AI’s Hidden Footprint: Media Framing of Rural Data Center Expansion

AI’s Hidden Footprint: Media Framing of Rural Data Center Expansion | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article AI’s Hidden Footprint: Media Framing of Rural Data Center Expansion Jingyuan Zhang, Rui Liu, Sunil Dhoubhadel, Rodolfo Nayga This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9442287/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The artificial intelligence boom relies on the rapid expansion of data centers, which provide the infrastructure necessary for data storage and computational power to support AI services and development. These facilities consume substantial resources to sustain their operations. Technology companies are increasingly locating data centers in rural regions to take advantage of resource availability and lower costs, creating new opportunities while also introducing challenges for local communities. This study provides an initial exploration of the environmental and socioeconomic implications of this trend by examining how media frame rural data center expansion in its early stages. We find that media coverage has grown rapidly, with an increasing dominance of negative framing. Media narratives highlight conflicts between data center development and rural residents over resource competition, environmental concerns, community identity, and local governance processes. These dynamics carry significant environmental and socioeconomic implications for rural regions. They may alter human-nature interactions, disrupt agricultural production, and affect income stability and living standards, while potentially exacerbating economic inequality and social justice concerns. Our findings highlight the need for future research and policy efforts to better balance technological advancement with rural development, including supporting community sustainability and improving transparency in governance processes to promote accountability and public trust. Agricultural Economics & Policy Environmental Policy Social Policy Environmental Economics Data center expansion rural development agriculture energy-water-land nexus resource competition media framing Figures Figure 1 Figure 2 Figure 3 Figure 4 1. Introduction Artificial Intelligence (AI) has experienced rapid growth in recent years. In 2024 alone, corporate investment in AI exceeded 252.3 billion U.S. dollars (Stanford Institute for Human-Centered AI 2025 ), and chatbot applications such as ChatGPT have surpassed 2 billion visits and 500 million users worldwide (Liu and Wang 2026 ). Advances in machine learning, data processing, and computing power have enabled technology companies to integrate AI into a wide range of applications, transforming technological progress, business operations, and social development. In business markets, AI has improved production processes, decision-making, market analytics, and automation. For the general population, AI is increasingly embedded in everyday life through personalized services, influencing how people access information, interact with digital platforms, and make social decisions. The infrastructure underlying this AI boom is the data centers, which store vast amounts of data and provide the computational power required for AI operation and development. Data centers are large-scale industrial facilities that house high-density computing systems, servers, and supporting equipment, including generators, power supplies, and environmental control systems. Maintaining their continuous operation requires substantial resource inputs, particularly energy and water (Lei et al. 2025 ; Jegham et al. 2025 ). For example, a single large data center can consume tens to hundreds of megawatts of electricity and use millions of gallons of water per day for cooling, levels comparable to those of a medium-sized city. To meet the growing demand for AI services, technology companies have been rapidly expanding data center infrastructure, emphasizing its importance for market competitiveness, technological advancement, and national security (Makholm et al. 2025 ). In the United States, which accounts for 54% of global data center capacity, the number of data center facilities has more than doubled since 2020, with both their number and scale increasing substantially (Synergy Research Group 2025 ). While the early development of these facilities was concentrated in urban areas, recent trends show a growing shift toward rural regions, including small towns and communities (Lee 2026 ; Shin 2026 ; Hassan 2026 ), a pattern often attributed to greater resource availability and lower costs. For rural counties and towns, data centers represent extremely high-value investments, often ranging from hundreds of millions to billions of dollars. Their development is therefore expected to generate employment opportunities and increase tax revenues, supporting local infrastructure and broader economic growth. However, their expansion into rural regions also introduces important trade-offs. The key resources required for data centers, namely the energy-water-land nexus, are fundamental to rural landscapes and livelihoods. Introducing data centers that demand these resources at the scale of a medium-sized city into rural regions, where infrastructure is designed for dispersed populations, can strain local systems, compromise infrastructure reliability, intensify resource competition, and increase living costs. Developing data centers in rural regions reflects an emerging trend with significant natural and socioeconomic implications. It introduces a new dimension to human-nature interactions by placing pressure on rural areas, which host a large share of natural capital, including land and water resources, in most countries, and may exacerbate existing resource constraints. This expansion can directly affect the economic backbone of rural economies, namely agricultural production, by raising farmers’ production costs and weakening the economic capacity of local populations. The resulting impacts may also extend to the social sphere, raising concerns about distributional justice and the disproportionate burden placed on rural communities, as capital-intensive technology companies take advantage of low-cost rural resources. Existing literature on the impacts of AI and data centers has primarily focused on energy use, natural resource consumption, and environmental concerns. For example, Jegham et al. ( 2025 ) assess the environmental footprint of AI by examining data center consumption of energy, water, and carbon emissions. They find that even short queries at large scale, such as 700 million queries per day, can consume electricity comparable to that used by 35,000 U.S. households annually, require water equivalent to the needs of 1.2 million people, and generate carbon emissions that would require a Chicago-sized forest to offset. They argue that AI development drives disproportionately high resource consumption and poses challenges for sustainable development. Similarly, Siddik et al. ( 2021 ) discuss the substantial resource demands of data centers in the United States, showing that many facilities draw water from medium- and highly water-stressed watersheds and generate substantial greenhouse gas emissions, raising concerns about long-term sustainability. On the other hand, the socioeconomic dimensions of data center expansion remain largely underexplored. There is no research examining the roles, challenges, and evolving dynamics faced by rural populations, who have historically been economically vulnerable in many countries and underrepresented in academic research. What are the primary concerns of rural communities regarding this trend? What types of interests and conflicts emerge in local decision-making processes? What are the broader social and economic implications? Without a clear understanding of the full scope of this issue, it is difficult for researchers and policymakers to develop informed research agendas and effective policy responses. This study aims to fill this gap by providing an initial exploration of AI’s infrastructural externalities and the broader natural and socioeconomic impacts of data centers on human–environment interactions. In the absence of a well-established academic literature, we conduct the first systematic descriptive media analysis, which leverages the important role of media in reflecting stakeholder concerns, shaping public perceptions, and influencing political outcomes (Andsager 2000 ; Tewksbury et al. 2000 ; Tan and Weaver 2009 ). Specifically, we examine how media outlets frame the rural expansion of data centers and analyze their thematic focus, using a dataset of newspaper articles published between January 2024 and February 2026 that captures the early stage of this expansion. Our findings provide timely insights to guide future research on the natural and socioeconomic implications of data centers and help policymakers consider trade-offs when designing balanced regulatory frameworks that support technological advancement alongside sustainable environmental and social development. 2. Data The media data are obtained from Nexis Uni, a comprehensive database that archives newspaper articles published worldwide. To focus on rural data center expansion, we adopt a two-pillar keyword strategy[1]. The first pillar captures the object of interest using the term “data center.” The second pillar captures the rural context using the following keywords: “rural,” “small town,” “small towns,” “agricultur*,” “farm*,” “unincorporated,” and “community.” The wildcard operator (*) allows the search to capture variations such as “agriculture,” “agricultural,” “farm,” and “farming.” Articles are retained only if keywords from both pillars appear in either the headline or the lead paragraph. The search is restricted to newspaper articles published in the United States, written in English, and dated between January 1, 2024, and February 28, 2026. Using this query, a total of 1,482 articles were retrieved in March 2026. To ensure that data centers are the primary focus of the articles, rather than a passing reference or part of a broader discussion, we apply a term frequency threshold (Zhang et al. 2012; Ibrahim and Landa-Silva 2016). Specifically, we retain articles in which the term “data center” appears at least three times in the body text. This threshold is informed by the average article length of 1,078 words. The final sample consists of 519 newspaper articles[2]. The distribution of the sample by publication year and geographic location is reported in Table 1. Table 1. Sample characteristics Articles (N = 519) Publication Year 2024 89 (17.1%) 2025 291 (56.1%) 2026 139 (26.8%) Geographic Region Midwest 162 (31.2%) South 141 (27.2%) Northeast 102 (19.7%) West 35 (6.7) Not labeled 79 (15.2%) [1] A three-pillar keyword strategy incorporating a sentiment dimension, such as keywords including “support,” “oppose,” or “protest,” is not adopted because it may introduce selection bias in the retrieval of articles. [2] As a robustness check, we conduct a validation exercise in which a random sample of 100 articles is drawn from the full set of 1,482 retrieved articles. Each article is manually reviewed to determine whether data centers constitute the primary focus. The human validation yields an 87.0% agreement rate with the term frequency method. 