What Data Center Developers Could Learn From a Decade of Solar's Hardest Fights

On April 16, the rural Wisconsin town of Cassville voted 44-0 to ban data center development inside its limits. The project on the table was worth a billion dollars. It promised 50 jobs and $5.5 million a year in property taxes — meaningful money for a town of about 800 people in the Driftless Area. The vote took place in a town garage, with an attorney reading the ordinance from a laptop balanced on a snowplow.

The proposed developer was anonymous. Nobody in the room could tell you which company actually wanted to build the data center, only that someone did, and that they wanted 500 acres and 400-500 megawatts of power.

The vote wasn't really about the data center. It was about who would build it, and the fact that nobody in Cassville knew.

Anyone who watched utility-scale solar fight its way through rural America between roughly 2018 and 2022 has seen this movie before. The opposition arguments are nearly identical. The procedural mistakes developers make are nearly identical. The outcomes are starting to look identical too. The data center industry is repeating the original solar industry's playbook, including the parts that didn't work — and at this point, hyperscalers don't even have the excuse of going first. Solar developers already wrote the lessons learned. They wrote them the hard way, by losing.

The Pattern Is Familiar

In Q2 2025 alone, an estimated $98 billion in data center projects were blocked or delayed by community opposition, moratoriums, and litigation — more than every previous quarter combined since 2023. The full-year 2025 number reached $156 billion. The trend is accelerating.

It is also, almost beat-for-beat, the same trend the solar industry lived through between roughly 2018 and 2022.

The arguments are nearly word-for-word. Solar opponents talked about glare, visual impact, drainage, decommissioning uncertainty, and the conversion of farmland to industrial use. Substitute noise, light pollution, water consumption, and warehouse-scale buildings, and you have the data center opposition's bullet points. Both industries face the same skepticism about construction jobs that vanish on schedule, leaving a fenced site staffed by a handful of technicians. Both face the same complaints about utility rate impacts — solar through interconnection upgrade costs, data centers through the load growth that is already pushing residential bills up by $16 to $18 per month in parts of PJM. The specific physical concerns differ. The structure of the objection is the same.

The opposition tactics are also nearly identical. Citizen groups form around a single proposed project, then evolve into permanent local organizations that show up at every planning meeting. Model ordinances banning or restricting development circulate online. Once one county passes a moratorium, neighboring counties consider their own — often using the first county's ordinance as a template. The county-by-county map of solar moratoriums that took shape between 2019 and 2022, concentrated heavily in the Midwest and Southeast, looks remarkably similar to the data center moratorium map taking shape now. Some of the same counties even appear on both.

The demographic pattern matches too. Both industries are running into the hardest opposition in rural and exurban communities with strong agricultural identity, where the proposed project is the largest single land-use change anyone in the room has voted on in a generation. The political coalitions that organize against utility-scale solar — farmland preservation groups, conservative property-rights advocates, environmental neighbors worried about local impacts, residents who simply don't like change — show up in the same combinations against data centers.

And the developer-side mistakes are familiar enough that the playbook could be written from memory. Land gets acquired through anonymous LLCs. Community engagement begins after zoning is already on the calendar. The public pitch leads with construction jobs that everyone in the room understands are temporary, while concrete mitigation for water, noise, or electricity concerns gets treated as a topic for later rather than something committed to in writing now. The planning commission hearing becomes the place where the developer first hears what the community actually thinks — which is much too late.

The solar industry made every one of these mistakes, often repeatedly, in different jurisdictions. It eventually changed how it operated because the alternative was watching pipelines die county by county. Some of those changes are now standard practice in mature solar development. Almost none have been adopted by the data center industry yet.

Lesson 1: Put Your Name on the Project

The Cassville project never had a public sponsor. Neighboring Van Buren Township, Michigan is fighting a 1-gigawatt facility called "Project Cannoli." Other recent projects have circulated under similarly opaque code names. The shell-LLC pattern is now standard in data center site selection.

It's also the single most reliable way to lose a community fight.

Solar developers tried this in the early days too. Buying land through anonymous LLCs to avoid alerting competitors or driving up prices made sense from a business development perspective. It also generated the exact distrust that translates into 44-0 ban votes. The industry largely abandoned the practice for projects above a certain scale, because the cost of community backlash exceeded the cost of bidding against a known competitor.

Microsoft announced in March that it would stop using NDAs with local governments globally for data center projects. The framing was generous to the company — "strengthening public trust" and "meaningful engagement." A more honest framing came from Bill Lueders of the Wisconsin Freedom of Information Council: it's happening only after their attempts to operate behind NDAs blew up in their face.

Solar reached the same conclusion roughly five years ago. Data centers are catching up.

Lesson 2: Lead With Tax Revenue, Not Jobs

Cassville was offered 50 jobs at a $1 billion facility. That's not a typo. And it's exactly why "jobs" is usually the worst argument a data center developer can lead with.

Data centers are not labor-intensive in operation. The construction phase creates substantial short-term employment, but those jobs leave when the construction does. The permanent operational staff at a hyperscale facility is small — typically a few dozen technicians for a multi-hundred-megawatt site. That ratio is just structural to the technology.

