How to Review Water Risk for an AI Data Center

Water use depends on the facility, cooling system, climate, power source, and operating schedule. A workload-level analogy cannot replace a site-level water review.

Define the measurement boundary

• Withdrawal: Water taken from a source.
• Consumption: Water not returned to the same local watershed after use.
• Discharge: Water returned after treatment or cooling.
• Indirect use: Water associated with electricity generation outside the site.

State which values are measured and which are estimated. Do not combine withdrawal and consumption in one total.

Review the cooling design

Confirm the expected rack density, supply temperatures, cooling mode, seasonal operating range, and backup mode. Air cooling, closed-loop liquid cooling, and evaporative systems have different electricity and water trade-offs. The equipment name alone does not establish site water use.

Verify the local constraints

• Source capacity and drought restrictions
• Withdrawal, discharge, and construction permits
• Municipal allocation and competing users
• Water quality and treatment requirements
• Limits that apply during heat or drought events
• Cost and schedule for required water infrastructure

Ask for operating evidence

For an existing facility, request monthly withdrawal and consumption data with weather, load, and cooling-mode context. For a proposed facility, retain the engineering basis, permit status, utility correspondence, and each forecast assumption.

Put water in the site decision

Water availability, permit limits, cooling performance, and backup operation belong in site diligence. Record who owns each mitigation, what it costs, and whether it changes the power or construction schedule.

For current source-backed AI infrastructure records, browse the greencio intelligence feed.