3. Sentiment Analysis Evaluating the tone of newspaper coverage and its evolution over time provides insights into how the media perceive rural data center expansion and how public perceptions may be shaped. We assess media tone using a sentiment analysis approach. Two independent coders reviewed the sample articles following a standardized protocol and classified each article into one of three mutually exclusive categories: negative, positive, or neutral. An article was coded as negative if it primarily emphasized adverse impacts, local conflicts, or risks associated with data center development, with substantially less or no discussion of potential benefits. Conversely, an article was coded as positive if it highlighted benefits, such as economic opportunities and gains, with limited discussion of potential drawbacks. Articles were classified as neutral if they presented information in a factual tone without emphasizing either positive or negative aspects, or if both perspectives were discussed in a balanced manner. The resulting confusion matrix is presented in Table 2. The independent coding produced largely consistent results, with an inter-coder agreement rate of 87.0% and a Cohen’s Kappa of 0.79. To avoid overstating either positive or negative coverage, articles with coder disagreement were classified as neutral. The final classification of the 519 articles includes 250 negative, 110 positive, and 159 neutral articles. Table 2. Confusion matrix of the sentiment analysis Coder1/Coder2 Negative Neutral Positive Negative 250 13 0 Neutral 17 93 15 Positive 0 26 110 Figure 1 presents the monthly publication frequency of articles by tone category. Prior to June 2025, the number of articles discussing rural data center expansion remained relatively low and stable, with similar levels of negative, neutral, and positive coverage. This period likely reflects the early stage of rural data center expansion, when local communities had not yet formed strong or consistent reactions and were exposed to both positive and negative narratives. During this stage, many projects were still under development and had not yet fully affected local communities. Since June 2025, however, the number of negative and neutral articles has increased substantially, while the number of positive articles has declined. Several factors may contribute to this shift. As large-scale data center development accelerated across multiple rural regions in early 2025, increased demand for energy and water resources, particularly during peak summer periods, may have made resource constraints and associated impacts more visible to local communities. Additionally, the rapid scaling of computing technologies and AI infrastructure may have intensified resource consumption relative to earlier periods, reinforcing negative perceptions. Figure 2 presents the geographic distribution of articles and their average tone by state. The Mid-Atlantic and Midwest regions account for the largest share of media coverage, while the South and West contribute fewer articles. This pattern is consistent with the geographic concentration of early data center development. Across most states, coverage is predominantly negative, with positive tones appearing only in a few states with relatively lower publication volumes. 4. Thematic Analysis 4.1. Keywords Analyzing article keywords helps identify the main topics discussed in the media and provides insights into how these topics are interconnected. We conduct a word co-occurrence network analysis (Radhakrishnan et al. 2017), which identifies high-frequency terms that appear together within the same article and computes pairwise correlations among them.[3] The results are presented in Figure 3. The analysis reveals two primary clusters of keywords. The cluster located in the upper-right region of Figure 3 consists of articles focused on the energy dimension of data center expansion in rural areas, with prominent terms such as electricity, power, energy, and gas. The second cluster, located in the lower-left region, reflects broader discussions across a more diverse set of topics, including resource-related terms such as land and water, community-related terms such as residents and people, and local governance issues including councils and zoning. 4.2. Thematic Dimensions Several dimensions of rural data center expansion are of particular interest to researchers and policymakers. Investigating how the media frame these dimensions and what newspaper articles address within them can provide insights into public perceptions and concerns. Based on the most frequently occurring keywords identified in Figure 3 and prior studies on the resource demands of data centers (Jegham et al. 2025; Lei et al. 2025), we focus on the following dimensions: Jobs- referring to employment opportunities generated by data center development; Taxation- referring to local tax revenues associated with data center investments; Water, Energy, and Land- capturing resource consumption required for data center operations; Environment- referring to non-consumptive environmental impacts such as pollution, waste, and emissions; Community Identity- capturing nonmarket impacts on rural culture, heritage, and lifestyle; and Local Governance- referring to municipal decision-making processes such as zoning changes, council deliberations, planning approvals, and moratoriums. Each article is reviewed to determine whether it addresses these thematic dimensions based on topical salience (Boguraev and Kennedy 2002). For each dimension, articles are coded as either “yes” or “no.” An article is coded as “yes” only if it contains substantial discussion of the dimension, defined as the dimension serving as a primary narrative driver of the article, appearing in a dedicated paragraph, or recurring throughout the article. Detailed definitions, inclusion criteria, and exclusion rules are provided in Table 3. Table 3. Structured codebook for thematic dimensions Thematic Dimensions Definition and Inclusion Criteria Exclusion Criteria Jobs Employment associated with data center development, including job creation and local hiring initiatives. No discussion of the inclusion criteria. Taxation Local tax revenue and public funding generated by data centers. No discussion of the inclusion criteria. Water Water scarcity resulting from data center development and consumption, such as depleted wells. No discussion of the inclusion criteria, or references to water pollution or contamination. Energy Energy impacts associated with data centers. No discussion of the inclusion criteria. Land Land use changes and property value impacts resulting from data center development. No discussion of the inclusion criteria, or zoning discussions limited solely to governance procedures. Environment Environmental impacts of data centers, including pollution, waste, and greenhouse gas emissions. No discussion of the inclusion criteria, or discussions limited only to resource consumption and scarcity issues. Community Identity Impacts on rural community character, such as agricultural heritage, rural lifestyles, or concerns about visual landscape changes due to data centers. No discussion of the inclusion criteria. Local Governance Local governance processes related to data centers, including planning decisions, zoning votes, moratoriums, and municipal decision-making. No discussion of the inclusion criteria. Figure 4 presents the distribution of articles across thematic dimensions. The left panel reports the number of articles addressing each dimension, ranked from the most frequently discussed dimension, local governance, to the least discussed, land. The right panel presents the average difference in tone between articles that address a given dimension and those that do not. Because a single article may address multiple dimensions simultaneously, counts across dimensions are not mutually exclusive and may sum to more than the total number of articles (519). In the following sections, each thematic dimension is discussed in detail. 4.2.1. Jobs and Taxation Job creation and tax revenue are often viewed as key drivers of early local support for data centers. In the sample, 39.9% of articles discuss job creation, and 33.1% discuss tax revenue related to rural data center expansion. As shown in the right panel of Figure 4, although both groups of articles exhibit an overall negative tone, those that address jobs and taxation are, on average, more positive than those that do not. Discussions of job creation largely reflect statements from developers and local policymakers during project planning. However, a substantial share of articles question the effectiveness and magnitude of these benefits. Many argue that most jobs created by data centers are temporary construction jobs, such as electricians and construction workers, with relatively few permanent operational positions. For example, a proposed $10 billion Meta data center project announced in June 2025 near Rayville is expected to create 5,600 jobs, of which approximately 5,300 are temporary and only 300 are permanent. Articles further suggest that, due to the highly automated nature of data centers, permanent positions may be limited in number and scope, leading to concerns that employment benefits are overstated. A similar debate emerges regarding taxation. While many articles emphasize the potential for millions to billions of dollars in long-term tax revenues, others argue that tax incentives used to attract data centers, often lasting two to ten years, may offset expected fiscal gains. As a result, the net economic benefits to local communities may be more modest than initially anticipated. 4.2.2. Energy-Water-Land Nexus Energy, water, and land are often discussed jointly as part of a resource nexus that supports rural livelihoods and agricultural production. The expansion of data centers, which relies heavily on these same resources, has therefore raised concerns about potential disruptions to the local economy. The energy dimension is the most frequently discussed theme within this nexus, appearing in 63.4% of the sample articles. Articles addressing energy tend to exhibit a more negative tone than those that do not. Media discussions often focus on potential increases in local electricity prices and pressure on grid capacity when data centers connect to existing power systems. Coverage frequently highlights that infrastructure upgrades required to support these facilities may be passed on to rural residents through higher utility rates. In cases where data centers develop their own power sources, reports note that their reliance on fuels such as natural gas, diesel, and coal may further increase local energy prices and raise the cost of living and production for rural households. Water is discussed in 41.4% of the articles, and those addressing water tend to exhibit a more negative tone on average. Media coverage emphasizes the substantial water demands associated with traditional evaporative cooling systems used in data centers. These discussions frequently link data center development to aquifer depletion, declining groundwater levels, and increased costs of maintaining local water infrastructure. Land is the least frequently discussed dimension within the nexus, appearing in 25.6% of the sample articles. Articles addressing land use also tend to exhibit a more negative tone. Media coverage raises concerns about the conversion of agricultural land for data center development. Additional concerns include rising land prices driven by large-scale data center investment, which may increase property tax burdens in surrounding areas. At the same time, some residents express concern that data centers may reduce nearby property values due to aesthetic considerations and perceived environmental impacts. 4.2.3. Environment The environment dimension captures non-consumptive environmental impacts associated with data centers, primarily including pollution, waste, and greenhouse gas (GHG) emissions. In total, 39.5% of the sampled articles address this dimension, and these articles exhibit a substantially more negative tone. Data centers can generate pollution affecting local air, water, and land resources through cooling processes and energy use. Media coverage reports that cooling water discharged into local systems is often significantly warmer than natural conditions and may contain chemicals used to control mineral buildup and bacterial growth. This may contribute to both thermal and chemical pollution in local streams and groundwater systems. In addition, data centers frequently rely on backup generators or dedicated private power sources that use fossil fuels. These sources are often portrayed as more difficult to monitor and regulate, contributing to air pollution, water contamination, and increased GHG emissions. Articles also highlight concerns related to noise pollution, electronic waste from frequent equipment replacement, and risks associated with fuel storage and potential chemical leakage. These issues are commonly framed as threats to local environmental quality and public health. 