Water use and cooling capacity have moved from engineering footnotes to central site-selection criteria for any organization evaluating data center space. The convergence of AI workloads, denser compute architectures, and regional water stress has forced real estate and infrastructure teams to treat utility infrastructure with the same analytical rigor they would apply to lease economics or zoning compliance.
Why Cooling Constraints Define Operational Risk
Data center operators have long measured performance in terms of power delivery and network connectivity. Cooling infrastructure, by contrast, was treated as a building system to be inherited rather than interrogated. That assumption no longer holds in markets where air-cooled facilities are approaching their thermal limits and where municipalities are issuing water withdrawal permits with meaningful conditions attached.
The shift is partly driven by compute density. A rack configured for general server workloads might draw three to five kilowatts. A rack purpose-built for large-language-model training or inference can draw thirty to fifty kilowatts or more. The cooling system designed for the former cannot serve the latter without significant mechanical intervention, and that intervention may be constrained by local water availability.
Understanding cooling constraints is not simply a facilities question. It is a strategic real estate question that belongs inside the site evaluation framework from the first day a requirement is defined. Organizations that defer this analysis until after a letter of intent is executed routinely discover constraints that renegotiate the economics of the deal or disqualify the site entirely.
The Mechanics of Water-Cooled Data Center Infrastructure
Cooling towers, evaporative condensers, and chilled-water loops all consume water, but they consume it in meaningfully different ways. A cooling tower rejects heat by evaporating a portion of its circulating water into the atmosphere. That evaporated volume is called evaporative loss, and it represents the primary consumptive use — water that leaves the local watershed permanently rather than returning via discharge.
Makeup water replenishes evaporative loss plus blowdown, which is the intentional discharge of a portion of the circulating water to prevent mineral concentration buildup. The ratio of evaporative loss to total water withdrawal defines the consumptive fraction. Facilities with high cycles of concentration — meaning the system recirculates water many times before blowdown — consume a higher fraction of what they draw, which matters to regulators counting net withdrawals against an aquifer or river basin.
Liquid cooling systems, including direct-to-chip and immersion configurations, also require water or other coolants, though the consumptive profile differs. A closed-loop direct-to-chip system can operate with dramatically lower evaporative loss than an evaporative tower, particularly when the rejected heat is transferred to a dry cooler or a facility chiller plant. Understanding which cooling modality a facility uses — and what its water intensity profile looks like under peak load — is foundational to evaluating operational risk.
Power Usage Effectiveness, or PUE, is a commonly cited efficiency metric, but it does not capture water consumption. Water Usage Effectiveness, or WUE, measures the volume of water consumed per unit of IT energy delivered, expressed in liters per kilowatt-hour. Organizations should request WUE data alongside PUE when evaluating facilities, understanding that WUE varies seasonally in climates where economizer modes are used.
Mapping Regional Water Availability and Regulatory Exposure
The local watershed context shapes every other aspect of a cooling evaluation. A facility drawing from a water-stressed aquifer in an arid region faces a materially different risk profile than a facility in a water-abundant river basin with established discharge rights. The first step in regional mapping is identifying the water source: municipal supply, groundwater well, surface water diversion, or recycled water.
Municipal water supply introduces dependency on the utility's own infrastructure resilience and rate-setting authority. A municipality experiencing population growth or drought conditions may impose water-use restrictions, rate increases, or allocation caps that directly affect data center operations. Reviewing the utility's long-range water supply plan and understanding its drought contingency protocols is a necessary diligence step.
Groundwater users face a different set of exposures. Where state agencies manage groundwater through permit systems, a facility's extraction rights may be capped, transferable, or subject to curtailment during declared drought emergencies. Some jurisdictions require metering and reporting at the well level, creating a regulatory compliance obligation that the occupier inherits through the lease or service agreement.
Surface water diversion rights operate under state-law doctrines that vary significantly across the country. In prior-appropriation states, the age of a facility's water right determines its priority during shortage events. Junior rights can be curtailed entirely while senior rights continue. This legal structure should be investigated as part of any site evaluation in markets governed by prior appropriation.
Recycled water is increasingly available in markets where utilities have invested in reclaimed water infrastructure. Facilities that connect to recycled water for cooling tower makeup can materially reduce their draw on potable supply, which both lowers operational cost and reduces regulatory exposure in water-stressed markets. Availability, pressure, and water quality specifications for recycled water should all be confirmed during site diligence.
