Imagine walking into a polling station in a quiet Midwest town of fewer than a thousand people and casting a vote that somehow becomes the only green light the data center industry received across the entire country on Election Day. That is exactly what happened in Harwood, North Dakota. Out of 45 separate ballot measures dealing with data centers or large industrial projects tied to them, this one small community stood alone in siding with the developers. The rest of the nation sent a clear message of caution, delay, or outright rejection.
A Rare Yes in a Sea of No
I keep coming back to that number. Forty-five votes. One approval. It feels almost statistical noise until you dig into the details of what actually played out in Harwood. The town sits about ten miles from Fargo and counts roughly 940 residents. On election night, around 430 people showed up to vote. Three hundred twenty-five of them rejected a charter amendment that would have forced extra independent reviews and public hearings for big industrial projects. That works out to about 75 percent against the measure. In other words, the majority chose not to throw new procedural roadblocks in front of large developments already in motion.
The target of the proposed change was Applied Digital’s Polaris Forge 2 project, a planned $3 billion AI campus expected to draw 280 megawatts of power. Construction had already started. The company had locked in a long-term lease arrangement worth roughly $5 billion over fifteen years for 200 megawatts with a major hyperscaler, and financing of more than $2 billion had closed. Local officials had approved the project months earlier, and the concrete was already pouring.
How the Measure Reached the Ballot
A local group called Voices of Harwood drafted the amendment early in the year. They collected more than twice the required signatures in just three days. Their organizer, a software developer who has lived in the area for nearly two decades, later noted that the final language on the ballot looked quite different from the version his group originally submitted. Still, the intent remained clear: give residents a stronger formal voice before projects of this scale move forward.
The group insisted the fight was never purely about stopping one data center. It was about process. People wanted meaningful input on developments that reshape land use, power demand, and the character of a small community. One resident put it plainly: people deserve a say in projects of this scope. That sentiment echoes across many of the other 44 measures that failed for the industry elsewhere.
Local Leaders and Business Interests Push Back
Town officials and the regional chamber of commerce actively opposed the amendment. They warned that passage could cost taxpayers hundreds of thousands of dollars in lost opportunities and higher annual taxes. The mayor publicly sided against the measure. Council members expressed concern that the company might try to de-annex the project site and take the anticipated tax revenue with it. Officials projected the data center could generate $1.65 million in taxes by 2028. Those numbers carried weight in a community of this size.
I’ve watched similar debates unfold in other towns. The promise of stable property tax revenue often becomes the decisive argument, especially when budgets feel tight and schools or services need support. In Harwood, that argument appears to have landed. Residents ultimately decided the existing approval process was enough and that additional layers of review carried more risk than reward.
Why North Dakota Looked Different
North Dakota has cultivated a reputation as a place that welcomes large energy-related projects. Cooler climate, available power, and a generally pro-business regulatory posture all play roles. The state even offers sales tax exemptions on computers, cooling systems, and servers for qualifying facilities. Company representatives specifically cited those factors when explaining the decision to locate in Harwood.
That backdrop matters. In many other states, voters and local boards have grown wary of the water use, noise, visual impact, and strain on the electrical grid that large AI campuses bring. Some communities have passed temporary moratoriums. Others have tightened zoning rules or demanded more detailed environmental reviews. The broader political climate after recent elections has only heightened those concerns among certain groups of investors and developers.
Yet in this corner of North Dakota, the combination of early construction momentum, projected tax benefits, and a community that seemed less inclined toward pure opposition produced a different outcome. One local discussion program even asked why the national wave of resistance seemed to stop at the state line. The conversation circled around orderly development versus fear-driven reactions. Residents in Harwood apparently preferred the former.
The Bigger Picture for AI Infrastructure
Data centers have become the physical backbone of the artificial intelligence boom. Every major model training run and every cloud inference workload eventually lands in a building full of servers that consume enormous amounts of electricity and require sophisticated cooling. Hyperscalers and specialized operators are racing to secure sites, power contracts, and financing. That race collides with local democracy in real time.
When 44 out of 45 measures go against the industry, something fundamental is shifting. Communities are no longer accepting these facilities as automatic economic wins. They want clearer answers about long-term electricity rates, water consumption, noise levels at night, and what happens to the land after the useful life of the facility ends. Some also question whether the permanent jobs created justify the scale of the investment and the public incentives often attached.
In my view, the Harwood result does not reverse that broader trend. It simply shows that outcomes still depend heavily on local context. Timing matters. Construction already underway changes the political calculation. Credible tax projections change it further. A state tax policy that actively encourages these projects changes it again. Remove any of those factors and the vote might have flipped.
Process Versus Outcome
One of the more interesting threads in the Harwood story is the gap between what the citizen group originally proposed and what finally appeared on the ballot. Process disputes often decide these fights as much as the underlying merits. When residents feel the language was altered or that the window for input closed too quickly, trust erodes. When officials can point to existing public meetings and formal approvals, they gain ground.
The software developer who helped organize the signature drive emphasized long-term implications that still needed fuller examination. His perspective reflects a common tension. Once a multi-billion-dollar project breaks ground, reversing course becomes extremely difficult. The moment for careful scrutiny often arrives before the first shovel hits the dirt, not after financing closes and concrete trucks arrive.