4.2.4. Community Identity Community identity represents another prominent dimension, discussed in 27.2% of the sample articles. Articles addressing this dimension tend to exhibit a more negative tone on average. Media coverage within this theme often focuses on perceived impacts of data center expansion on nonmarket amenities associated with rural communities. These include changes to the visual landscape due to large industrial-scale facilities, perceived disruptions to rural lifestyles, and concerns about preserving local agricultural heritage. Articles frequently describe how residents view data center development as altering the traditional character of rural communities. 4.2.5. Local Governance Local governance is among the most frequently discussed dimensions, appearing in 67.1% of all articles. Media coverage within this theme primarily focuses on policy decision-making processes and conflicts surrounding data center development in rural municipalities. Because much rural land is zoned for agricultural use, data center projects often require zoning changes, special permits, or other forms of local approval. Articles frequently report conflicts surrounding these approval processes, particularly related to transparency and representation. For example, some coverage highlights the use of Non-Disclosure Agreements (NDAs) that limit public access to project details during early planning stages. Other articles describe tensions between local officials and residents, where policymakers are perceived as favoring developer interests. In some cases, reports suggest that impact assessments may emphasize economic benefits while downplaying potential costs, contributing to public skepticism. These governance challenges may erode public trust in local decision-making processes and have been associated with increasing community opposition to data center projects. Media narratives often portray these conflicts as a key driver of resistance among rural residents. [3] Standard English stop words, along with contextual terms such as “data centers,” “million,” “billion,” “company,” “said,” and “site,” are removed from the analysis. 5. Discussion The rapid growth of AI and the expansion of data center infrastructure into rural regions have attracted increasing attention and intensified interactions with local communities, resulting in significant environmental and socioeconomic implications. 5.1. Environment Environmental impacts can be broadly categorized into two dimensions: resource depletion, and contamination and emissions. Regarding resource depletion, the magnitude and scale of resource use by data centers, along with the associated consequences for national and global resource reserves, require further quantitative examination to evaluate the broader environmental footprint of data center infrastructure. In addition, the geographic distribution of data centers and their environmental footprint deserve close attention. An uneven concentration of data centers in resource-scarce regions, such as water-stressed areas, could result in excessive resource extraction and even permanent depletion. Given these resource challenges, together with the substantial capital capacity of technology companies, relying solely on market mechanisms to allocate resources may risk compromising environmental sustainability and leading to long-term consequences. Future research should therefore explore effective resource management strategies that complement market forces and help mitigate potential externalities. Possible approaches include zoning policies that guide data center development toward more suitable locations and infrastructure planning that improves the mobility and transfer of key resources across regions. Such measures could support more sustainable environmental development and more balanced human-nature interactions. On the other hand, contamination and emissions represent another major component of the environmental footprint of data centers. Data center infrastructure can generate significant local and regional environmental consequences by degrading air, water, and soil quality, undermining emission-reduction goals, and, in some cases, creating public health risks. The spatial distribution of these contamination and emission externalities may be even more important and more difficult to monitor, given the financial capacity of technology companies and the self-sufficient nature of some remote data center developments. When such facilities are built with dedicated power sources and independent data transmission infrastructure, such as satellite-based services, they may create concentrated and insufficiently monitored pollution burdens in areas with high-quality natural resource endowments, leading to even broader environmental impacts. In light of these concerns, future researchers and policymakers should work toward a more comprehensive environmental regulatory framework. This may include pretreatment requirements, emission quotas, and stronger incentives for clean energy use, together with effective enforcement mechanisms to ensure that these facilities operate within controlled environmental limits. 5.2. Socioeconomics The expansion of data centers into rural regions reveals socioeconomic implications across three primary dimensions: (1) agriculture, which supports rural income and regional community development; (2) living standards, particularly the cost of living and quality of life for local populations; and (3) local governance. 5.2.1. Agriculture Agriculture serves as the economic backbone and a primary source of income in rural regions across many countries. Populations in these areas have long relied on farm production for their livelihoods, which depend on a stable and affordable supply of resource inputs to sustain the thin profit margins. In particular, agricultural production requires extensive land for crops and livestock, reliable water resources for irrigation and animal consumption, and energy inputs, including fossil fuels and electricity, to operate machinery and equipment. Data center development may interfere with agricultural production in three key ways. First, it creates direct competition for resource inputs, raising production costs and undermining farm profitability, which in turn reduces the income of rural populations. Second, for resources with quality differentiation, data centers may acquire higher-quality resources due to their significantly greater financial capacity than local populations. This process can systematically push agricultural producers toward lower-quality inputs, resulting in reduced productivity and lower farm income. For example, data centers may occupy prime agricultural land that is flat and located near water sources, permanently converting it into industrial facilities and removing it from agricultural use. Third, in regions where resources are scarce, data centers may take up a large share of limited resources due to their financial advantage. For instance, they may secure water rights in drought-prone or semi-arid areas, thereby fully displacing nearby farms. Indeed, the expansion of data centers may erode the agricultural economic foundation of rural regions and weaken the income base of local populations. This may accelerate the exit of agricultural populations and propagate effects through the broader food supply chain, raising concerns about national and global food security and price stability, with disproportionate impacts on low-income consumers who are more sensitive to food price increases. 5.2.2. Living Standards Data center development also directly affects local populations through changes in the cost of living and overall quality of life. These facilities may increase living costs by intensifying competition for key resources, such as raising local electricity rates, fuel prices, and the maintenance costs of water systems. In addition, the construction of large industrial facilities with continuous lighting and operational noise can alter the visual and cultural landscape of rural communities. These changes may degrade the surrounding environment and social atmosphere, compromise living conditions, and lead to nonmarket losses associated with rural lifestyles, disrupting established ways of life and local heritage. Combined with the negative impacts on rural income, data center development may undermine both the economic viability and quality of life of local populations. This process may contribute to the marginalization of these communities, potentially leading to outmigration and household displacement, and ultimately weakening regional social development. In addition, these dynamics raise important concerns regarding economic equality and social justice. Rural populations are often economically disadvantaged relative to urban populations, a condition reflected in the lower cost of rural resources. The development of data centers in these regions to take advantage of lower resource costs may therefore reflect a pattern in which private firms exploit resources located in economically vulnerable communities. The operational, financial, and social costs associated with data center activities may thus be disproportionately borne by these populations through nonmarket externalities and rising resource prices in shared markets. As a result, the burdens of technological advancement may shift toward these already vulnerable groups, while the benefits are concentrated among large capital owners, potentially widening economic disparities and reinforcing inequities. Given these concerns, future research should further examine the economic and social trade-offs between AI advancement, data center expansion, and rural development. Efforts should focus on identifying protective measures that ensure stable and affordable access to resources for local communities and residents. Policymakers should consider strategies to mitigate the impacts of data centers on rural livability and income stability, while promoting a more equitable distribution of costs and benefits and supporting sustainable social development. 5.2.3. Local Governance The development of data centers also has significant implications for local governance, exposing institutional vulnerabilities in regional regulatory bodies when confronted with large-scale investments. Data center projects, often involving investments ranging from millions to billions of dollars, introduce substantial pressures on local townships and counties, which were originally designed to manage dispersed populations and routine infrastructure needs. Documented tensions between residents and local officials, along with concerns about the transparency of policy decision-making processes, may further complicate governance outcomes. If not properly addressed, these issues can weaken public trust and generate additional social challenges. Given these dynamics, future research should examine how local governance structures can be strengthened to better manage large-scale development projects. In particular, zoning decisions and approval processes, which may permanently alter regional development patterns, should be conducted with greater transparency and accountability. Policymakers may also consider promoting public participation and developing more comprehensive planning frameworks that protect local communities while accommodating socioeconomic and technological development. 