Data Center Water Use Requirements: Cooling, Power and Community Risk

Water is becoming a front-end development constraint for AI data center campuses.

Data center water use requirements cover the water a facility consumes directly for cooling and the water consumed indirectly to generate the electricity that powers it. For AI campuses, both numbers matter. The project may buy utility power, but local communities still see the combined burden: power demand, water demand, land use and infrastructure strain.

That is why water has moved from sustainability appendix to site selection gate.

The clearest recent signal is Texas. A May 2026 University of Texas at Austin white paper estimated that data centers could account for 3% to 9% of Texas water use by 2040, up from less than 1% today. The researchers also noted that more than 400 data centers are operating or under construction in Texas. Those numbers make water diligence a development issue, not an ESG talking point.

Start with the cooling concept

A developer cannot evaluate water demand without knowing the cooling strategy. The same MW load can produce very different water requirements depending on system design.

The early diligence package should define:

1. Air cooling, liquid cooling or hybrid approach

2. Evaporative cooling assumptions

3. Chilled-water plant concept

4. Heat rejection equipment

5. Expected annual water use

6. Peak day water demand

7. Makeup water quality requirements

8. Wastewater and blowdown handling

AI workloads make this harder because rack densities are rising. A conventional enterprise facility and a GPU-heavy AI facility may have similar building envelopes but very different heat profiles. The cooling system is not a late design choice. It affects site area, utility service, permitting, community acceptance and operating cost.

AI can help compare cooling options against climate data, water availability and projected load. Engineers still have to select the system. The model can organize the evidence, but it cannot certify the design.

Model indirect water use from power

Direct facility water use is only part of the picture. The UT Austin report explicitly includes water used to produce the power needed by data centers. That matters because power generation can be water intensive, especially where thermal generation is part of the supply mix.

For developers, the diligence question is not simply, 'Can the municipality serve our cooling demand?' The better question is, 'What water burden does this campus create across the local infrastructure stack?'

The site team should ask:

• What generation mix is likely to serve the load?

• Are local power plants water constrained?

• Is new generation part of the power strategy?

• Does onsite generation require water?

• Will water stress affect future power reliability?

• Are water and power approvals reviewed by different agencies?

This is where power and water diligence have to be joined. A site can look attractive because power is theoretically available, but if the implied generation path creates water risk, the public approval story becomes fragile.

Check municipal capacity before the public process

Municipal water capacity is often presented as a yes-or-no utility question. That is too shallow.

A data center developer needs to understand:

1. Current treatment capacity

2. Existing peak demand

3. Planned population growth

4. Drought restrictions

5. Industrial water allocations

6. Wastewater treatment capacity

7. Capital improvement plans

8. Whether new infrastructure is required

The answer may be 'yes, with upgrades'. That is not the same as yes. Upgrades mean cost, schedule, political exposure and sometimes a public approval process.

The project team should also check whether the water source is potable water, reclaimed water, groundwater, surface water or private supply. Each source creates different permitting, quality and resilience questions. Reclaimed water may reduce political pressure but introduce treatment and availability constraints. Groundwater may work technically but trigger community resistance in stressed basins.

Price community risk as a development constraint

Water is emotional because it is visible. Residents may not understand PUE, redundancy or utility tariff structures. They understand wells, low pressure, drought restrictions and who gets priority in a dry year.

That makes water a community-risk issue. The diligence team should map:

• Nearby residential areas

• Agricultural users

• existing drought plans

• Recent water restrictions

• Local press coverage

• Water authority meeting minutes

• Public objections to industrial growth

• Tribal, environmental or watershed stakeholders

The goal is not to bury the issue. The goal is to know whether the site can support a credible explanation before the entitlement process begins.

A weak water story can turn a technically feasible data center into a politically expensive one. A strong water story is specific: cooling design, peak demand, water source, conservation measures, wastewater handling and infrastructure commitments.

Use AI to maintain the water risk register

Water diligence produces scattered evidence. Engineering memos, utility emails, municipal plans, drought maps, meeting minutes, state water plans and cooling vendor assumptions rarely live in one clean file.

This is where AI is useful. An agentic workflow can:

1. Extract water assumptions from engineering documents

2. Pull municipal capacity references into a tracker

3. Compare demand against drought plans

4. Flag missing wastewater analysis

5. Monitor local news and meeting agendas

6. Update the project risk register as the design changes

The human role is judgment. Developers need engineers, counsel and local advisors to confirm feasibility and guide the public process. AI should make the work faster, not make the decision look easier than it is.

The practical checklist

Before a data center site reaches LOI, the water diligence checklist should answer seven questions:

1. What cooling system is assumed?

2. What is annual and peak water demand?

3. What water source will serve the site?

4. Is wastewater capacity available?

5. What indirect water demand comes from power generation?

6. What upgrades are required?

7. What community objections are likely?

If those answers are missing, the site is underwritten on hope. Power may be the first constraint everyone sees. Water is the constraint that can still kill the project after the power story looks solved.