How Town Planning Works | TPW-0053 — The Data Centre District: How Digital Infrastructure Rewrites Power, Water, Land, Noise and Local Value

A data centre can look deceptively simple from the road.

It is a large building, usually quiet from a distance, with relatively few people entering and leaving compared with an office tower, shopping centre or factory.

That visual simplicity hides an unusually demanding urban system.

A large data centre can require enormous and highly reliable electricity supply, multiple fibre routes, backup power, cooling infrastructure, secure perimeters, substations, water or heat-rejection systems, stormwater capacity, emergency access and substantial construction logistics. It may generate major capital investment and tax revenue while creating far fewer long-term jobs than similarly large industrial developments. Its impacts can be highly concentrated in the host community even when the digital services it supports are used globally.

This is why data centres have become a town-planning issue rather than merely a technology-industry issue.

The American Planning Association’s 2026 Trend Report identifies both a data-centre boom and growing community pushback as immediate planning trends. The International Energy Agency’s 2026 outlook estimates that global data-centre electricity demand is rising rapidly and could roughly double between 2025 and 2030. In the United States, the IEA expects data centres to account for about half of electricity-demand growth through 2030. The planning challenge is therefore no longer hypothetical: towns are deciding where these facilities belong, what infrastructure they require, who pays for that infrastructure and how much local value they actually create.

The first planning mistake is treating a data centre like an ordinary warehouse

Data centres are often built in industrial-looking structures: large floorplates, few windows, loading access and extensive mechanical equipment.

That can tempt zoning systems to classify them alongside warehouses.

The land-use behaviour is different.

A warehouse typically consumes land and road capacity. A data centre consumes land and an exceptional amount of electrical capacity. Some facilities also create significant water demand, continuous mechanical noise and large demands for backup generation and utility redundancy.

The correct planning category should therefore reflect the infrastructure profile, not merely the building appearance.

Communities such as Loudoun County, Virginia, and Sugar Grove, Illinois, now use data-centre-specific zoning standards or districts that address siting, setbacks, building design, equipment, lighting, parking and performance. The logic is simple: when a land use has system impacts unlike ordinary industrial activity, it deserves its own planning questions.

Power is the real site-selection map

A data centre can buy land almost anywhere. It cannot obtain large volumes of reliable electricity everywhere.

Electricity therefore becomes one of the strongest spatial determinants of location.

Developers look for substations, transmission capacity, interconnection potential, redundant feeds and the ability to expand. Large campuses may require hundreds of megawatts. Some proposed projects are now planned at gigawatt scale.

This means the planning map and the grid map must be read together.

A site may be zoned perfectly and electrically impossible. Another site may have excellent grid access but lie near homes, water constraints or ecologically sensitive land.

The town-planning job is to find places where the digital use and the physical energy system can fit without simply transferring costs to everyone else.

Grid capacity is a public resource even when the building is private

Electricity networks are shared infrastructure.

If one project consumes a large share of available capacity, other growth may be delayed. Housing, factories, transit electrification and ordinary businesses can all depend on the same substations and transmission system.

The 2026 U.S. Department of Energy transmission-needs work explicitly identifies data centres among the new large loads increasing pressure on the grid.

The planning question is therefore not merely whether the operator can pay its electricity bill.

It is whether the grid can serve the facility while preserving reliability and development capacity for the wider town.

This may require utility impact studies, phased interconnection, dedicated generation, storage, transmission upgrades or cost-sharing arrangements.

Without that analysis, a planning approval can accidentally allocate future urban capacity through one private project.

Who pays for the wires matters

Large electrical upgrades are expensive.

If new substations, lines or generators are required, the financial structure determines who benefits and who carries risk.

A project can appear fiscally attractive to a municipality while shifting part of its infrastructure cost into utility rates paid across a wider region.

That does not mean data centres necessarily increase household electricity prices. Grid economics are complex, and new generation or transmission can also create wider benefits.

The important planning principle is transparency.

Before approval, decision-makers should understand what infrastructure is required, who finances it, what capacity remains afterward and what happens if the data-centre project is delayed or cancelled after the public system has already invested.

Water is a local question, not a universal answer

Data-centre water use varies enormously.

Some facilities use evaporative cooling that can consume substantial water. Others use closed-loop or air-cooled systems that reduce operational water demand but may increase electricity use or capital cost.

Climate matters. Technology matters. Workload matters. Water source matters.

This makes blanket arguments about whether data centres are “water intensive” too crude for planning.

The correct question is what the specific facility will use under local climatic conditions and peak operating scenarios.

EPA guidance on data-centre redevelopment notes that some large facilities may require major water supplies and that reclaimed water can be a useful alternative where systems are available. Quincy, Washington, provides a well-known example of water reuse supporting a community with significant data-centre and industrial demand.

Planning should therefore ask about potable water, reclaimed water, drought conditions, wastewater, cooling technology and future expansion together.

The peak day matters more than the annual average

Infrastructure often fails at the peak.

A data centre may have manageable annual water consumption and create a difficult peak during hot weather when households and other industries also consume more.

The electricity system faces the same issue.