6. Conclusions and Remarks The rapid advancement of AI has been accompanied by substantial infrastructure demands through the expansion of data centers. The growing presence of these facilities in rural regions carries important environmental and socioeconomic implications that warrant greater attention from researchers and policymakers. This study provides a first systematic descriptive analysis of this issue from a media perspective. We find a rapid increase in media coverage of rural data center expansion in recent years, accompanied by an intensification of public debate. Although a significant share of newspaper articles highlights potential benefits, such as job creation and tax revenue, these benefits are increasingly questioned in terms of their scale and distribution. At the same time, concerns have intensified regarding pressures on energy, water, and land resources, as well as impacts on environmental quality, community identity, and local governance. These issues are widely reported as affecting rural communities through both environmental and socioeconomic channels, with implications for agricultural production and rural livelihoods. If left unaddressed, these externalities may generate broader challenges for social development and equity, as economically disadvantaged rural communities may disproportionately bear the costs of technological advancement. Given these developments, there is a clear need for future research that systematically evaluates the environmental and socioeconomic impacts, as well as the associated trade-offs, of data center expansion in rural regions. Researchers and policymakers should work toward developing governance frameworks that support evidence-based decision making and balance technological advancement with social development. 7. Limitations This study has two main limitations. First, media coverage may not represent all perspectives, and further research is needed to incorporate views from stakeholders across diverse demographic and socioeconomic backgrounds. Second, this study employs a simplified sentiment analysis based on tone classification. While this approach provides a useful overview, it may oversimplify complex narratives. Despite these limitations, this study provides a valuable foundation for informing future research and offers preliminary yet policy-relevant insights that can help guide more data-driven decision-making as rural data center development continues to evolve. 8. Statements and Declarations Data availability The datasets analysed during the current study are available in the supplementary file. Competing interests The authors declare no competing interests. Funding No funding was received for this study. Ethics declarations Ethical approval was not required as no human participants were involved in this research. References Andsager JL (2000) How Interest Groups Attempt to Shape Public Opinion with Competing News Frames. Journal Mass Commun Q 77:577–592. https://doi.org/10.1177/107769900007700308 Boguraev B, Kennedy C (2002) Salience-based Content Characterisation of Text Documents Hassan A (2026) ‘Nobody Owns Us’: How Plans for a Google Data Center Roiled an Oklahoma Town. N. Y. Times Ibrahim OAS, Landa-Silva D (2016) Term frequency with average term occurrences for textual information retrieval. Soft Comput 20:3045–3061. https://doi.org/10.1007/s00500-015-1935-7 Jegham N, Abdelatti M, Koh CY, et al (2025) How Hungry is AI? Benchmarking Energy, Water, and Carbon Footprint of LLM Inference Lee N (2026) Supporting rural communities amid the data center boom. Brookings Lei N, Lu J, Shehabi A, Masanet E (2025) The water use of data center workloads: A review and assessment of key determinants. Resour Conserv Recycl 219:108310. https://doi.org/10.1016/j.resconrec.2025.108310 Liu Y, Wang H (2026) Who on earth is using Generative AI? World Dev 199:107260. https://doi.org/10.1016/j.worlddev.2025.107260 Makholm JD, Olive LTW, Beiser EA (2025) Hyperscale Data Centers, Regulatory Institutions, and US Economic Growth. Clim Energy 41:21–27. https://doi.org/10.1002/gas.22441 Radhakrishnan S, Erbis S, Isaacs JA, Kamarthi S (2017) Novel keyword co-occurrence network-based methods to foster systematic reviews of scientific literature. PLOS ONE 12:e0172778. https://doi.org/10.1371/journal.pone.0172778 Shin R (2026) Trump’s AI drive is running headlong into his most reliable voters. POLITICO Siddik MAB, Shehabi A, Marston L (2021) The environmental footprint of data centers in the United States. Environ Res Lett 16:064017. https://doi.org/10.1088/1748-9326/abfba1 Stanford Institute for Human-Centered AI (2025) Artificial Intelligence Index Report 2025 Synergy Research Group (2025) Hyperscale Data Center Count Hits 1,136; Average Size Increases; US Accounts for 54% of Total Capacity. Reno, NV Tan Y, Weaver DH (2009) Local Media, Public Opinion, and State Legislative Policies: Agenda Setting at the State Level. Int J Press 14:454–476. https://doi.org/10.1177/1940161209336225 Tewksbury D, Jones J, Peske MW, et al (2000) The Interaction of News and Advocate Frames: Manipulating Audience Perceptions of a Local Public Policy Issue. Journal Mass Commun Q 77:804–829. https://doi.org/10.1177/107769900007700406 Zhang H, Wang D, Wu W, Hu H (2012) Term frequency – function of document frequency: a new term weighting scheme for enterprise information retrieval. Enterp Inf Syst 6:433–444. https://doi.org/10.1080/17517575.2012.665945 Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9442287","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":624564483,"identity":"0ba12ac7-9fcf-45a5-ace1-e129170318d8","order_by":0,"name":"Jingyuan Zhang","email":"","orcid":"","institution":"Prairie View A\u0026M University","correspondingAuthor":false,"prefix":"","firstName":"Jingyuan","middleName":"","lastName":"Zhang","suffix":""},{"id":624564484,"identity":"79d85248-b690-42bc-8322-c0bc9278f602","order_by":1,"name":"Rui Liu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAnElEQVRIiWNgGAWjYBACPmYQaWAD5bIRoYUNoiVNggQtEOowKVrYecw+8xScr+OfdsaA4UPZYWIcxmM8m8fgtoTE7RwDxhnniNTCDNLCANTCzNtGvJZzEvIgLX9J0HJAwgCkhZE4LWzFjHMMkiU33k4rONhzLp2wFn7+w5sZ3vyx45e7nbzxwY8ya8JaUMABEtWPglEwCkbBKMAFAPqDLH/YeVU6AAAAAElFTkSuQmCC","orcid":"","institution":"Texas A\u0026M University","correspondingAuthor":true,"prefix":"","firstName":"Rui","middleName":"","lastName":"Liu","suffix":""},{"id":624564485,"identity":"2ddbdf35-9ffd-4661-8d72-d4c562d264dc","order_by":2,"name":"Sunil Dhoubhadel","email":"","orcid":"","institution":"Prairie View A\u0026M University","correspondingAuthor":false,"prefix":"","firstName":"Sunil","middleName":"","lastName":"Dhoubhadel","suffix":""},{"id":624564486,"identity":"ed5f56a8-b01b-45a4-9c34-c1f91a6dbf52","order_by":3,"name":"Rodolfo Nayga","email":"","orcid":"","institution":"Texas A\u0026M University","correspondingAuthor":false,"prefix":"","firstName":"Rodolfo","middleName":"","lastName":"Nayga","suffix":""}],"badges":[],"createdAt":"2026-04-16 22:09:45","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-9442287/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9442287/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107290056,"identity":"3c25d63c-22f9-4963-a53c-54155a574c12","added_by":"auto","created_at":"2026-04-20 05:21:03","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":112026,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMonthly newspaper article publication frequency by tone\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-9442287/v1/0edb561561e1c5341a112000.png"},{"id":107485521,"identity":"5881d3cb-853c-464b-a620-b3062474e5ac","added_by":"auto","created_at":"2026-04-22 02:35:18","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":134421,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGeographic distribution and tone of newspaper articles\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote: Bubble size indicates article volume; Color indicates average tone.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-9442287/v1/ada64364d25e4d25d6b26da5.png"},{"id":107290058,"identity":"4f2f9d60-1649-4ec6-b63f-3a04af7829c0","added_by":"auto","created_at":"2026-04-20 05:21:03","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":102198,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eWord co-occurrence network\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-9442287/v1/920b03f62497bbd2a6bf9c81.png"},{"id":107485484,"identity":"256780d5-200d-4dd1-b1fc-c1ba63b19789","added_by":"auto","created_at":"2026-04-22 02:35:09","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":72037,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFrequency of thematic dimensions and corresponding tone\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eNote:\u003c/strong\u003e The left panel shows the number of articles that address each specific dimension. The right panel compares the average tone of articles that discuss a given dimension (solid circles) with those that do not (open circles). For example, articles that discuss the local governance dimension are, on average, more negative (solid circles) than those that do not discuss this dimension (open circles). The average tone is calculated by coding articles with a negative tone as -1 and those with a positive tone as 1, and then taking the mean across each group. The vertical dashed line represents the overall average tone across all analyzed articles.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-9442287/v1/cbaddcb37b768fa8564096fe.png"},{"id":107487270,"identity":"d34634f1-ed3a-4059-846a-d4d185d87da7","added_by":"auto","created_at":"2026-04-22 02:40:18","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":697262,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9442287/v1/2002e629-57d7-4327-86f7-9f80cc6a81e4.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eAI’s Hidden Footprint: Media Framing of Rural Data Center Expansion\u003c/p\u003e","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eArtificial Intelligence (AI) has experienced rapid growth in recent years. In 2024 alone, corporate investment in AI exceeded 252.3\u0026nbsp;billion U.S. dollars (Stanford Institute for Human-Centered AI \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), and chatbot applications such as ChatGPT have surpassed 2\u0026nbsp;billion visits and 500\u0026nbsp;million users worldwide (Liu and Wang \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). Advances in machine learning, data processing, and computing power have enabled technology companies to integrate AI into a wide range of applications, transforming technological progress, business operations, and social development. In business markets, AI has improved production processes, decision-making, market analytics, and automation. For the general population, AI is increasingly embedded in everyday life through personalized services, influencing how people access information, interact with digital platforms, and make social decisions.\u003c/p\u003e \u003cp\u003eThe infrastructure underlying this AI boom is the data centers, which store vast amounts of data and provide the computational power required for AI operation and development. Data centers are large-scale industrial facilities that house high-density computing systems, servers, and supporting equipment, including generators, power supplies, and environmental control systems. Maintaining their continuous operation requires substantial resource inputs, particularly energy and water (Lei et al. \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Jegham et al. \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). For example, a single large data center can consume tens to hundreds of megawatts of electricity and use millions of gallons of water per day for cooling, levels comparable to those of a medium-sized city.\u003c/p\u003e \u003cp\u003eTo meet the growing demand for AI services, technology companies have been rapidly expanding data center infrastructure, emphasizing its importance for market competitiveness, technological advancement, and national security (Makholm et al. \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). In the United States, which accounts for 54% of global data center capacity, the number of data center facilities has more than doubled since 2020, with both their number and scale increasing substantially (Synergy Research Group \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). While the early development of these facilities was concentrated in urban areas, recent trends show a growing shift toward rural regions, including small towns and communities (Lee \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Shin \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Hassan \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2026\u003c/span\u003e), a pattern often attributed to greater resource availability and lower costs.