The Site Evaluation Scoring Framework
Answering the question of How Data Center Users Should Evaluate Water Use and Cooling Constraints requires a structured scoring methodology, not an ad hoc checklist. The framework should weight cooling and water criteria alongside power, connectivity, seismic, and economic factors within a consistent scoring matrix so that trade-offs across candidate sites can be compared objectively.
The first scoring dimension is water source security: the reliability, legal protections, and redundancy of the facility's water supply. A facility served by a single municipal connection with no on-site storage receives a lower score than one with dual-source supply and significant buffer storage. Sites with access to recycled water should receive additional credit where drought risk is high.
The second dimension is water intensity, measured through WUE across the annual operating cycle. A facility reporting a WUE of 0.5 liters per kilowatt-hour in its documented disclosures represents a meaningfully lower water risk than one reporting 2.0 liters per kilowatt-hour. Where the operator has not disclosed WUE, that absence of transparency is itself a scoring input.
The third dimension is cooling scalability: whether the facility's cooling infrastructure can support the occupier's projected load growth without requiring material capital investment or additional water permits. A facility operating near the rated capacity of its chiller plant with no additional cooling capacity under construction offers limited runway. Teams should model hypothetical growth scenarios — labelled as such — to understand at what point a facility's cooling headroom becomes the binding constraint.
Conducting the Technical Diligence Walk
A scoring matrix built from disclosed data must be supplemented by a physical inspection. The technical diligence walk for cooling and water infrastructure should follow a defined protocol that covers the mechanical, electrical, and water systems in sequence.
Begin at the cooling tower or dry cooler yard. Count the installed units, confirm the rated capacity of each, and identify how many are actively serving the existing load. The gap between installed capacity and active load is the facility's thermal headroom. Document whether expansion pads exist for additional cooling equipment and confirm that permits for such expansion have been approved or applied for.
Move to the makeup water connection and storage infrastructure. Where is the connection point, and what is the rated flow? Is there an on-site storage tank, and what is its volume? For facilities in markets with high drought risk, storage that can sustain operations for forty-eight to seventy-two hours without an external supply connection is a materially different risk posture than a facility with no on-site storage.
Inspect the chemical treatment system for the cooling tower circuit. The sophistication of water treatment directly affects cycles of concentration and therefore consumptive use. Facilities operating advanced conductivity-based controllers with real-time blowdown management consume less water per unit of heat rejected than facilities relying on time-based controls. This operational detail rarely appears in marketing materials but is visible during a walk-through.
Review the blowdown discharge pathway. Where does cooling tower blowdown go? Facilities in jurisdictions with pretreatment requirements for discharge to municipal sewer must maintain compliance with the local industrial pretreatment program. Non-compliance with discharge permits creates a regulatory liability that can affect lease continuity.
Evaluating Liquid Cooling Readiness for Dense Compute
As rack densities continue to escalate, the question of liquid cooling readiness is inseparable from the water evaluation. A facility built exclusively for air cooling cannot physically accommodate direct-to-chip or immersion configurations without significant structural and mechanical modification. Organizations planning to deploy high-density AI or high-performance computing workloads must assess liquid cooling infrastructure as a primary site criterion.
Direct-to-chip cooling delivers coolant directly to the processor, capturing heat at the source. The facility must supply chilled water or a facility coolant loop to each cabinet, which requires in-row or above-floor distribution manifolds and appropriate pressure and temperature setpoints. Confirming that the facility's secondary distribution piping is sized for the target rack density is a non-trivial engineering task that belongs in the diligence scope.
Immersion cooling — in which servers are submerged in dielectric fluid — introduces a different set of facility requirements. The dielectric fluid, while not water, must still be managed for spill containment, fire suppression compatibility, and disposal. Some facilities have invested in immersion-ready infrastructure; others have not. The distinction is determinative for certain workload types.
Where a facility does support liquid cooling, review the fluid-to-water heat exchanger specifications and the cooling distribution unit ratings against the occupier's projected load. Confirm that the facility's chiller plant setpoints are compatible with the required supply temperature for the specific cooling technology the occupier intends to deploy.
Negotiating Water and Cooling Provisions in the Lease or Service Agreement
The evaluation process culminates in translating findings into contractual protections. Standard wholesale colocation agreements and retail colocation contracts are not drafted with water risk in mind; the occupier must negotiate specific provisions if any protection is desired.
