Data Centers Are Coming. How We Develop Them Matters.

A Landscape Architect's Perspective on Growth, Community and Development Stewardship

Walk into any public hearing on a data center these days and you'll notice something before the staff report even finishes: half the room has already made up its mind. Nobody's talking about the site. They're talking about how they feel about it.

Depending on who's doing the talking, these facilities are either the indispensable backbone of America's technological future or an environmental and community disaster waiting to happen. Neither camp is entirely wrong, and neither is entirely right. I suspect, as with most things I've dealt with in thirty years of this work, the truth is sitting somewhere in the dirt between the two extremes — which, come to think of it, is usually where the truth lives on any job site.

As a landscape architect, I care about the land these facilities sit on, the water and ecological systems around them, the infrastructure needed to support them and the communities that have to live next to them. I've come to look at questions like this through something I call Development Stewardship — the idea that development is going to happen whether we like it or not, but we have a responsibility to understand its consequences and pursue it wisely.

So I'm not here to cheerlead for data centers, and I'm not here to run them out of town either. I'm asking a simpler question:

How do we develop them well?

Why Data Centers Matter

A data center isn't just a warehouse full of blinking servers. It's the physical plant behind nearly everything modern life runs on — banking, communications, health records, government systems, cloud storage, scientific research, national defense, and now, increasingly, artificial intelligence.

That matters beyond convenience. Domestic computing capacity, cloud infrastructure, data security and advanced AI now carry real economic and national-security weight. Federal policy explicitly treats AI data centers and the infrastructure that supports them as central to American technological leadership, economic competitiveness and national security (White House, "Accelerating Federal Permitting of Data Center Infrastructure").

None of that means every proposed data center deserves a permit. But it does mean "we just don't need these" isn't a realistic starting position. Our appetite for digital services isn't shrinking. It's accelerating. And the infrastructure to feed that appetite has to live somewhere.

The Concerns Are Real

Acknowledging that data centers matter doesn't mean we look away from what they cost.

Electricity is the obvious one. The U.S. Department of Energy reported data centers consumed roughly 4.4 percent of total U.S. electricity in 2023, with earlier projections putting that share at 6.7 to 12 percent by 2028 (U.S. Department of Energy, "DOE Releases New Report Evaluating Increase in Electricity Demand from Data Centers"). Lawrence Berkeley National Laboratory's 2025 update, published in 2026, estimates data centers could reach roughly 11.8 percent of total U.S. electricity consumption by 2030, with modeled scenarios running anywhere from 9.5 to 15.3 percent (Berkeley Lab, "United States Data Center Energy Usage Report: 2025 Update"). That range isn't a typo or a hedge — it's a reminder that these are projections shaped by uncertain growth, shifting technology and how facilities are actually operated, not a fixed number carved in stone.

Water is the other big one, especially where evaporative cooling is involved. But it's not a uniform cost across every facility. Berkeley Lab researchers found water use tied to computing workloads swings enormously depending on server efficiency, cooling systems, climate, operating practices and even where the electricity itself comes from (Berkeley Lab, "The Water Use of Data Center Workloads").

Beyond those two headline issues, you're also looking at:

• Land clearing and habitat disturbance

• Wetland and watershed impacts

• Stormwater runoff

• New transmission lines and substations

• Backup-generator emissions

• Mechanical noise

• Lighting and visual impacts

• Demands on public roads and utilities

• Compatibility with neighboring land uses

• Who ends up footing the bill for infrastructure — the developer, the utility, or the ratepayer next door

  
  

None of that is irrational opposition. It's a punch list, and it deserves the same scrutiny we'd give any punch list. But a long list of concerns isn't automatically a reason to kill a project either.

Data Centers Aren't the Only Ones at the Table

Here's something that tends to get lost in all the noise: data centers didn't invent industrial development, and they won't be the last item on the agenda either.

