Rural communities are increasingly leveraging zoning authority and public opposition to block or delay large-scale AI data center developments.

Proponents of today’s data centers argue they bring much-needed jobs and economic boosts to local communities. Rural America has heard similar pitches before. Nearly four decades ago, concentrated animal feeding operations (CAFOs) arrived barn by barn, backed by bold promises of employment and expanded county tax bases. By the time residents organized, right-to-farm laws and state preemption had stripped them of leverage. Meanwhile, manure lagoons leached into groundwater and polluted the air, reducing life expectancy among farmworkers and downwind residents. Organizing continues to yield tangible victories, from North Carolina’s 1997 CAFO moratorium to Chris Jones’s current campaign for Iowa agriculture secretary on a clean-water platform. The lessons surrounding CAFOs arrived too late for many, but in the age of artificial intelligence, those same experiences can equip rural communities with a strategic advantage.

Unlike the gradual spread of CAFOs, data centers are arriving in a land rush. That velocity is both alarming and advantageous. Texas alone has approximately 248 proposed data center projects, despite polls showing 56% of Texans oppose having one nearby. Consequently, rural communities are raising questions before agreements are finalized. According to the Pew Research Center, of the more than 1,500 data centers proposed or under development nationwide, 67% are located in rural areas. The ultimate question remains: Do rural residents have the power to say no?

The answer is unfolding across the country. In November 2025, roughly 800 residents packed a board meeting to oppose a billion-dollar Meta-linked project in Howell Township, Michigan, prompting the developer to withdraw. In February, the San Marcos, Texas, city council rejected a $1.5 billion campus following nearly nine hours of public testimony, subsequently becoming the first Texas municipality to ban data centers via zoning code. Fort Worth is currently drafting its own data center zoning regulations. Nationally, the poorest census tracts resist data center proposals at five times the rate of the wealthiest. Projects facing opposition are canceled at several times the rate of those that proceed unchallenged. Industry trackers report tens of billions of dollars in blocked or delayed investments. Opposition to data centers remains one of the few issues commanding broad American consensus. A June Reuters/Ipsos poll revealed that only 14% of respondents supported building a data center in their community. On July 18, organizers coordinated 142 protests across 42 states in the first nationwide demonstration against data centers and artificial intelligence.

Hyperscalers are attempting to replicate the CAFO playbook. West Virginia has stripped counties and municipalities of regulatory authority over data centers, mirroring how the state’s right-to-farm law shields agricultural operations from local civil or criminal enforcement so long as they comply with state and federal permits and regulations. In Texas, when Hill County passed the state’s first county-level moratorium in May, a developer sued for $100 million, forcing the county to rescind the measure within a week—a direct result of Texas law prohibiting counties from zoning against either data centers or feedlots. Hood County later withdrew its own moratorium after State Senator Paul Bettencourt (R-Houston) questioned whether counties possessed that authority; Senator Bettencourt has also declared San Marcos’s ban illegal. Right-leaning advocacy groups are pursuing similar strategies. The American Legislative Exchange Council (ALEC), which authored pro-CAFO right-to-farm model legislation beginning in the 1990s, appears equally determined to restrict local control over data center development. At its July annual meeting, ALEC reviewed a model legislative framework designed to override local bans and moratoriums, granting data centers the “clear authority to secure power…subject only to objective safety and interconnection standards,” according to the Center for Media and Democracy. The event included speakers from Google and Nvidia and was sponsored by Meta.

Surging opposition ahead of the midterm elections has compelled political leaders to confront local control. This marks a key divergence from the CAFO era. In June, Texas Governor Greg Abbott—who previously described Texas as “the epicenter of AI development”—called for banning AI data centers in rural neighborhoods and reducing their billion-dollar annual tax exemption amid rare bipartisan pressure. Florida Governor Ron DeSantis signed legislation preserving local authority over data centers. Michigan’s gubernatorial candidates remain divided over server facilities expanding between Lansing and Detroit. Illinois recently reversed its data center policy, scaling back subsidies initially introduced in 2019. In July, New York became the first state to implement a one-year moratorium.

Which communities have been unable to say no? Researchers mapped CAFO and data center locations in North Carolina and Texas, where data is readily accessible, and utilized Claude to overlay the sites against demographic and county boundaries. Mapping the distribution reveals stark contrasts. In North Carolina, the pattern is unmistakable. The state’s 2,482 permitted animal operations concentrate in the Black Belt coastal plain, the same region where epidemiologist Steve Wing documented hog lagoons generations ago. When weighted by permit volume, these counties are 46% people of color, compared to 34% statewide across North Carolina’s 100 counties. Duplin County alone accounts for over 500 permits and 2.2 million allowable hogs in a county of 49,000 residents, nearly half of whom are people of color. Data centers follow an inverse geographic pattern. North Carolina’s facilities are located in counties whiter than the state average, including Piedmont towns that lost their historic furniture and textile industries, as well as affluent Wake and Mecklenburg counties, which contain Raleigh and Charlotte, respectively. Two suspended projects are situated in neighboring Edgecombe and Pitt Counties, where residents drew upon their historical experience opposing CAFOs to organize resistance. This history of resistance proved decisive. In 1995, predominantly Black residents of Kingsboro in Edgecombe County successfully prevented a major meatpacker from constructing a 300-acre slaughterhouse by mobilizing to rezone the land. Three decades later, when a $19 billion, 400-acre data center proposal emerged, residents mounted a unified opposition. Developer Energy Storage Solutions withdrew its application in July, and by August, Edgecombe County had implemented a two-year moratorium.

Texas presents a more complex demographic picture. While rural Texas generally skews whiter than the state average—which is moderated by majority-Hispanic urban centers—the Panhandle feedlot region is majority-Hispanic meatpacking territory. Counties such as Deaf Smith and Castro feature workforces living adjacent to the feedyards they operate. Similarly, rural counties currently hosting data centers are less white than rural Texas overall. Other notable outliers exist nationwide. The world’s densest concentration of data centers sits outside Washington, D.C., in Loudoun County, Virginia, one of the nation’s wealthiest regions. Conversely, xAI’s Colossus facility operates adjacent to Boxtown, a majority-Black neighborhood in Memphis. In Texas, the Stargate Oracle AI campus near Abilene will occupy approximately 4 million square feet and be operated by Oracle for OpenAI.

The geographic distribution and political dynamics surrounding data centers differ significantly from those of CAFOs. Historically, CAFOs were sited alongside marginalized communities lacking political leverage. Most counties proposing data centers lack entrenched local billionaires or multi-generational landholding families willing to champion these projects and shield vulnerable populations from disproportionate harm—a dynamic that repeatedly characterized CAFO expansion. Site selectors are rapidly targeting data center locations based on inexpensive land and water, affordable energy, and minimal opposition. Unlike CAFO developers, however, hyperscalers lack entrenched local advocates wielding significant political influence. Race and class remain the oldest American indicators of which communities possess the power to say no—and which do not.