Solar developers learned this lesson early. Selling a utility-scale solar project on operations-phase jobs almost guaranteed losing the argument, because the numbers were never going to look great relative to the project footprint. The pivot was to lead with tax revenue and infrastructure investment — both of which utility-scale solar genuinely delivers, often at scales that dwarf comparable industrial development.

Data centers have an even stronger version of this argument available. A $1 billion data center generates more property tax revenue than almost any other land use a rural county can attract. That's the headline. Promising jobs you can't deliver, while burying the tax revenue that you actually can, gets the argument backwards. It also gives opponents an easy line of attack when the operations staffing announcement inevitably underwhelms.

Lesson 3: Address the Real Concerns, With Real Mitigation

The community concerns about data centers aren't speculative. Water consumption at a 1-gigawatt facility can run 2-3.6 million gallons per day. Goldman Sachs analysis found data center demand is on track to push residential electricity bills up by $16-18 per month in some PJM-region states. Backup generator noise and emissions are real. Truck traffic during construction is real.

The solar industry handled its analogous list — visual impact, glare, agricultural land conversion, decommissioning concerns — by developing standardized mitigation. Setbacks from property lines. Vegetation buffers. Pollinator habitat. Bonded decommissioning agreements. Agrivoltaic options. None of these eliminated opposition, but they took the easy arguments off the table and forced the conversation onto narrower ground.

Data centers have analogous mitigation available, and the industry's failure to standardize and lead with it is one of the avoidable mistakes. Closed-loop cooling reduces water consumption dramatically. Behind-the-meter generation can isolate a project from grid impacts. Sound-attenuated generator enclosures exist. Setback distances and visual screening for warehouse-scale buildings are well-understood. The technology is there. The willingness to commit to it in writing, before zoning approval rather than after, is what's missing.

Microsoft's five-point Community-First AI Infrastructure Plan is the most concrete example so far of a major operator codifying these commitments — paying electricity rates high enough to cover the new infrastructure they require, replenishing more water than they use, declining property-tax abatements. It is also, notably, the work of a single company. The rest of the industry has not converged on the same standards.

Lesson 4: Show Up Before Zoning, Not At Zoning

The most common solar industry failure of the 2018-2020 era was arriving at a planning commission hearing with a fully designed project, no prior community engagement, and a presentation deck. The result was almost always a no vote, sometimes a moratorium that locked the developer out of the entire jurisdiction for two years or more.

The fix was pre-application engagement: listening sessions, individual meetings with neighbors, visits to schools and fire departments, modifications to the project based on what came out of those conversations. The cost in developer hours was significant. The cost of skipping it was projects getting killed at zoning, which is far more expensive.

Data center developers are largely repeating the original solar industry mistake. The Cassville vote happened in the absence of any meaningful public engagement by the project sponsor — because there wasn't a publicly identified project sponsor in the first place. Hermantown, Minnesota residents are suing the city over a Google data center comprehensive plan amendment they argue was passed "in secret and in bad faith to assist the developer." That's the Cassville sequence with a different cast.

The industry-wide pattern is that meaningful community engagement begins after a moratorium has already been proposed, when the developer is reactively trying to save a project. By then, trust is gone and the conversation is structurally adversarial.

Lesson 5: Build Coalitions Instead of Just Defending Yourself

Solar developers learned, slowly, that being in a fight by themselves was nearly impossible to win. The successful playbooks involved building actual local coalitions — partnerships with school districts that benefited from tax revenue, fire departments that benefited from training and equipment investment, farm bureaus, county agricultural extension offices, sometimes labor unions for the construction phase.

Data center developers have done remarkably little of this. There's a reason: the model assumes that hyperscaler money plus property tax revenue is enough to carry the day, and that local political dynamics will cooperate. Cassville's 44-0 vote is one of dozens of recent counterexamples. The economic argument by itself doesn't beat a coalition of organized opponents. The economic argument plus a coalition of organized supporters can.

Building those coalitions takes time, attention, and a willingness to operate in local politics that most hyperscaler corporate development teams have not invested in. Solar developers had to learn it the hard way too.

The Window Is Narrower Than It Looks

The instinct in the data center industry is to assume the current backlash is a blip. AI demand is real and accelerating. Federal policy is broadly supportive. Some communities will say no, but plenty will say yes. The math, on a national scale, still works.

This is roughly what utility-scale solar told itself in 2018, and it was both true and irrelevant. Yes, the macro demand was real. Yes, plenty of communities approved projects. But every county-level moratorium that passed permanently took development capacity off the table for the duration of that moratorium, and frequently triggered cascading moratoriums in adjacent counties. The opposition organized, networked, and shared model ordinances. The local fights compounded into structural constraints on where and how fast development could happen.

That same dynamic is now visible in the data center sector. State legislatures are introducing data center moratorium and disclosure bills. Local moratoriums are spreading. The projects that get blocked don't just disappear — they shift the political environment for the next project, and the next.

The solar industry took roughly a decade to figure out community engagement. It didn't have a choice. Data centers don't have a decade. They have, at best, the next 18 to 24 months before the patchwork of state and local restrictions hardens into something much more difficult to develop around.

The lessons are sitting there. Whether the industry adopts them in time is a different question.