\u003c/p\u003e \u003cp\u003eFor rural counties and towns, data centers represent extremely high-value investments, often ranging from hundreds of millions to billions of dollars. Their development is therefore expected to generate employment opportunities and increase tax revenues, supporting local infrastructure and broader economic growth. However, their expansion into rural regions also introduces important trade-offs. The key resources required for data centers, namely the energy-water-land nexus, are fundamental to rural landscapes and livelihoods. Introducing data centers that demand these resources at the scale of a medium-sized city into rural regions, where infrastructure is designed for dispersed populations, can strain local systems, compromise infrastructure reliability, intensify resource competition, and increase living costs.\u003c/p\u003e \u003cp\u003eDeveloping data centers in rural regions reflects an emerging trend with significant natural and socioeconomic implications. It introduces a new dimension to human-nature interactions by placing pressure on rural areas, which host a large share of natural capital, including land and water resources, in most countries, and may exacerbate existing resource constraints. This expansion can directly affect the economic backbone of rural economies, namely agricultural production, by raising farmers\u0026rsquo; production costs and weakening the economic capacity of local populations. The resulting impacts may also extend to the social sphere, raising concerns about distributional justice and the disproportionate burden placed on rural communities, as capital-intensive technology companies take advantage of low-cost rural resources.\u003c/p\u003e \u003cp\u003eExisting literature on the impacts of AI and data centers has primarily focused on energy use, natural resource consumption, and environmental concerns. For example, Jegham et al. (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) assess the environmental footprint of AI by examining data center consumption of energy, water, and carbon emissions. They find that even short queries at large scale, such as 700\u0026nbsp;million queries per day, can consume electricity comparable to that used by 35,000 U.S. households annually, require water equivalent to the needs of 1.2\u0026nbsp;million people, and generate carbon emissions that would require a Chicago-sized forest to offset. They argue that AI development drives disproportionately high resource consumption and poses challenges for sustainable development. Similarly, Siddik et al. (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) discuss the substantial resource demands of data centers in the United States, showing that many facilities draw water from medium- and highly water-stressed watersheds and generate substantial greenhouse gas emissions, raising concerns about long-term sustainability.\u003c/p\u003e \u003cp\u003eOn the other hand, the socioeconomic dimensions of data center expansion remain largely underexplored. There is no research examining the roles, challenges, and evolving dynamics faced by rural populations, who have historically been economically vulnerable in many countries and underrepresented in academic research. What are the primary concerns of rural communities regarding this trend? What types of interests and conflicts emerge in local decision-making processes? What are the broader social and economic implications? Without a clear understanding of the full scope of this issue, it is difficult for researchers and policymakers to develop informed research agendas and effective policy responses.\u003c/p\u003e \u003cp\u003eThis study aims to fill this gap by providing an initial exploration of AI\u0026rsquo;s infrastructural externalities and the broader natural and socioeconomic impacts of data centers on human\u0026ndash;environment interactions. In the absence of a well-established academic literature, we conduct the first systematic descriptive media analysis, which leverages the important role of media in reflecting stakeholder concerns, shaping public perceptions, and influencing political outcomes (Andsager \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2000\u003c/span\u003e; Tewksbury et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2000\u003c/span\u003e; Tan and Weaver \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Specifically, we examine how media outlets frame the rural expansion of data centers and analyze their thematic focus, using a dataset of newspaper articles published between January 2024 and February 2026 that captures the early stage of this expansion. Our findings provide timely insights to guide future research on the natural and socioeconomic implications of data centers and help policymakers consider trade-offs when designing balanced regulatory frameworks that support technological advancement alongside sustainable environmental and social development.\u003c/p\u003e"},{"header":"2. Data","content":"\u003cp\u003eThe media data are obtained from Nexis Uni, a comprehensive database that archives newspaper articles published worldwide. To focus on rural data center expansion, we adopt a two-pillar keyword strategy[1]. The first pillar captures the object of interest using the term \u0026ldquo;data center.\u0026rdquo; The second pillar captures the rural context using the following keywords: \u0026ldquo;rural,\u0026rdquo; \u0026ldquo;small town,\u0026rdquo; \u0026ldquo;small towns,\u0026rdquo; \u0026ldquo;agricultur*,\u0026rdquo; \u0026ldquo;farm*,\u0026rdquo; \u0026ldquo;unincorporated,\u0026rdquo; and \u0026ldquo;community.\u0026rdquo; The wildcard operator (*) allows the search to capture variations such as \u0026ldquo;agriculture,\u0026rdquo; \u0026ldquo;agricultural,\u0026rdquo; \u0026ldquo;farm,\u0026rdquo; and \u0026ldquo;farming.\u0026rdquo; Articles are retained only if keywords from both pillars appear in either the headline or the lead paragraph. The search is restricted to newspaper articles published in the United States, written in English, and dated between January 1, 2024, and February 28, 2026. Using this query, a total of 1,482 articles were retrieved in March 2026.\u003c/p\u003e\n\u003cp\u003eTo ensure that data centers are the primary focus of the articles, rather than a passing reference or part of a broader discussion, we apply a term frequency threshold (Zhang et al. 2012; Ibrahim and Landa-Silva 2016). Specifically, we retain articles in which the term \u0026ldquo;data center\u0026rdquo; appears at least three times in the body text. This threshold is informed by the average article length of 1,078 words. The final sample consists of 519 newspaper articles[2]. The distribution of the sample by publication year and geographic location is reported in Table 1.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" class=\"fr-table-selection-hover\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" style=\"width: 623px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTable 1. Sample characteristics\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003eArticles (N = 519)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003ePublication Year\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e89 (17.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e2025\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e291 (56.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e2026\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e139 (26.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003eGeographic Region\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003eMidwest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e162 (31.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003eSouth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e141 (27.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003eNortheast\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e102 (19.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003eWest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e35 (6.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003eNot labeled\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 312px;\"\u003e\n \u003cp\u003e79 (15.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e[1] A three-pillar keyword strategy incorporating a sentiment dimension, such as keywords including \u0026ldquo;support,\u0026rdquo; \u0026ldquo;oppose,\u0026rdquo; or \u0026ldquo;protest,\u0026rdquo; is not adopted because it may introduce selection bias in the retrieval of articles.\u003c/p\u003e\n\u003cp\u003e[2] As a robustness check, we conduct a validation exercise in which a random sample of 100 articles is drawn from the full set of 1,482 retrieved articles. Each article is manually reviewed to determine whether data centers constitute the primary focus. The human validation yields an 87.0% agreement rate with the term frequency method.\u003c/p\u003e"},{"header":"3. Sentiment Analysis","content":"\u003cp\u003eEvaluating the tone of newspaper coverage and its evolution over time provides insights into how the media perceive rural data center expansion and how public perceptions may be shaped. We assess media tone using a sentiment analysis approach.\u003c/p\u003e\n\u003cp\u003eTwo independent coders reviewed the sample articles following a standardized protocol and classified each article into one of three mutually exclusive categories: negative, positive, or neutral. An article was coded as negative if it primarily emphasized adverse impacts, local conflicts, or risks associated with data center development, with substantially less or no discussion of potential benefits. Conversely, an article was coded as positive if it highlighted benefits, such as economic opportunities and gains, with limited discussion of potential drawbacks. Articles were classified as neutral if they presented information in a factual tone without emphasizing either positive or negative aspects, or if both perspectives were discussed in a balanced manner.\u003c/p\u003e\n\u003cp\u003eThe resulting confusion matrix is presented in Table 2. The independent coding produced largely consistent results, with an inter-coder agreement rate of 87.0% and a Cohen\u0026rsquo;s Kappa of 0.79. To avoid overstating either positive or negative coverage, articles with coder disagreement were classified as neutral. The final classification of the 519 articles includes 250 negative, 110 positive, and 159 neutral articles.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" style=\"width: 623px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTable 2. Confusion matrix of the sentiment analysis\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003eCoder1/Coder2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003eNegative\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003eNeutral\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003ePositive\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003eNegative\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003e250\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003e13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003eNeutral\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003e17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003e93\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003ePositive\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003e26\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 156px;\"\u003e\n \u003cp\u003e110\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eFigure 1 presents the monthly publication frequency of articles by tone category. Prior to June 2025, the number of articles discussing rural data center expansion remained relatively low and stable, with similar levels of negative, neutral, and positive coverage. This period likely reflects the early stage of rural data center expansion, when local communities had not yet formed strong or consistent reactions and were exposed to both positive and negative narratives. During this stage, many projects were still under development and had not yet fully affected local communities.