Data Centres and the New Environmental Test for Digital Infrastructure
1. A Planning Question That Signals a Wider Shift
A recent question in the UK Parliament captures one of the most important issues now facing data centre development: before a hyperscale AI data centre is approved, how clearly should its water demand be disclosed and assessed?
The government’s answer was careful, but revealing. It stated that environmental impacts, including water consumption, must be properly assessed before planning decisions are made. For developments progressing through the Nationally Significant Infrastructure Projects regime, applicants must assess impacts on water resources through Environmental Impact Assessment and supporting documentation. AI Growth Zone applicants must also provide details of anticipated water requirements and evidence from the relevant water supplier on the availability of supply.
That answer gives a useful starting point for the wider debate. Data centres are no longer viewed only as digital infrastructure. They are power infrastructure, cooling infrastructure, water infrastructure and, increasingly, strategic national infrastructure.
The question is no longer whether we need more data centres. The question is how we design, approve and operate them with clear evidence of their energy, water and environmental impact.
This is an important shift. For many years, data centre performance was mainly discussed through uptime, resilience, security, capacity and energy efficiency. Those remain critical. But the rapid growth of AI has widened the conversation. The physical consequences of digital demand are becoming more visible: electricity connection, grid reinforcement, cooling strategy, water availability, heat rejection, land use, local environmental burden and long-term operational resilience.
For developers, this creates a new kind of challenge. A data centre proposal can no longer be judged only by its compute capacity or investment value. It also needs to show how it will behave as a physical system in a specific location, under real operating conditions, with real constraints from the grid, the water network and the surrounding environment.
2. The Global Scrutiny Around Data Centre Growth
The UK is trying to accelerate data centre development, particularly around AI Growth Zones. The government has described onshore data centre capability as essential for AI adoption, resilience and national strategic capacity. Its AI Growth Zones document says these zones are intended to unlock large-scale compute capacity by addressing two of the biggest barriers to investment: slow planning processes and delays in access to power.
Planning policy is already moving in that direction. The House of Commons Library notes that, following December 2024 reforms to the National Planning Policy Framework, local authorities in England are required to consider the need for data centres when setting local policies and deciding planning applications. It also notes that data centres can opt into the Nationally Significant Infrastructure Projects regime, where applications are determined by the Secretary of State rather than local planning authorities.
At the same time, the pressure points are becoming harder to ignore. The same House of Commons Library briefing notes that data centres currently consume around 2.5% of the UK’s electricity, with the sector’s electricity use expected to rise significantly by 2030. It also highlights concerns around grid capacity, energy costs and water consumption.
The UK is not alone in facing this tension between growth and scrutiny.
In Texas, the environmental permitting debate has become more immediate. In September 2026, the Texas Commission on Environmental Quality was directed to halt the issuance of environmental permits for infrastructure related to data centres, including air, water and waste permits. The directive followed earlier orders to audit data centres requesting grid connections, with a focus on grid reliability, public utility use, water use and electricity reporting requirements.
In the EU, the approach is moving toward transparency and comparability. The European Commission has proposed a common rating scheme for data centres, covering facilities above 500 kW. The scheme is intended to make energy and water use more transparent, cover data centres’ contribution to the grid, and support sustainable integration into Europe’s energy system. The Commission also notes that the EU aims to triple its data centre capacity over the next five to seven years.
These examples are different in policy detail, but they point to the same underlying trend. Governments want data centre capacity. Economies need compute. AI growth depends on physical infrastructure. Yet the approval, design and operation of that infrastructure is entering a more accountable phase.
The next generation of data centres will be judged by more than their ability to host servers. They will be judged by how convincingly they deal with energy, water, heat, resilience and local environmental effects.
3. Why the Environmental Scale Has Changed
The scale of the environmental question has changed because the scale of demand has changed.
The United Nations University report Environmental Cost of AI’s Energy Use estimates that global data centres consumed 448 TWh of electricity in 2025. If data centres were treated as a country, that would place them 11th globally for electricity consumption. The report projects that data centre electricity demand could exceed 945 TWh by 2030, accounting for almost 3% of projected global electricity use. AI is becoming a major driver of that growth. The same report estimates that AI workloads accounted for more than 20% of data centre electricity use in 2025, with projections suggesting this could rise to 40% by 2030. The report is also useful because it pushes the discussion beyond carbon. It argues that AI’s environmental footprint should be understood through carbon, water and land impacts, since the environmental result depends heavily on where electricity is generated and what resources are used to support it. That is a crucial point for data centre development. A project can look favourable under one metric while transferring pressure elsewhere.