Refineries are expanding or coming online. Paper mills are permitting new lines or retrofitting old ones. Power generation projects — gas, solar, battery storage, sometimes all three on the same site — are working their way through the same planning departments. Sawmills, chemical plants, manufacturing facilities, distribution centers — the list of industrial and heavy-commercial projects either operating, under construction, or sitting in some planning commission's queue right now is a lot longer than the evening news would suggest. Data centers just happen to be the ones getting the headlines at the moment, because they're new, they're big, and they're tied to a technology everybody has an opinion about.

I bring this up because it matters for how we think about all of it. A community that spends its entire attention span litigating data centers while a refinery expansion or a new mill sails through with far less scrutiny isn't necessarily protecting itself — it may just be looking at the wrong door. Industrial development, broadly, is not new and it is not slowing down. The same Development Stewardship questions — where, what systems does it touch, who bears the cost, what gets avoided and mitigated — apply whether the sign out front says "Data Hall 3" or "Paper Mill Expansion" or "Combined-Cycle Generating Station." The technology changes. The land-use fundamentals don't.

National Statistics Don't Decide Local Land-Use Questions

Big numbers tell you the scale of a problem. They don't tell you whether a specific project, on a specific parcel, in a specific community, is responsible development.

Say someone tells you: a data center can use as much water as thousands of homes. That might be true under the right circumstances. But it's the opening question of an investigation, not the closing argument. What cooling system is actually proposed? How much water would the facility really use, and is that demand year-round or seasonal? Is it potable water, reclaimed water, surface water, groundwater? What's the local system's capacity, and who else depends on it? What happens during a drought? Could a different cooling technology cut the demand in half?

Same discipline applies to power. What's the projected load? Is generation already sitting there, or does someone have to build it? What transmission and substation work gets triggered, and who pays for it? Can the facility flex its demand during grid stress, or does it just show up and take? That last question has gotten serious enough that federal policy discussions now address whether data-center operators should cover their own power-delivery infrastructure rather than passing the tab to ordinary ratepayers (White House, "Ratepayer Protection Pledge").

The claimed benefits deserve the same treatment. If a developer waves around a multibillion-dollar investment figure, ask how much of that actually stays local. How many construction jobs, and for how long — this isn't a three-day install, but it's not permanent either. How many permanent jobs, at what wages? What tax revenue survives after the exemptions get carved out? According to the National Conference of State Legislatures, 38 states now offer some form of preferential tax treatment for data centers, and the strings attached — or not attached — vary a great deal from state to state (NCSL, "Subsidizing Servers: How States Are Competing to Attract Data Centers").

A big capital number can deliver real community benefit. But capital investment, permanent jobs, tax revenue and public cost are four different animals, and Development Stewardship means we look at each one on its own, not as a package deal.

Maybe We Should Look at Data Centers as Landscapes

Here's where I think landscape architecture actually earns its seat at this table.

Most people describe a data center as a building, or a piece of tech infrastructure. We tend to see something wider than that — land, water, ecology, infrastructure and people, all working as one interconnected system.

A data center occupies land. It changes drainage. It draws power, sometimes water. It throws off heat. It needs roads, utilities, security. It might clear vegetation or disturb habitat. It has neighbors, and it sits inside somebody's viewshed. Its mechanical equipment makes noise you can hear from the property line. It touches public infrastructure, the local economy and the character of the place it's dropped into.

In other words, a data center isn't just a building full of computers.

It's a landscape intervention. Full stop.

A 300-acre campus built on already-disturbed industrial land next to adequate transmission is a completely different proposition from the identical building requiring forest clearing, wetland fill, new transmission corridors and heavy groundwater withdrawal next to an established neighborhood. Same blueprints. Same server racks. Different landscape entirely.

Where Matters

When people hear "hundreds of acres" or "substantial water and power demand," it's natural to picture it landing in their backyard. But development shouldn't happen in a regulatory vacuum — that's what zoning ordinances, comprehensive plans, environmental regulations, setbacks and buffers exist for. Different land uses carry different impacts, and they're not equally appropriate everywhere.