People deserve a say in projects of this scope.
That simple statement captures the core demand running through most of these local contests. It is not always a demand to stop development. Frequently it is a demand for transparency, independent analysis, and genuine opportunity to influence the final shape of a project before it becomes a fait accompli.
Economic Arguments That Still Carry Weight
Tax revenue remains the most potent local selling point. In a town of 940 people, an additional $1.65 million a year by 2028 is not abstract. It can mean better roads, stronger schools, or lower pressure on residential property taxes. Chamber of commerce materials that warned of hundreds of thousands in extra costs if the measure passed likely reinforced that message.
At the same time, the industry has learned that pure economic promises no longer close every deal. Residents increasingly ask whether the tax revenue will offset higher electricity rates for everyone else, or whether water withdrawals will affect nearby wells and agricultural users. Those questions rarely receive simple answers, which is why so many ballot measures have tilted toward caution.
I’ve found that the most successful projects tend to address these concerns early and specifically rather than treating them as afterthoughts. When a developer arrives with detailed power procurement plans, water recycling commitments, and clear noise mitigation designs, the conversation changes. When those details remain vague, skepticism grows.
What the National Pattern Suggests
Looking across the full set of 45 measures, a pattern emerges. Communities with strong existing industrial bases or energy production histories sometimes prove more receptive. Places already strained by rapid growth or water scarcity prove far less so. Political messaging around energy costs and corporate influence also shapes the atmosphere. After recent election results, some market observers openly worried that more jurisdictions would move toward temporary pauses or stricter review standards.
Harwood’s decision offers a counter-example rather than a new national direction. It demonstrates that when construction has begun, financing is secured, and local officials remain unified in support, voters can still choose continuity over additional process. It does not prove that the broader resistance has faded.
Perhaps the most interesting aspect is how quickly these local decisions now ripple into financial markets. Large technology and infrastructure stocks have shown sensitivity to the political risk surrounding data center permitting. A single high-profile rejection can move sentiment. A solitary approval, by contrast, rarely reverses the wider narrative.
Lessons for Future Projects
Developers watching this outcome will likely draw several practical conclusions. First, early and continuous engagement with residents matters more than ever. Second, state-level incentives and tax treatment can still tilt the balance in favorable jurisdictions. Third, once physical work is underway, the political dynamics shift in the project’s favor. Fourth, citizen-led ballot measures remain a real risk in many places and require careful legal and communications responses.
Communities, for their part, continue refining the tools they use to assert control. Charter amendments, zoning overlays, temporary moratoriums, and enhanced review requirements all appear with growing frequency. The Harwood vote shows that these tools do not always succeed, yet their presence alone changes how companies approach site selection.
- Secure power and water agreements with transparent local impact assessments
- Begin formal community outreach well before seeking final approvals
- Quantify tax benefits in concrete, multi-year projections
- Address noise, lighting, and visual screening as core design features
- Anticipate citizen ballot initiatives in jurisdictions that allow them
None of these steps guarantee success. They simply raise the odds in an environment where local consent can no longer be taken for granted.
The Human Element Behind the Numbers
Behind every ballot total sit individual stories. Long-time residents who worry about the character of the place they raised their children. Newer arrivals who moved for quiet and open space. Local business owners who see potential customers or suppliers. Elected officials balancing budgets against quality-of-life concerns. Software developers who suddenly find themselves organizing signature drives. All of them showed up in Harwood.
The 75 percent margin suggests that, on balance, the majority concluded the existing path forward made sense. That conclusion may look different in other towns facing similar proposals. Geography, culture, energy markets, and recent experience with large projects all shape the final judgment.
What remains consistent is the underlying tension. Massive capital projects that support the digital economy now routinely confront the slower, more deliberative pace of local democracy. Sometimes the capital side wins. Sometimes the deliberative side prevails. On this particular Election Day, the capital side won exactly once.
Looking Ahead
The AI infrastructure build-out is far from finished. Power constraints, chip availability, and financing conditions will continue to drive site selection. Local political risk has become a permanent variable in that equation. Companies that treat community relations as a compliance checkbox rather than a strategic priority will face more nights like the 44 that went against them. Those that invest early in understanding local priorities may find more outcomes resembling Harwood’s.
I do not expect a sudden wave of approvals simply because one small North Dakota town voted the way it did. The broader pattern of caution looks durable. Yet the Harwood result reminds us that local conditions still matter enormously. Cool climate, available power, favorable tax treatment, and a project already under construction created a unique alignment. Elsewhere the alignment is different, and the votes have reflected that reality.
In the end, the story is less about a single yes vote and more about the growing friction between national technology ambitions and the places asked to host them. That friction will keep producing ballot measures, council debates, and late-night signature drives. Most of those contests will remain contested. A few will end the way Harwood did. Tracking which ones succeed, and why, offers the clearest window into how the physical foundations of artificial intelligence will actually get built.
The next round of local decisions is already forming. Developers are scouting new sites. Citizen groups are studying charter language. Officials are calculating tax tables. Somewhere in another small town, another group of residents will soon decide whether the promised benefits outweigh the permanent changes. Their answer will join the tally that currently stands at one yes and forty-four expressions of hesitation. The final score of this larger contest remains very much unwritten.