\u003c/p\u003e\n\u003cp\u003eSince June 2025, however, the number of negative and neutral articles has increased substantially, while the number of positive articles has declined. Several factors may contribute to this shift. As large-scale data center development accelerated across multiple rural regions in early 2025, increased demand for energy and water resources, particularly during peak summer periods, may have made resource constraints and associated impacts more visible to local communities. Additionally, the rapid scaling of computing technologies and AI infrastructure may have intensified resource consumption relative to earlier periods, reinforcing negative perceptions.\u003c/p\u003e\n\u003cp\u003eFigure 2 presents the geographic distribution of articles and their average tone by state. The Mid-Atlantic and Midwest regions account for the largest share of media coverage, while the South and West contribute fewer articles. This pattern is consistent with the geographic concentration of early data center development. Across most states, coverage is predominantly negative, with positive tones appearing only in a few states with relatively lower publication volumes.\u003c/p\u003e"},{"header":"4. Thematic Analysis","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003e4.1. Keywords\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAnalyzing article keywords helps identify the main topics discussed in the media and provides insights into how these topics are interconnected. We conduct a word co-occurrence network analysis (Radhakrishnan et al. 2017), which identifies high-frequency terms that appear together within the same article and computes pairwise correlations among them.[3] The results are presented in Figure 3.\u003c/p\u003e\n\u003cp\u003eThe analysis reveals two primary clusters of keywords. The cluster located in the upper-right region of Figure 3 consists of articles focused on the energy dimension of data center expansion in rural areas, with prominent terms such as electricity, power, energy, and gas. The second cluster, located in the lower-left region, reflects broader discussions across a more diverse set of topics, including resource-related terms such as land and water, community-related terms such as residents and people, and local governance issues including councils and zoning.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003e4.2. Thematic Dimensions\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSeveral dimensions of rural data center expansion are of particular interest to researchers and policymakers. Investigating how the media frame these dimensions and what newspaper articles address within them can provide insights into public perceptions and concerns.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eBased on the most frequently occurring keywords identified in Figure 3 and prior studies on the resource demands of data centers (Jegham et al. 2025; Lei et al. 2025), we focus on the following dimensions: Jobs- referring to employment opportunities generated by data center development; Taxation- referring to local tax revenues associated with data center investments; Water, Energy, and Land- capturing resource consumption required for data center operations; Environment- referring to non-consumptive environmental impacts such as pollution, waste, and emissions; Community Identity- capturing nonmarket impacts on rural culture, heritage, and lifestyle; and Local Governance- referring to municipal decision-making processes such as zoning changes, council deliberations, planning approvals, and moratoriums.\u003c/p\u003e\n\u003cp\u003eEach article is reviewed to determine whether it addresses these thematic dimensions based on topical salience (Boguraev and Kennedy 2002). For each dimension, articles are coded as either \u0026ldquo;yes\u0026rdquo; or \u0026ldquo;no.\u0026rdquo; An article is coded as \u0026ldquo;yes\u0026rdquo; only if it contains substantial discussion of the dimension, defined as the dimension serving as a primary narrative driver of the article, appearing in a dedicated paragraph, or recurring throughout the article. Detailed definitions, inclusion criteria, and exclusion rules are provided in Table 3.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"3\" style=\"width: 623px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTable 3. Structured codebook for thematic dimensions\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eThematic Dimensions\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 284px;\"\u003e\n \u003cp\u003eDefinition and Inclusion Criteria\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eExclusion Criteria\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eJobs\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 284px;\"\u003e\n \u003cp\u003eEmployment associated with data center development, including job creation and local hiring initiatives.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eNo discussion of the inclusion criteria.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eTaxation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 284px;\"\u003e\n \u003cp\u003eLocal tax revenue and public funding generated by data centers.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eNo discussion of the inclusion criteria.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eWater\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 284px;\"\u003e\n \u003cp\u003eWater scarcity resulting from data center development and consumption, such as depleted wells.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eNo discussion of the inclusion criteria, or references to water pollution or contamination.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eEnergy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 284px;\"\u003e\n \u003cp\u003eEnergy impacts associated with data centers.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eNo discussion of the inclusion criteria.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eLand\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 284px;\"\u003e\n \u003cp\u003eLand use changes and property value impacts resulting from data center development.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eNo discussion of the inclusion criteria, or zoning discussions limited solely to governance procedures.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eEnvironment\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 284px;\"\u003e\n \u003cp\u003eEnvironmental impacts of data centers, including pollution, waste, and greenhouse gas emissions.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eNo discussion of the inclusion criteria, or discussions limited only to resource consumption and scarcity issues.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eCommunity Identity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 284px;\"\u003e\n \u003cp\u003eImpacts on rural community character, such as agricultural heritage, rural lifestyles, or concerns about visual landscape changes due to data centers.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eNo discussion of the inclusion criteria.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eLocal Governance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 284px;\"\u003e\n \u003cp\u003eLocal governance processes related to data centers, including planning decisions, zoning votes, moratoriums, and municipal decision-making.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 208px;\"\u003e\n \u003cp\u003eNo discussion of the inclusion criteria.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eFigure 4 presents the distribution of articles across thematic dimensions. The left panel reports the number of articles addressing each dimension, ranked from the most frequently discussed dimension, local governance, to the least discussed, land. The right panel presents the average difference in tone between articles that address a given dimension and those that do not. Because a single article may address multiple dimensions simultaneously, counts across dimensions are not mutually exclusive and may sum to more than the total number of articles (519). In the following sections, each thematic dimension is discussed in detail.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e4.2.1. Jobs and Taxation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eJob creation and tax revenue are often viewed as key drivers of early local support for data centers. In the sample, 39.9% of articles discuss job creation, and 33.1% discuss tax revenue related to rural data center expansion. As shown in the right panel of Figure 4, although both groups of articles exhibit an overall negative tone, those that address jobs and taxation are, on average, more positive than those that do not.\u003c/p\u003e\n\u003cp\u003eDiscussions of job creation largely reflect statements from developers and local policymakers during project planning. However, a substantial share of articles question the effectiveness and magnitude of these benefits. Many argue that most jobs created by data centers are temporary construction jobs, such as electricians and construction workers, with relatively few permanent operational positions. For example, a proposed $10 billion Meta data center project announced in June 2025 near Rayville is expected to create 5,600 jobs, of which approximately 5,300 are temporary and only 300 are permanent. Articles further suggest that, due to the highly automated nature of data centers, permanent positions may be limited in number and scope, leading to concerns that employment benefits are overstated.\u003c/p\u003e\n\u003cp\u003eA similar debate emerges regarding taxation. While many articles emphasize the potential for millions to billions of dollars in long-term tax revenues, others argue that tax incentives used to attract data centers, often lasting two to ten years, may offset expected fiscal gains. As a result, the net economic benefits to local communities may be more modest than initially anticipated.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e4.2.2. Energy-Water-Land Nexus\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eEnergy, water, and land are often discussed jointly as part of a resource nexus that supports rural livelihoods and agricultural production. The expansion of data centers, which relies heavily on these same resources, has therefore raised concerns about potential disruptions to the local economy.\u003c/p\u003e\n\u003cp\u003eThe energy dimension is the most frequently discussed theme within this nexus, appearing in 63.4% of the sample articles. Articles addressing energy tend to exhibit a more negative tone than those that do not. Media discussions often focus on potential increases in local electricity prices and pressure on grid capacity when data centers connect to existing power systems. Coverage frequently highlights that infrastructure upgrades required to support these facilities may be passed on to rural residents through higher utility rates. In cases where data centers develop their own power sources, reports note that their reliance on fuels such as natural gas, diesel, and coal may further increase local energy prices and raise the cost of living and production for rural households.\u003c/p\u003e\n\u003cp\u003eWater is discussed in 41.4% of the articles, and those addressing water tend to exhibit a more negative tone on average. Media coverage emphasizes the substantial water demands associated with traditional evaporative cooling systems used in data centers. These discussions frequently link data center development to aquifer depletion, declining groundwater levels, and increased costs of maintaining local water infrastructure.