F E A T U R E

These are different regulatory and policy routes, but together they reflect a wider shift. As AI-driven development accelerates globally, data centre water demand, cooling architecture, source availability, reuse and resilience are receiving greater scrutiny earlier in planning, alongside mainstream reporting and community debate.

The question is not whether water matters, but whether it is understood early enough to influence key design decisions while meaningful options remain open.

Beyond the basics

The basics of data centre design – power capacity, land, connectivity and project delivery timelines – have historically dominated early discussions. Water is also site-specific yet has received less attention, particularly when viewed through the lens of what could be described as the new age of data centres. In practice, water demand and treatment requirements are increasingly affected by geographical location and climate, local water availability and quality, IT load and cooling architecture and the operator’ s resilience requirements and operating model. Headline consumption figures are not a good substitute for the water balance of a specific project. At the site-planning and design-engineering stage, the water supply position needs to be clear enough to inform:

• Treatment and reuse routes

• Space, utilities and interfaces with wider MEP systems

• Discharge and contingency arrangements

• Whole-life cost, maintenance and future expansion

Contributions by mechanical, electrical and plumbing( MEP) consultants, water utilities and, where appropriate, local authorities, might be necessary – because resolving issues early on helps to avoid costly redesign, rework and programme disruption.

Bringing water into the planning stage

Since September 2026, Scotland has required an Environmental Impact Assessment( EIA) for applications for new data centres exceeding 50 MW power capacity. The Scottish Government Chief Planner letter and EIA Direction reference AI-related water consumption, potential impacts on water resources and the role closed-loop cooling can play in significantly reducing the amount of water required where water is used for cooling.

The Scottish Government also emphasises identifying EIA at the earliest possible stage, when mitigation can be integrated into the design. This does not prescribe a single cooling or water solution; it reinforces the value of establishing the water position before key design decisions become harder or more expensive to change.

American markets – acting in the same vein

In Texas, an August 2026 data centre audit directive instructed the Public Utility Commission of Texas and Electric Reliability Council of Texas( ERCOT) to obtain information – from projects progressing through ERCOT ' s interconnection process – including projected annual and peak water consumption, anticipated water sources and proposed cooling technology. A follow-up in the next month told the Texas Water Development Board to enforce existing water-use reporting requirements and coordinate with ERCOT on additional data centre water information.

Virginia is taking a different approach. From January 2027, amended water-withdrawal reporting requirements require certain water providers to categorise and report potable and reclaimed water supplied to data centres with Department of Environmental Quality air permits. Separate 2026 budget provisions also place water-efficiency requirements on new data centres in defined cooling water scarcity areas and the Eastern Virginia Groundwater Management Area.

Although the regulatory routes differ, both focus on the same practical shift – water is being considered earlier in data centre development. That means giving greater attention to how much may be needed, where it will come from, how cooling choices affect demand and what role reuse or more water-efficient approaches could play.

Planning ahead

The same early-planning principle is in the Alliance for Water Efficiency ' s 2026 Data Center Primer for Water Service Providers, developed to help providers engage earlier in the planning and design of data centres. It highlights missed opportunities for efficiency and reuse when water providers are brought in late in the cycle. A practical early-stage checklist is helpful for establishing the site-wide water situation and should cover:

• Expected average and peak water demand

• Cooling architecture, including liquid-cooling interfaces, and associated water-quality and filtration requirements

• Mains, groundwater, reclaimed water, surface water or other technically suitable source-water options, including whether rainwater harvesting or specialist surface-water treatment is appropriate

• Opportunities for data centre water reuse, recovery and recycling, where appropriate

• Opportunities to reduce feedwater demand and wastewater volumes through high-recovery water treatment or, where justified, Zero Liquid Discharge( ZLD)

36 www. intelligentdatacentres. com