Those safeguards aren't perfect. Zoning gets outdated. Regulations weren't written with this scale of facility in mind. Local governments sometimes lack the technical horsepower to properly evaluate a proposal this complex. Even a project that's technically zoned correctly can still create impacts that spill past the property line.

But the answer to a bad fit isn't to blacklist an entire category of development — it's figuring out where that use actually belongs, and what conditions ought to travel with it. A data center on the right industrial parcel, with real infrastructure and honest separation from residential uses, is not the same conversation as the same facility proposed next to an established neighborhood or inside an environmentally sensitive landscape. That determination gets made case by case, every time, with no shortcuts.

Development Stewardship Starts Without a Predetermined Answer

We live in a culture that rewards the instant take. You're expected to declare yourself for or against something before you've had time to actually understand it. That might work fine on social media. It's a terrible way to make land-use decisions.

Development Stewardship starts from a blank page. It accepts that growth is coming and that not every project belongs everywhere or should get built exactly as first drawn. The goal isn't to rubber-stamp a project or torpedo it — it's to understand it well enough to know whether it belongs, what it will actually cost, and how it might be made better. That means asking:

• What's actually being proposed?

• Where would it sit?

• What land and environmental systems does it touch?

• What infrastructure does it require?

• What does the community actually gain?

• What burdens does the community carry?

• What can be avoided? What can be minimized? What remaining impact needs to be mitigated?

• Who pays for the improvements it triggers?

• What commitments are actually enforceable over the life of the project — not just promised at the ribbon-cutting?

  
  

The U.S. Government Accountability Office has flagged that definitive conclusions about AI's environmental footprint are hard to pin down, since the technology moves fast and private developers don't always disclose the technical details that would let anyone check their math (GAO, "Generative AI's Environmental and Human Effects"). That uncertainty isn't a license to ignore the impacts — it's a reason communities should demand project-specific numbers instead of settling for generalized comfort or generalized panic.

The conclusion follows the analysis. Not the other way around.

What Wise Development Might Look Like

Development Stewardship takes vague worry and turns it into planning and design questions you can actually answer.

Water a concern? Look at reclaimed water, a closed-loop system, or a less thirsty cooling technology. Tree clearing an issue? Concentrate development on the parts of the site already disturbed, and hold onto the habitat, drainage corridors and mature canopy elsewhere. Noise? Use distance, existing landforms, sound barriers, or just reposition the mechanical equipment. Community character on the line? Setbacks, preserved vegetation, buffers, lighting standards and architectural requirements go a long way. Power demand? Ask about generation, storage and demand-flex measures that ease grid stress instead of adding to it — and make sure any required infrastructure upgrades strengthen the surrounding grid rather than just servicing the new tenant.

And if incentives are on the table, tie them to something measurable — local hiring, minimum investment thresholds, permanent job counts, water-efficiency standards, infrastructure contributions, environmental performance. Not vibes. Numbers.

All of this runs on a hierarchy most of us in this field already know by heart:

AVOID → MINIMIZE → MITIGATE → COMPENSATE

Can the impact be designed or sited away entirely? If not, can it be reduced? What's left after that gets mitigated. And whatever burden remains after all three steps — that's what gets compensated.

The Trade-Off Nobody's Negotiating Yet

There's a piece of this conversation I think gets skipped too often, and it's the one I find most interesting: the negotiation itself.

Too many of these discussions get treated as a binary — approve it as submitted, or turn it away. But a rezoning, a conditional use permit, an incentive package — these are all negotiations, whether anyone in the room wants to call them that or not. And a developer sitting on a multibillion-dollar capital investment has room to negotiate. That's not a knock on the developer. That's just where the leverage sits.

So use it.

