-specific expectations, not a universal legal test.

Build an evidence-backed site scorecard

Use one scorecard for every candidate and mark each item as confirmed, conditional, or unverified. Record the evidence date, assumptions, responsible party, next action, and consequence if the item fails. The following fields synthesize the planning factors identified by ASHRAE, EPRI, OCS, and Infrastructure NSW.

Do not average away a fatal uncertainty. A low-cost parcel with unconfirmed power delivery should remain a conditional option until the utility resolves the connection, not receive a high score because its land price is attractive.

Compare candidates with workload-specific weighting

Rank sites on power deliverability and schedule; water and cooling feasibility; network diversity and latency; expansion potential; hazards and resilience; approvals and community acceptance; and whole-life cost. Weight those axes for the workload. A latency-sensitive service may place connectivity first, while a water-intensive cooling design in a drought-prone region may make water the gating criterion. A project with a fixed launch date may value schedule certainty over a modest land-cost difference.

For every favorable claim, require a document: a utility letter rather than a map pin, a water allocation or service confirmation rather than a nearby main, a surveyed route plan rather than a carrier count, and a permit schedule rather than an informal planning conversation. Keep unresolved items visible through acquisition, design freeze, financial close, and construction.

A practical diligence sequence

1. Define the operating case. Freeze the load, ramp, rack density, cooling modes, redundancy, latency, sustainability, and launch assumptions used to test every parcel.
2. Screen fatal constraints. Remove sites with no plausible utility path, prohibited land use, unacceptable hazard exposure, impossible access, or no legal water and wastewater solution.
3. Request utility and carrier evidence. Obtain capacity, voltage, milestones, upgrade scope, route maps, points of presence, duct data, and expansion terms in writing.
4. Develop a concept plan. Place buildings, substations, generators, cooling equipment, water and wastewater systems, roads, security, stormwater, and future phases on the actual parcel.
5. Commission targeted studies. Complete electrical, water, cooling, geotechnical, flood, environmental, traffic, acoustic, emissions, and network-route investigations proportionate to the decision.
6. Build the scorecard and risk register. Mark each fact confirmed, conditional, or unverified; assign an owner, due date, mitigation, and financial or schedule exposure.
7. Engage authorities and neighbors early. Validate zoning, approvals, infrastructure obligations, workforce plans, and community expectations before committing to an irreversible layout.
8. Stress-test failure cases. Model delayed grid upgrades, drought restrictions, water-quality failure, fiber-route loss, extreme weather, road closure, and construction-phase outages.
9. Make the land decision with conditions. Tie closing, exclusivity, or major design spend to measurable utility, water, route, title, hazard, and approval milestones.
10. Revalidate before construction. Reconfirm volatile tariffs, lead times, utility schedules, water rules, carrier routes, permits, and equipment availability at design freeze and procurement.

Common mistakes and how to correct them

Choosing a substation on a map

Failure: The parcel is marketed as power-rich because a substation is nearby.

Correction: Obtain the utility’s available capacity, required upgrades, voltage, interconnection queue position, and delivery date. Treat proximity as a screening clue only.

Counting carriers instead of routes

Failure: Several providers are present, but their cables share one corridor or building entrance.

Correction: Trace physical paths, common points, points of presence, duct occupancy, and repair access. Require independent failure analysis.

Using an annual water estimate without a peak case

Failure: Annual consumption looks manageable, but design-day demand exceeds mains, treatment, or sewer capacity.

Correction: Test peak heat, drought restrictions, water quality, blowdown, storage, alternate modes, and wastewater discharge with the authorities.

Leaving approvals until after design

Failure: A preferred layout conflicts with zoning, emissions, noise, environmental, or infrastructure-funding requirements.

Correction: Create the permit matrix at site-screening stage and let approval dependencies shape the concept plan.

Scoring uncertain facts as favorable

Failure: An indicative promise is treated as committed capacity or a route is assumed diverse without proof.

Correction: Keep the item conditional or unverified, name the responsible confirmer, and attach a go/no-go condition to the investment decision.

What the guidance does—and does not—establish

ASHRAE’s site-planning guidance states: "Prioritize power availability and grid capacity early in site selection, as utility constraints and interconnection timelines are now primary drivers of feasibility and schedule risk." That principle supports early utility diligence, but it does not set a universal megawatt threshold, approval duration, water allowance, or fiber count.

EPRI, ASHRAE, OCS, and Infrastructure NSW describe interacting siting factors rather than one universal score or numerical threshold. The legal and numeric details in the examples are jurisdiction-specific: Infrastructure NSW reflects New South Wales policy, and the OCS material is focused on India. Utility capacity, water rights, tariffs, fiber routes, construction dates, and permits change by site and over time. Confirm them with the responsible local providers and authorities; none of the guidance by itself is a parcel-specific feasibility determination or engineering certification.

Quick Recap

Bestseller No. 1

Spanish -- Thermal Guidelines for Data Processing Environments, 4th ed.

Bestseller No. 2

Thermal Guidelines for Data Processing Environments by TC9.9 MISSION CRITICAL FACILITIES (2004-06-30)

$147.83

Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.