\u003c/p\u003e\n\u003cp\u003eLand is the least frequently discussed dimension within the nexus, appearing in 25.6% of the sample articles. Articles addressing land use also tend to exhibit a more negative tone. Media coverage raises concerns about the conversion of agricultural land for data center development. Additional concerns include rising land prices driven by large-scale data center investment, which may increase property tax burdens in surrounding areas. At the same time, some residents express concern that data centers may reduce nearby property values due to aesthetic considerations and perceived environmental impacts.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e4.2.3. Environment\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe environment dimension captures non-consumptive environmental impacts associated with data centers, primarily including pollution, waste, and greenhouse gas (GHG) emissions. In total, 39.5% of the sampled articles address this dimension, and these articles exhibit a substantially more negative tone.\u003c/p\u003e\n\u003cp\u003eData centers can generate pollution affecting local air, water, and land resources through cooling processes and energy use. Media coverage reports that cooling water discharged into local systems is often significantly warmer than natural conditions and may contain chemicals used to control mineral buildup and bacterial growth. This may contribute to both thermal and chemical pollution in local streams and groundwater systems. In addition, data centers frequently rely on backup generators or dedicated private power sources that use fossil fuels. These sources are often portrayed as more difficult to monitor and regulate, contributing to air pollution, water contamination, and increased GHG emissions. Articles also highlight concerns related to noise pollution, electronic waste from frequent equipment replacement, and risks associated with fuel storage and potential chemical leakage. These issues are commonly framed as threats to local environmental quality and public health.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e4.2.4. Community Identity\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eCommunity identity represents another prominent dimension, discussed in 27.2% of the sample articles. Articles addressing this dimension tend to exhibit a more negative tone on average. Media coverage within this theme often focuses on perceived impacts of data center expansion on nonmarket amenities associated with rural communities. These include changes to the visual landscape due to large industrial-scale facilities, perceived disruptions to rural lifestyles, and concerns about preserving local agricultural heritage. Articles frequently describe how residents view data center development as altering the traditional character of rural communities.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e4.2.5. Local Governance\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eLocal governance is among the most frequently discussed dimensions, appearing in 67.1% of all articles. Media coverage within this theme primarily focuses on policy decision-making processes and conflicts surrounding data center development in rural municipalities. Because much rural land is zoned for agricultural use, data center projects often require zoning changes, special permits, or other forms of local approval.\u003c/p\u003e\n\u003cp\u003eArticles frequently report conflicts surrounding these approval processes, particularly related to transparency and representation. For example, some coverage highlights the use of Non-Disclosure Agreements (NDAs) that limit public access to project details during early planning stages. Other articles describe tensions between local officials and residents, where policymakers are perceived as favoring developer interests. In some cases, reports suggest that impact assessments may emphasize economic benefits while downplaying potential costs, contributing to public skepticism. These governance challenges may erode public trust in local decision-making processes and have been associated with increasing community opposition to data center projects. Media narratives often portray these conflicts as a key driver of resistance among rural residents.\u003c/p\u003e\n\u003cp\u003e[3] Standard English stop words, along with contextual terms such as \u0026ldquo;data centers,\u0026rdquo; \u0026ldquo;million,\u0026rdquo; \u0026ldquo;billion,\u0026rdquo; \u0026ldquo;company,\u0026rdquo; \u0026ldquo;said,\u0026rdquo; and \u0026ldquo;site,\u0026rdquo; are removed from the analysis.\u003c/p\u003e"},{"header":"5. Discussion","content":"\u003cp\u003eThe rapid growth of AI and the expansion of data center infrastructure into rural regions have attracted increasing attention and intensified interactions with local communities, resulting in significant environmental and socioeconomic implications.\u003c/p\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e5.1. Environment\u003c/h2\u003e \u003cp\u003eEnvironmental impacts can be broadly categorized into two dimensions: resource depletion, and contamination and emissions. Regarding resource depletion, the magnitude and scale of resource use by data centers, along with the associated consequences for national and global resource reserves, require further quantitative examination to evaluate the broader environmental footprint of data center infrastructure. In addition, the geographic distribution of data centers and their environmental footprint deserve close attention. An uneven concentration of data centers in resource-scarce regions, such as water-stressed areas, could result in excessive resource extraction and even permanent depletion.\u003c/p\u003e \u003cp\u003eGiven these resource challenges, together with the substantial capital capacity of technology companies, relying solely on market mechanisms to allocate resources may risk compromising environmental sustainability and leading to long-term consequences. Future research should therefore explore effective resource management strategies that complement market forces and help mitigate potential externalities. Possible approaches include zoning policies that guide data center development toward more suitable locations and infrastructure planning that improves the mobility and transfer of key resources across regions. Such measures could support more sustainable environmental development and more balanced human-nature interactions.\u003c/p\u003e \u003cp\u003eOn the other hand, contamination and emissions represent another major component of the environmental footprint of data centers. Data center infrastructure can generate significant local and regional environmental consequences by degrading air, water, and soil quality, undermining emission-reduction goals, and, in some cases, creating public health risks. The spatial distribution of these contamination and emission externalities may be even more important and more difficult to monitor, given the financial capacity of technology companies and the self-sufficient nature of some remote data center developments. When such facilities are built with dedicated power sources and independent data transmission infrastructure, such as satellite-based services, they may create concentrated and insufficiently monitored pollution burdens in areas with high-quality natural resource endowments, leading to even broader environmental impacts.\u003c/p\u003e \u003cp\u003eIn light of these concerns, future researchers and policymakers should work toward a more comprehensive environmental regulatory framework. This may include pretreatment requirements, emission quotas, and stronger incentives for clean energy use, together with effective enforcement mechanisms to ensure that these facilities operate within controlled environmental limits.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e5.2. Socioeconomics\u003c/h2\u003e \u003cp\u003eThe expansion of data centers into rural regions reveals socioeconomic implications across three primary dimensions: (1) agriculture, which supports rural income and regional community development; (2) living standards, particularly the cost of living and quality of life for local populations; and (3) local governance.\u003c/p\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003e5.2.1. Agriculture\u003c/h2\u003e \u003cp\u003eAgriculture serves as the economic backbone and a primary source of income in rural regions across many countries. Populations in these areas have long relied on farm production for their livelihoods, which depend on a stable and affordable supply of resource inputs to sustain the thin profit margins. In particular, agricultural production requires extensive land for crops and livestock, reliable water resources for irrigation and animal consumption, and energy inputs, including fossil fuels and electricity, to operate machinery and equipment.\u003c/p\u003e \u003cp\u003eData center development may interfere with agricultural production in three key ways. First, it creates direct competition for resource inputs, raising production costs and undermining farm profitability, which in turn reduces the income of rural populations. Second, for resources with quality differentiation, data centers may acquire higher-quality resources due to their significantly greater financial capacity than local populations. This process can systematically push agricultural producers toward lower-quality inputs, resulting in reduced productivity and lower farm income. For example, data centers may occupy prime agricultural land that is flat and located near water sources, permanently converting it into industrial facilities and removing it from agricultural use. Third, in regions where resources are scarce, data centers may take up a large share of limited resources due to their financial advantage. For instance, they may secure water rights in drought-prone or semi-arid areas, thereby fully displacing nearby farms.\u003c/p\u003e \u003cp\u003eIndeed, the expansion of data centers may erode the agricultural economic foundation of rural regions and weaken the income base of local populations. This may accelerate the exit of agricultural populations and propagate effects through the broader food supply chain, raising concerns about national and global food security and price stability, with disproportionate impacts on low-income consumers who are more sensitive to food price increases.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e \u003ch2\u003e5.2.2. Living Standards\u003c/h2\u003e \u003cp\u003eData center development also directly affects local populations through changes in the cost of living and overall quality of life. These facilities may increase living costs by intensifying competition for key resources, such as raising local electricity rates, fuel prices, and the maintenance costs of water systems. In addition, the construction of large industrial facilities with continuous lighting and operational noise can alter the visual and cultural landscape of rural communities. These changes may degrade the surrounding environment and social atmosphere, compromise living conditions, and lead to nonmarket losses associated with rural lifestyles, disrupting established ways of life and local heritage.\u003c/p\u003e \u003cp\u003eCombined with the negative impacts on rural income, data center development may undermine both the economic viability and quality of life of local populations. This process may contribute to the marginalization of these communities, potentially leading to outmigration and household displacement, and ultimately weakening regional social development.