A low-carbon energy source may still have water or land implications. A water-efficient cooling strategy may increase power demand. A high-density data hall may improve space utilisation while increasing local thermal risk. Additional redundancy may strengthen resilience while increasing equipment load, embodied carbon and operational complexity.
The idea of a “green data centre” therefore needs more careful handling. It cannot be reduced to one number or one claim. Environmental performance depends on the relationship between compute load, cooling architecture, grid mix, local weather, water availability, heat rejection, operating strategy and future capacity growth.
That makes early estimation essential. Once a site is selected, infrastructure is fixed, cooling plant is procured and operating assumptions are embedded, many of the largest environmental decisions have already been made.
4. From Disclosure to Engineering Evidence
A planning statement may report projected power demand or water use, but those figures need a credible technical basis. The more important question is how those estimates were produced, how sensitive they are to operating conditions, and whether alternative design choices could reduce energy or water demand before the project becomes fixed.
Data centres are coupled engineering systems. Rack density, airflow management, liquid cooling, CDUs, chilled-water networks, pumps, valves, thermal storage, controls, external heat rejection and weather conditions all interact.
A change in one part of the facility can shift demand elsewhere. Better containment may reduce recirculation but change pressure behaviour. A liquid cooling deployment may reduce air-side cooling load while introducing new technology-loop and heat-rejection requirements. A chiller or pump strategy may improve performance in one operating mode and create inefficiency in another. A facility designed around today’s rack densities may face very different constraints when future AI loads are added.
These interactions are difficult to understand from static assumptions. They need scenario-based assessment.
How will the facility respond to a rapid AI workload ramp?
Where will the first thermal constraints appear?
How much cooling capacity is unavailable because of poor flow distribution?
How does the system behave during a fan, pump, CDU or chiller failure?
What changes between winter operation, summer peaks and partial-load conditions?
How do energy use, water use and resilience trade against each other under different cooling strategies?
These are engineering questions, but they are increasingly becoming planning, investment and regulatory questions. If a developer is asked to disclose environmental impact, the disclosure has to be supported by methods that can evaluate real system behaviour.
That is where technical assessment becomes valuable early in the project. Not as an afterthought. Not as a final validation exercise. As part of the design process itself.
5. Growth With Technical Accountability
For Mansim, this is the part of the debate that deserves more attention.
The future of data centre development will depend on the ability to grow with technical accountability. That means more than reporting headline figures. It means understanding how a facility behaves across the full cooling chain, from IT equipment to racks, technology loops, data halls, chilled-water systems, thermal storage, external plant and day-to-day operations.
Mansim’s own data centre work reflects this broader view. The company’s data centre capability spans cold plates, processors, GPUs and immersion cooling; rack manifolds, CDUs and technology-loop hydraulics; data-hall airflow, containment and rack inlet temperatures; chilled-water networks, pumps, valves and thermal storage; and external plant including chillers, dry coolers, hybrid coolers and heat rejection. It also covers operational questions such as capacity planning, failure analysis and Reality DC digital twin workflows.
The methods are wider than CFD alone. They include hydraulic network modelling, transient thermal-hydraulic modelling, conjugate heat transfer, multiphysics simulation, sensitivity studies, design optimisation and digital twin modelling.
That combination is important. Optimisation rarely comes from one isolated improvement. It comes from understanding how the whole facility behaves across scales: chip to rack, rack to hall, hall to plant, plant to grid and operations.
Tools such as Cadence Reality DC sit within that wider engineering picture. Their value is not the digital model for its own sake. The value is being able to test capacity changes, rack deployment, maintenance plans, failure scenarios and design options before they become physical constraints. Mansim describes this operational use of validated models as supporting rack deployment, planned maintenance, facility changes and ongoing optimisation.
This is where policy and engineering meet. If planning authorities, regulators, utilities and communities ask for clearer evidence of environmental impact, the industry needs better ways to produce that evidence. If energy and water transparency become more prominent, the figures must be tied to real engineering behaviour.
Data centre growth is necessary. AI infrastructure will continue to expand, and the strategic case for more compute is clear. But growth without technical accountability will face increasing friction from planning systems, electricity networks, water constraints and local communities.
The strongest projects will be those that can show, before construction or expansion, how they will use energy and water, how they will manage heat, how they will remain resilient, and how their environmental effects have been estimated and reduced.
That is the new environmental test for digital infrastructure.
Comments