Annual energy consumption describes quantity. Peak demand describes capacity.

Town planning should therefore require infrastructure assessments that test coincidence: what happens when the facility reaches high demand at the same time the wider town is experiencing heat, drought or other peak conditions?

This is another application of The Hidden Town. Utilities have capacity limits even when those limits are invisible from the street.

The employment story needs to separate construction from operation

Data-centre projects can produce large construction workforces.

Operational staffing is often much smaller.

This distinction matters because local economic-development debates can combine temporary and permanent employment into one impressive number.

Construction employment is real value. It supports trades, suppliers and local spending.

But a town considering the long-term use of hundreds of acres should also ask how many stable jobs remain after opening, what skills they require and whether local workers can realistically fill them.

The American Planning Association’s 2026 data-centre trend coverage highlights this tension directly: major investment can arrive with relatively few permanent jobs.

The right comparison is not “jobs or no jobs.” It is value per unit of scarce land, power, water and public infrastructure.

Tax revenue can be enormous—and still deserves stress testing

Data centres can generate major property, equipment or business tax revenue depending on local law.

This can transform a municipal budget.

The opportunity is real, but so is concentration risk.

If one industry becomes a large share of the tax base, policy changes, depreciation rules, corporate restructuring or technology shifts can create fiscal volatility.

Tax incentives can also reduce the public benefit substantially.

A town should therefore model gross revenue, incentive cost, infrastructure cost, ongoing public-service cost and downside scenarios.

The best economic-development deal is not necessarily the project with the largest announced investment. It is the project whose long-term local value remains strong after public obligations are counted.

Noise comes from the mechanical system, not the people

A data centre may have few employees and still create continuous noise.

Fans, chillers, cooling towers, transformers and backup generators can operate for long periods. Tonal or low-frequency noise can travel differently from ordinary traffic noise.

Setbacks alone may not solve the problem.

Planning should consider equipment placement, acoustic enclosures, building orientation, screening, topography and nighttime background noise.

Performance standards are often more useful than arbitrary distance because two facilities using different cooling technologies may behave very differently.

The applicant should therefore model expected sound at property boundaries and nearby sensitive uses under realistic operating conditions.

Backup generators turn rare events into planning conditions

Data centres require extraordinary reliability.

That often means batteries and backup generators.

The generators may operate rarely, but they still matter for emissions, noise, fuel storage, testing and emergency planning.

A facility with dozens or hundreds of generators can behave like a small power station during an outage.

Planning review should therefore include testing schedules, emission permits, fuel delivery, fire protection and outage scenarios.

The edge case is part of the land use.

Building height is not the only visual issue

Data centres can be physically large without being tall.

Long blank facades, rooftop mechanical systems, substations, security fencing and transmission infrastructure can dominate a landscape.

Good design standards can reduce visual impact without pretending the building is something else.

Facade articulation, landscape buffers, equipment screening, coordinated fencing and thoughtful site layout can help.

But the planning system should avoid cosmetic requirements that consume money while ignoring the real issues of power, water, noise and infrastructure.

The design review should address both what the building looks like and what the facility does to the town.

Brownfields can be logical data-centre sites

Former industrial or contaminated land can offer large parcels, infrastructure access and separation from homes.

The U.S. Environmental Protection Agency published 2026 guidance specifically on redeveloping Superfund and brownfield sites as AI data centres.

This can be a powerful land-recycling strategy when site conditions fit.

A brownfield location may reduce pressure to consume farmland or greenfield land. Existing transmission, water or industrial access can lower infrastructure cost.

But brownfield reuse is not automatically sustainable.

Cleanup requirements, flood risk, environmental justice and infrastructure constraints still matter.

The relevant connection is The Brownfield Town: reuse is valuable when the new land use genuinely fits the site and wider system.

Farmland and rural communities face a different tradeoff

Large data-centre campuses increasingly target suburban and rural land because parcels are large and cheaper.

This can create major tax revenue for places that have lost industrial employment.

It can also convert farmland, change landscape character and require new transmission infrastructure.

The planning question should therefore compare the project with the full opportunity cost of land conversion.

Is the site uniquely suitable because of grid and fibre access? Could a lower-value brownfield site work instead? Is agricultural land strategically important? What infrastructure follows the development?

Land that looks empty on a satellite image may already be performing food, drainage, habitat or landscape functions.

Heat rejection is an urban-design issue

Computing produces heat.

Cooling moves that heat somewhere else.

Most planning debates focus on energy and water, but the heat itself can become useful or problematic.

In colder climates, waste heat can potentially serve district-heating networks, greenhouses or nearby buildings if temperatures and distances are suitable.

In hot climates, adding large quantities of waste heat to an already warm industrial district may worsen local conditions if not managed carefully.

The strongest planning approach asks whether the thermal output can be integrated into another urban system rather than simply discharged.

Fibre redundancy shapes location too

Electricity is not the only network.

Data centres need high-capacity telecommunications and ideally diverse fibre routes.

A facility with two fibre connections that share the same physical trench does not have true redundancy.

Town planning can help protect corridor diversity by coordinating roadworks, utility easements and right-of-way access.

