If Not a Data Center, Then What?

The Cost of Saying Yes, No—or Nothing at All

Earlier this month, Frederick County, Maryland, appeared to offer one of the clearest examples yet of the emerging data-center bargain.

Developers behind the proposed Frederick Digital Campus offered more than $110 million in community benefits, including investments in schools, recreation, workforce training, agricultural preservation, community solar, and public safety. The proposal also called for reducing planned data-center square footage by nearly 20%, cutting potable-water use by 80%, and preserving 433 acres as a nature reserve.

Two weeks later, the county executive rejected the associated development-rights agreement and extended a pause on new data-center applications until July 1, 2027.

That does not mean the negotiations were pointless. It means negotiation and approval are two different things. A better proposal can still be rejected. A community can decide that even after concessions, a particular project does not fit its long-term plans.

But that raises another question—one public debates often skip:

If not this, then what?

And beyond the individual parcel:

What happens if enough communities make the same choice?

That is the final question of The Data-Center Divide.

I. COMPARE: THE ALTERNATIVE IS RARELY “NOTHING”

Public debate often compares a proposed data center with an imaginary alternative.

• No giant building.
• No electrical demand.
• No water use.
• No construction traffic.
• No noise.
• No change.

But land-use decisions rarely offer a choice between development and perfection. A parcel suitable for a large data center might alternatively become housing, a warehouse or distribution complex, manufacturing, retail, another industrial use, preserved agriculture—or remain undeveloped for years.

Each alternative has its own ledger.

Housing may create badly needed homes while also adding students, traffic, roads, utilities, parks, fire protection, and other public-service requirements.

Manufacturing may produce more permanent jobs but also bring industrial traffic, emissions, utility demand, and incentives.

Warehousing may consume significant acreage and generate heavy truck traffic while producing a different tax profile.

Agriculture or open space may preserve community character, environmental value, and optionality—benefits that are real even when they are difficult to express in dollars.

Doing nothing can itself have value.

The point is not that one use is inherently superior.

The point is that every choice should be compared with the realistic alternatives available to that community.

The proper comparison is not a data center versus perfection. It is a data center versus what would realistically happen instead.

That changes the conversation.

Instead of asking only, “What are the problems with this proposal?”

We also ask:

What would replace it? What would that alternative contribute? What would it cost? And what community objective are we trying to achieve?

“Compared with what?” is not a rhetorical question.

It is an analytical one.

II. DECIDE: WHAT DOES “NO” ACTUALLY BUY?

Sometimes the answer should be no—a responsible no.

A project may sit too close to homes. Its power requirements may exceed what the local system can responsibly support. Its water demands may be poorly matched to local resources. The developer may refuse to provide sufficient information. The economics may depend on public concessions that exceed the value the community receives. Or residents may decide that the land serves a more important purpose.

Those are legitimate decisions.

The reason for saying no matters. A rejection should rest on clear, project-specific concerns and a defensible community outcome—not election-cycle pressure, partisan signaling, generalized distrust, or the political convenience of opposing whatever is unpopular at the moment. “No” should be a decision made for the community, not simply a position taken before the evidence is weighed.

“No” should mean something.

If rejecting a project preserves agricultural land that the community intentionally wants to keep, that is a definable outcome. If it protects a residential neighborhood from an incompatible industrial use, that is an outcome. If it prevents ratepayers from absorbing infrastructure costs that have not been properly allocated, that is an outcome.

But what happens afterward matters too.

• Does the land remain open?
• Does another industrial project arrive five years later?
• Does development simply move several miles away?
• Does the tax base remain unchanged?
• Does the community need to raise revenue elsewhere to fund schools, roads, or public safety?

A decision should be judged not simply by what it stops, but by what it produces.

“No” is a policy choice, not the absence of one.

Frederick County illustrates the point. County officials did not merely reject a benefits package. They extended their pause on new applications while Maryland continues examining the larger impact of data-center development. That may ultimately lead to stronger rules, different projects, different land uses—or fewer data centers.

The important thing is to understand that every one of those outcomes carries consequences.

Why you say no matters.
Show your work.
Then evaluate what that choice actually produces.

Local decisions are valid, but they are not consequence-free.

III. THE INVISIBLE COST OF NOT BUILDING

It is easy to oppose infrastructure when the consequences of not building it remain invisible.

Most people never see the physical systems behind the digital world. A phone backs up photographs. A GPS application calculates a route. A bank processes a payment. A hospital retrieves a medical record. A business analyzes customer data. A streaming platform delivers a movie. An AI system answers a question.

Behind all of them are buildings, servers, fiber networks, electrical systems, cooling systems, and people keeping the infrastructure operating.

Using those services does not obligate a community to approve a badly designed project. But collectively, increasing digital demand means the infrastructure has to exist somewhere.

And that demand remains unusually strong.

CBRE reported that vacancy across major North American data-center markets remained only 1.4% in the first half of 2026, while more than 80% of capacity already under construction was preleased. Less than 1,500 MW remained available for preleasing—roughly six months of supply at the prevailing pace of demand.

That does not mean demand will grow forever at today's rate. It does mean that canceling one project does not cancel the underlying need for computing capacity.