If a data center needs a rezoning or wants public incentives, that's the moment to ask for something in return that outlasts the ribbon-cutting. Maybe the developer funds a public park on an adjacent parcel instead of just landscaping their own frontage. Maybe they set aside and permanently protect a conservation easement — real acreage, not a retention pond with a plaque on it. Maybe they fund the restoration of a degraded wetland or stream corridor nearby, something the community has wanted fixed for years but never had the capital for. Maybe the water-efficiency measures they'd install anyway get formalized into a public commitment, with monitoring and teeth. Maybe the substation upgrade they need gets sized and located so it strengthens service to the whole area, not just their meter.

None of that happens by accident. It happens because somebody at the table asked for it, put it in writing, and made it enforceable.

I've spent thirty years designing landscapes, and one thing I've learned is that a site plan is never just about what goes inside the property line. It's about what that development gives back to everything around it. A public park funded by a data-center developer isn't charity — it's a fair trade for the land, water, power and tolerance a community is being asked to provide. A restored wetland isn't a consolation prize — it's mitigation that actually improves the watershed instead of just checking a box on a permit application.

This is where Development Stewardship gets practical instead of philosophical. It's not enough to ask what impacts can be avoided, minimized, mitigated and compensated. Ask what asset gets created in the process. Good negotiation doesn't just reduce harm — it can leave a community better off than it was before the project ever showed up. A conservation area that didn't exist before. A trail system. A cleaned-up creek. Funded environmental monitoring that outlives the permit. These are the kinds of things that turn a contentious hearing into a project people are actually proud to point to twenty years later.

Communities and their elected officials shouldn't be shy about this. The developer needs the site more than the site needs the developer — that's usually the actual balance of power, even when it doesn't feel that way in the hearing room. Wise development isn't just about saying yes or no. It's about knowing what to ask for before you say yes.

Sometimes the Answer Is Still No

Development Stewardship isn't a promise to find a "yes" for every project that walks in the door. Some sites don't have the water. Some don't have the grid or road capacity. Some sit on environmental resources too significant to trade away, or too close to neighbors who'll never coexist comfortably with the use. In those cases, the responsible call is to send the project back for a redesign, relocate it, or reject it outright.

That's not anti-development. That's wise development.

The reverse holds just as true. A genuinely good project shouldn't get killed because someone recited a scary national statistic that has nothing to do with the actual site in front of the commission. Sometimes a proposal is fine as submitted. Sometimes design changes get it there. Sometimes the site is just wrong. The analysis tells you which — not the loudest voice in the room.

The Technology Will Keep Changing

Somewhere down the line, a good chunk of our computing infrastructure might end up in orbit, running on solar power and sidestepping a lot of the land, water and community conflicts we're wrestling with today. Concepts for orbital data centers are already floating around, though the engineering, maintenance and economic hurdles are still substantial — we're not there yet, and I wouldn't hold my breath.

Long before that ever happens, improvements in computing efficiency, cooling, water reuse, energy storage and power generation could meaningfully shrink the physical and environmental footprint of the facilities we're building on the ground today.

I can't tell you exactly where the technology is headed. What I can tell you is that our need for computing capacity isn't going anywhere. Data centers, in some form, are going to keep powering a lot of our economic, scientific and technological progress.

Just saying "no" isn't a strategy. Neither is saying "build it" without asking what it costs. The real skill is learning how to say:

Yes — wisely.

A Better Question

Maybe communities should stop asking whether data centers are good or bad, and start asking something more useful:

What would responsible data-center development look like here?

Here matters. On this land. In this watershed. Connected to this grid. Drawing from this water supply. Beside these neighbors. Inside this economy. Under these environmental conditions.

Development is coming — and not just data centers. Refineries, mills, power plants and everything else on that long industrial queue are coming right along with them. Our job — as designers, planners, developers, public officials, environmental professionals and citizens — was never just to decide whether we're for or against growth.

It's to get a lot better at deciding how we grow, and a lot bolder about what we ask for in return.

That's Development Stewardship.

~ Eric