Power delivery metrics are routinely specified in data center agreements with uptime guarantees and credits. Cooling delivery metrics are less commonly specified. Where the occupier's load requires maintained supply-water temperatures within a defined range, that specification should appear in the service level agreement rather than be treated as an implicit obligation. Deviation from the specified parameters, if it results in workload degradation or hardware risk, should trigger a defined remediation and credit process.
Water-source change notifications are rarely addressed in standard agreements. If the facility intends to shift from potable municipal supply to recycled water, or to change cooling modalities, the occupier may have legitimate operational interests in advance notice. A thirty- or sixty-day notification covenant is a reasonable request and may be granted without significant negotiation if raised during the initial term sheet phase rather than during the formal lease negotiation.
Permit compliance representations should be included as a facility warranty. The operator should represent and warrant that it holds all permits required for the cooling and water operations as described, and that no permit amendments are pending that would materially reduce its operational capacity. A breach of this warranty, discovered post-execution, then has a defined contractual remedy rather than leaving the occupier in a pure tort context.
Expansion rights in data center agreements typically address power and space. Where cooling headroom is the occupier's binding constraint, expansion option language should reference the facility's available cooling capacity as a condition precedent to exercising the option. This prevents the operator from allocating cooling headroom to other tenants during the option period.
Integrating Environmental and Sustainability Commitments
Many organizations now carry public environmental commitments that extend to their data center supply chains. Water stewardship targets — pledges to operate at net-zero water withdrawal, to match water consumption with watershed replenishment credits, or to use exclusively recycled water by a target date — must be evaluated against the facility's actual water profile.
Water stewardship credits are an emerging mechanism analogous to renewable energy certificates. They represent investment in watershed restoration or conservation projects designed to offset consumptive water use. Some facility operators are beginning to offer these as part of a sustainability supplement to their service agreement, though the crediting methodologies are not yet standardized across jurisdictions.
Where an organization has made public commitments, the gap between those commitments and the actual water profile of a candidate facility represents a reputational and reporting risk. ESG reporting frameworks increasingly require disclosure of water withdrawal and consumption data for Scope 3 activities, which include leased data center infrastructure. Teams should confirm what consumption data the facility will make available for reporting purposes before executing an agreement.
Certifications such as LEED for Data Centers or ENERGY STAR for Data Centers capture some water-efficiency criteria, though neither is primarily designed around water stewardship. The Alliance for Water Stewardship standard provides a more water-specific framework, and some operators have begun pursuing it in markets where water risk is a recognized business issue. Certification status should be confirmed directly with the certifying body, not taken at face value from marketing materials.
Managing Decisions and Maintaining Continuity Across the Evaluation
A data center evaluation involving water and cooling diligence produces a substantial body of technical findings, permit documents, scoring analyses, and negotiating positions.
Permit documents obtained from one facility should be stored in a way that makes retrieval straightforward when negotiation counterparties raise questions months later. Without deliberate document management, findings that took days to assemble can become inaccessible at the moment they are most needed.
If a WUE benchmark was used as a scoring input during site selection, actual performance against that benchmark should be measured once operations begin.
The site evaluation process is not a discrete event that ends at lease execution; it is the beginning of a monitoring relationship that continues for the duration of the occupancy.
Building a Repeatable Evaluation Process
Each element should have a named owner and a defined review schedule so the template stays current with regulatory changes and emerging cooling technologies.
This continuity is not achievable through email threads alone.
Over time, that institutional knowledge becomes the organization's most valuable site-selection asset.
The discipline required to close that feedback loop — comparing initial findings against realized performance — is the same discipline that separates organizations with genuine site-selection capability from those that treat each evaluation as a standalone exercise. Building that capability requires process, tools, and a commitment to documentation that extends past the lease execution date and into the operational period where the real-world consequences of the initial analysis become visible.
About Advantai
Advantai is a commercial real estate intelligence and operations platform operated by ADVANTAGE AI LLC, a Delaware limited liability company. It connects client relationships, property research, documents and financial decisions in one workspace for commercial real estate teams — advisers and brokerage teams, occupier and facility teams, and portfolio teams. The platform covers CRM and origination, requirements and site selection, Property X-Ray (an interactive 3D building workspace), financial modeling and comparison, document intelligence, transactions and diligence, client collaboration, and portfolio strategy with critical dates. The optional Super Agent upgrade adds specialist, source-backed research and automated scenario analysis.
Get Started with Advantai
Ready to see your next move clearly? Go to advantaico.com, click Request a demo and tell us about your next project. Prefer to start with a single project? Visit advantaico.com/getting-started to plan your first one.