And the use cases are becoming broader than consumer convenience.

The FDA and European Medicines Agency now recognize AI's potential to transform drug development and evaluation. The Department of Energy's Genesis Mission is applying AI and advanced computing to biotechnology, materials science, advanced manufacturing, nuclear energy, grid modernization, and scientific discovery.

The infrastructure conversation is therefore no longer just about Netflix, cloud photos, or chatbots. It increasingly touches research, medicine, manufacturing, energy, science, and economic productivity.

The cost of inadequate infrastructure may not appear as a line item in next year's municipal budget. It may appear gradually—in delayed projects, constrained capacity, higher costs, slower deployment, or investment occurring elsewhere.

Those costs are harder to see. That does not make them unreal.

IV. “SOMEWHERE ELSE” CAN MEAN ANOTHER COUNTRY

Data centers are global infrastructure.

The United States remains the largest data-center electricity consumer, accounting for about 45% of global data-center electricity consumption in 2024, compared with roughly 25% for China and 15% for Europe. The International Energy Agency projects that the United States and China together will account for nearly 80% of global data-center electricity-demand growth through 2030.

The AI competition is similarly international.

Stanford's 2026 AI Index reports that the United States still produces more notable AI models and leads private AI investment, while China leads in AI publication volume, citations, and patent output. It also finds that the performance gap between leading U.S. and Chinese models has narrowed into the low single digits.

That does not mean every local data center has national-security significance. Nor does it mean a community should approve a project because another country is building one. But it does mean that America's infrastructure decisions occur within a global market.

Power availability has become particularly important. JLL identifies speed to power as the primary site-selection criterion for data centers, followed by community support, latency, and customer proximity.

And companies have alternatives.

Google recently announced a €13 billion investment in Finnish digital and AI infrastructure over the next two years. Amazon has announced €33.7 billion of planned investment to expand cloud and AI data-center infrastructure in Spain.

Those investments are not evidence that American companies are abandoning the United States. They demonstrate that capital can move toward places offering the combination of power, infrastructure, customers, land, climate, policy, and predictability that projects require.

That raises a practical question for Americans as well. When U.S. companies build major data-center campuses abroad, the computing capacity may still serve global customers—including Americans—but much of the associated construction spending, utility investment, local tax revenue, workforce development, and supporting infrastructure accrues to the host community. If America needs this capacity anyway, how much of that investment do we want to capture here—and how much are we willing to see built elsewhere?

“Somewhere else” may mean another county. It may mean another state. And sometimes it may mean another country.

The current federal government has explicitly made domestic data centers, semiconductor facilities, energy generation, and grid expansion part of its AI infrastructure strategy. Other administrations may choose different policies, but the underlying strategic question will remain:

How much advanced computing capacity does the United States want to be able to build at home?

Local consent and national capacity do not have to be opposing principles. In fact, durable national capacity probably depends on communities believing the projects built around them are responsible, transparent, and worthwhile.

V. THE BETTER QUESTION: WHAT KIND OF FUTURE ARE WE CHOOSING?

The first version of a development proposal is not necessarily the only version.

Article IV examined how negotiation can change the bargain.

• Smaller footprints.
• Better siting.
• Closed-loop or reclaimed-water systems.
• Developer-funded infrastructure.
• Ratepayer protections.
• Different tax arrangements.
• Schools.
• Workforce programs.
• Environmental protections.
• Decommissioning plans.

Maryland offers a timely reminder that even major concessions do not guarantee approval. But it also demonstrates something important:

Communities have agency before reaching yes or no.

Sometimes the best alternative to a bad data-center proposal is not simply no data center. Or it may be a better data-center proposal. And sometimes, after that negotiation, the answer may still be no. That is what responsible decision-making looks like.

The final test can be surprisingly simple:

What happens if we say yes?
What do we gain, what do we give up, who pays, and what protections exist?

What happens if we say no?
What are we preserving, and what opportunities do we forego?

What happens instead?
What is the realistic alternative for the land, infrastructure, and community?

And what happens if everyone makes the same choice?
Where does the investment, infrastructure, computing capacity, and innovation ultimately go?

None of those questions automatically dictates an answer. Together, they produce a better decision.

THE DATA-CENTER DIVIDE

This series began with a simple idea:

First understand the thing. Then judge it.

We learned that the cloud has a physical address.

We examined the honest ledger of benefits and costs.

We studied Loudoun County—not as a perfect model, but as a community with decades of experience from which others can learn.

We looked at how communities can negotiate a new bargain before construction begins.

Now we reach the question underneath all the others:

Compared with what?

America does not need communities that say yes to everything. Nor is reflexive opposition a strategy for building the infrastructure an increasingly digital economy requires.

It needs communities capable of distinguishing good projects from bad ones, protecting what should be protected, demanding better terms, comparing realistic alternatives, and understanding that today's local decisions accumulate into tomorrow's national capacity.

That is not a call to build everywhere. It is a call to decide wisely.

Understand. Evaluate. Learn. Negotiate. Compare. Decide. Build wisely.

Because the future will require infrastructure.

The question is whether we will shape it—or merely react to it.

— Matt Cucinotta | Growth Solutions KC | Inspire · Inform · Ignite