\u003c/p\u003e \u003cp\u003eIn addition, these dynamics raise important concerns regarding economic equality and social justice. Rural populations are often economically disadvantaged relative to urban populations, a condition reflected in the lower cost of rural resources. The development of data centers in these regions to take advantage of lower resource costs may therefore reflect a pattern in which private firms exploit resources located in economically vulnerable communities. The operational, financial, and social costs associated with data center activities may thus be disproportionately borne by these populations through nonmarket externalities and rising resource prices in shared markets. As a result, the burdens of technological advancement may shift toward these already vulnerable groups, while the benefits are concentrated among large capital owners, potentially widening economic disparities and reinforcing inequities.\u003c/p\u003e \u003cp\u003eGiven these concerns, future research should further examine the economic and social trade-offs between AI advancement, data center expansion, and rural development. Efforts should focus on identifying protective measures that ensure stable and affordable access to resources for local communities and residents. Policymakers should consider strategies to mitigate the impacts of data centers on rural livability and income stability, while promoting a more equitable distribution of costs and benefits and supporting sustainable social development.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section3\"\u003e \u003ch2\u003e5.2.3. Local Governance\u003c/h2\u003e \u003cp\u003eThe development of data centers also has significant implications for local governance, exposing institutional vulnerabilities in regional regulatory bodies when confronted with large-scale investments. Data center projects, often involving investments ranging from millions to billions of dollars, introduce substantial pressures on local townships and counties, which were originally designed to manage dispersed populations and routine infrastructure needs. Documented tensions between residents and local officials, along with concerns about the transparency of policy decision-making processes, may further complicate governance outcomes. If not properly addressed, these issues can weaken public trust and generate additional social challenges.\u003c/p\u003e \u003cp\u003eGiven these dynamics, future research should examine how local governance structures can be strengthened to better manage large-scale development projects. In particular, zoning decisions and approval processes, which may permanently alter regional development patterns, should be conducted with greater transparency and accountability. Policymakers may also consider promoting public participation and developing more comprehensive planning frameworks that protect local communities while accommodating socioeconomic and technological development.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"6. Conclusions and Remarks","content":"\u003cp\u003eThe rapid advancement of AI has been accompanied by substantial infrastructure demands through the expansion of data centers. The growing presence of these facilities in rural regions carries important environmental and socioeconomic implications that warrant greater attention from researchers and policymakers. This study provides a first systematic descriptive analysis of this issue from a media perspective. We find a rapid increase in media coverage of rural data center expansion in recent years, accompanied by an intensification of public debate.\u003c/p\u003e \u003cp\u003eAlthough a significant share of newspaper articles highlights potential benefits, such as job creation and tax revenue, these benefits are increasingly questioned in terms of their scale and distribution. At the same time, concerns have intensified regarding pressures on energy, water, and land resources, as well as impacts on environmental quality, community identity, and local governance. These issues are widely reported as affecting rural communities through both environmental and socioeconomic channels, with implications for agricultural production and rural livelihoods. If left unaddressed, these externalities may generate broader challenges for social development and equity, as economically disadvantaged rural communities may disproportionately bear the costs of technological advancement.\u003c/p\u003e \u003cp\u003eGiven these developments, there is a clear need for future research that systematically evaluates the environmental and socioeconomic impacts, as well as the associated trade-offs, of data center expansion in rural regions. Researchers and policymakers should work toward developing governance frameworks that support evidence-based decision making and balance technological advancement with social development.\u003c/p\u003e"},{"header":"7. Limitations","content":"\u003cp\u003eThis study has two main limitations. First, media coverage may not represent all perspectives, and further research is needed to incorporate views from stakeholders across diverse demographic and socioeconomic backgrounds. Second, this study employs a simplified sentiment analysis based on tone classification. While this approach provides a useful overview, it may oversimplify complex narratives. Despite these limitations, this study provides a valuable foundation for informing future research and offers preliminary yet policy-relevant insights that can help guide more data-driven decision-making as rural data center development continues to evolve.\u003c/p\u003e"},{"header":"8. Statements and Declarations","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eData availability\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets analysed during the current study are available in the supplementary file.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eCompeting interests\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eFunding\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNo funding was received for this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eEthics declarations\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEthical approval was not required as no human participants were involved in this research.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eAndsager JL (2000) How Interest Groups Attempt to Shape Public Opinion with Competing News Frames. Journal Mass Commun Q 77:577\u0026ndash;592. https://doi.org/10.1177/107769900007700308\u003c/li\u003e\n \u003cli\u003eBoguraev B, Kennedy C (2002) Salience-based Content Characterisation of Text Documents\u003c/li\u003e\n \u003cli\u003eHassan A (2026) \u0026lsquo;Nobody Owns Us\u0026rsquo;: How Plans for a Google Data Center Roiled an Oklahoma Town. N. Y. Times\u003c/li\u003e\n \u003cli\u003eIbrahim OAS, Landa-Silva D (2016) Term frequency with average term occurrences for textual information retrieval. Soft Comput 20:3045\u0026ndash;3061. https://doi.org/10.1007/s00500-015-1935-7\u003c/li\u003e\n \u003cli\u003eJegham N, Abdelatti M, Koh CY, et al (2025) How Hungry is AI? Benchmarking Energy, Water, and Carbon Footprint of LLM Inference\u003c/li\u003e\n \u003cli\u003eLee N (2026) Supporting rural communities amid the data center boom. Brookings\u003c/li\u003e\n \u003cli\u003eLei N, Lu J, Shehabi A, Masanet E (2025) The water use of data center workloads: A review and assessment of key determinants. Resour Conserv Recycl 219:108310. https://doi.org/10.1016/j.resconrec.2025.108310\u003c/li\u003e\n \u003cli\u003eLiu Y, Wang H (2026) Who on earth is using Generative AI? World Dev 199:107260. https://doi.org/10.1016/j.worlddev.2025.107260\u003c/li\u003e\n \u003cli\u003eMakholm JD, Olive LTW, Beiser EA (2025) Hyperscale Data Centers, Regulatory Institutions, and US Economic Growth. Clim Energy 41:21\u0026ndash;27. https://doi.org/10.1002/gas.22441\u003c/li\u003e\n \u003cli\u003eRadhakrishnan S, Erbis S, Isaacs JA, Kamarthi S (2017) Novel keyword co-occurrence network-based methods to foster systematic reviews of scientific literature. PLOS ONE 12:e0172778. https://doi.org/10.1371/journal.pone.0172778\u003c/li\u003e\n \u003cli\u003eShin R (2026) Trump\u0026rsquo;s AI drive is running headlong into his most reliable voters. POLITICO\u003c/li\u003e\n \u003cli\u003eSiddik MAB, Shehabi A, Marston L (2021) The environmental footprint of data centers in the United States. Environ Res Lett 16:064017. https://doi.org/10.1088/1748-9326/abfba1\u003c/li\u003e\n \u003cli\u003eStanford Institute for Human-Centered AI (2025) Artificial Intelligence Index Report 2025\u003c/li\u003e\n \u003cli\u003eSynergy Research Group (2025) Hyperscale Data Center Count Hits 1,136; Average Size Increases; US Accounts for 54% of Total Capacity. Reno, NV\u003c/li\u003e\n \u003cli\u003eTan Y, Weaver DH (2009) Local Media, Public Opinion, and State Legislative Policies: Agenda Setting at the State Level. Int J Press 14:454\u0026ndash;476. https://doi.org/10.1177/1940161209336225\u003c/li\u003e\n \u003cli\u003eTewksbury D, Jones J, Peske MW, et al (2000) The Interaction of News and Advocate Frames: Manipulating Audience Perceptions of a Local Public Policy Issue. Journal Mass Commun Q 77:804\u0026ndash;829. https://doi.org/10.1177/107769900007700406\u003c/li\u003e\n \u003cli\u003eZhang H, Wang D, Wu W, Hu H (2012) Term frequency \u0026ndash; function of document frequency: a new term weighting scheme for enterprise information retrieval. Enterp Inf Syst 6:433\u0026ndash;444. https://doi.org/10.1080/17517575.2012.665945\u003cstrong\u003e\u003c/strong\u003e\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Prairie View A\u0026M University","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Data center expansion, rural development, agriculture, energy-water-land nexus, resource competition, media framing","lastPublishedDoi":"10.21203/rs.3.rs-9442287/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9442287/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe artificial intelligence boom relies on the rapid expansion of data centers, which provide the infrastructure necessary for data storage and computational power to support AI services and development. These facilities consume substantial resources to sustain their operations. Technology companies are increasingly locating data centers in rural regions to take advantage of resource availability and lower costs, creating new opportunities while also introducing challenges for local communities. This study provides an initial exploration of the environmental and socioeconomic implications of this trend by examining how media frame rural data center expansion in its early stages. We find that media coverage has grown rapidly, with an increasing dominance of negative framing. Media narratives highlight conflicts between data center development and rural residents over resource competition, environmental concerns, community identity, and local governance processes. These dynamics carry significant environmental and socioeconomic implications for rural regions. They may alter human-nature interactions, disrupt agricultural production, and affect income stability and living standards, while potentially exacerbating economic inequality and social justice concerns. Our findings highlight the need for future research and policy efforts to better balance technological advancement with rural development, including supporting community sustainability and improving transparency in governance processes to promote accountability and public trust.\u003c/p\u003e","manuscriptTitle":"AI’s Hidden Footprint: Media Framing of Rural Data Center Expansion","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-20 05:20:59","doi":"10.21203/rs.3.rs-9442287/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f639b5c5-86a7-453e-ae98-73e47d23db63","owner":[],"postedDate":"April 20th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":66477422,"name":"Agricultural Economics \u0026 Policy"},{"id":66477423,"name":"Environmental Policy"},{"id":66477424,"name":"Social Policy"},{"id":66477425,"name":"Environmental Economics"}],"tags":[],"updatedAt":"2026-04-20T05:21:00+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-20 05:20:59","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9442287","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9442287","identity":"rs-9442287","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}