This is another reason data-centre planning should connect to utility-corridor planning rather than sit only inside the zoning department.

Construction traffic can be the largest short-term impact

Large campuses can require years of construction.

Concrete, steel, transformers, cooling equipment, generators and servers arrive by heavy vehicle.

Road damage, dust, noise and worker traffic can become significant even when the completed facility produces little daily traffic.

Planning conditions should therefore distinguish construction impacts from operational impacts.

Construction-management plans, haul routes, road upgrades, working hours and restoration obligations may matter more than permanent parking ratios.

Security should not produce a dead district

Data centres require security.

Large fenced compounds can also create long inactive edges.

If several facilities cluster together, the result can be a district with poor pedestrian permeability, few services and almost no public life.

That may be acceptable in a remote utility-industrial zone and damaging beside an urban centre or transit corridor.

The planner should therefore match the land use to the urban context.

Land near high-capacity transit, housing and active streets may be too valuable for low-employment secure compounds unless the project provides exceptional system value.

Cluster strategy matters

One data centre can be manageable.

Ten nearby data centres can change the regional electricity, water and land system.

Cumulative impact therefore matters more than parcel-by-parcel review.

A town expecting a cluster should plan at district scale.

Reserve substations and transmission corridors. Plan water and wastewater. Establish noise standards. Protect residential buffers. Coordinate road upgrades. Decide where future expansion may occur.

The district plan should also define a ceiling if infrastructure or environmental capacity becomes constrained.

Growth is easier to govern before the first project makes every later approval feel inevitable.

A data-centre overlay can be more useful than case-by-case bargaining

Where strong demand exists, towns may benefit from clear regulations before the next application arrives.

An overlay or dedicated district can identify suitable areas and establish predictable standards for power studies, setbacks, noise, cooling, water, landscaping, generators, lighting and public reporting.

This creates two advantages.

Developers know the rules earlier. Communities do not need to reinvent the same debate for every parcel.

Predictability is valuable to both sides when the rules are evidence-based.

Community benefit agreements need measurable obligations

Communities sometimes negotiate benefits in exchange for hosting large infrastructure.

These can include road improvements, workforce training, parks, energy projects or local funding.

The agreement should be connected to measurable commitments rather than promotional language.

How much investment? By when? Who is responsible? What happens if the project expands? Which obligations survive ownership changes?

A benefit agreement should also avoid substituting small visible amenities for large invisible system costs.

A new park does not compensate for an unplanned electrical bottleneck.

Economic resilience requires an exit scenario

Technology infrastructure can become obsolete.

A town approving a large campus should ask what happens if the operator leaves.

Can the building be reused? Who removes generators and fuel systems? Who restores land? What happens to dedicated utility infrastructure?

Decommissioning bonds or closure plans may be appropriate in some jurisdictions.

The aim is not to assume failure. It is to prevent the public from inheriting stranded infrastructure if the market changes.

The planning application should include a system budget

A useful way to evaluate a major data centre is to require a system budget.

Land budget: hectares occupied now and at full buildout.
Power budget: peak megawatts, annual consumption, grid connection and redundancy.
Water budget: normal and peak use by source and cooling technology.
Noise budget: predicted boundary and receptor levels, including emergency operation.
Carbon budget: operational energy source, backup generation and major embodied impacts.
Traffic budget: construction and operating trips.
Employment budget: construction, permanent jobs and skill requirements.
Fiscal budget: taxes, incentives, infrastructure costs and public obligations.
Risk budget: fire, outage, drought, heat and decommissioning scenarios.

This allows decision-makers to compare projects using the same physical and fiscal language.

A practical planning test for a data-centre site

Grid: Can the electrical system serve the project without crowding out other planned growth?
Cost: Who pays for generation, transmission and distribution upgrades?
Water: What cooling system is proposed and how does it perform in drought and peak heat?
Location: Is the land better suited to another high-value urban use?
Compatibility: Are nearby homes protected from noise, lighting and visual impact?
Construction: Can roads handle the buildout period?
Jobs: What permanent employment remains after construction?
Tax: What is the net public value after incentives and infrastructure?
Resilience: Are energy, fibre and cooling systems redundant?
Environment: What are the effects on water, emissions, land and habitat?
Expansion: Is the full campus buildout being assessed now?
Closure: What happens if the facility becomes obsolete?

The Data Centre District in the wider Town Planning series

This article owns the planning job created when digital infrastructure becomes a large physical land use. The Hidden Town explains utility corridors. The Brownfield Town explains land recycling. The Productive Town explains economic geography. The Data Gap explains uncertainty and evidence.

The Data Centre District adds a distinct question: when computing infrastructure becomes physically enormous, how should the town allocate the power, water, land and public capacity it consumes?

The computer is global. The infrastructure is local.

A person can use a digital service without knowing which town contains the machines running it.

The host town cannot be equally abstract.

It sees the substation, water demand, tax agreement, transmission line, generators, construction traffic and land conversion.

This is the central planning fact of digital infrastructure.

The service may live in the cloud.

The cloud still needs a place on the map.