As artificial intelligence reshapes the global economy, communities increasingly face a familiar question: who benefits when new infrastructure becomes essential to economic and civic life? The discussion surrounding data centers and AI infrastructure is often framed as a choice between public ownership and private investment. Advocates of municipal or public compute argue for greater public control of critical digital infrastructure, while investors point to the enormous capital requirements involved in building the next generation of computing capacity. That framing may be too narrow. A more productive question is whether the extraordinary value expected from AI infrastructure can create a durable public dividend alongside private returns.

The billions of dollars flowing into data centers, fiber networks, power systems, and AI platforms reflect a broad expectation that artificial intelligence will become increasingly important to economic activity. It is still too early to know whether AI will ultimately prove as consequential as electricity, transportation, telecommunications, or broadband, and that comparison should remain a proposition rather than a settled fact. Even so, the scale of infrastructure investment and resource demand makes the public-interest question immediate: if AI becomes foundational infrastructure, communities should have pathways to participate in the value it creates rather than simply absorb its impacts.

The Missing Conversation

Today’s data center expansion is generating substantial debate about electricity demand, water use, land development, environmental effects, public incentives, and local governance. The U.S. Government Accountability Office reports that generative AI uses significant energy and water resources, while noting that company disclosures remain incomplete and that the portion of data center consumption attributable specifically to generative AI is still unclear.1 The Congressional Research Service similarly emphasizes that the effects of data center water use depend on local cooling technology, water sources, climate, and competing regional demands.2 Those qualifications matter because not every facility creates the same burdens. Nevertheless, communities are increasingly being asked to host infrastructure that serves regional, national, and global markets, often before governance practices have caught up with the speed and scale of development.

Public discussion frequently focuses on mitigating those impacts rather than sharing in the resulting value. History does not provide a single model for resolving that problem, but it does offer relevant precedents. The Communications Act of 1934 established nationwide access to communications service at reasonable charges as a federal policy objective, and the Telecommunications Act of 1996 expanded universal-service policy to include advanced services for rural and high-cost areas, schools, libraries, and rural health care providers.3 Rural electrification followed a different institutional path: the Rural Electrification Act of 1936 provided financing that enabled rural communities to form consumer-owned cooperatives when incumbent utilities showed limited interest in serving sparsely populated areas.4 These examples do not prove that AI infrastructure should be governed in the same way. They do demonstrate that when infrastructure becomes socially and economically essential, institutions often emerge to address access, affordability, accountability, and public benefit.

The next generation of AI infrastructure may require a comparable negotiation. Rather than asking only whether compute should be publicly or privately owned, policymakers and communities should ask what public benefit society should expect from infrastructure that may become as consequential as electricity or broadband.

The Idea of an AI Infrastructure Dividend

One possible answer is an AI Infrastructure Dividend: a structured commitment through which a portion of the value associated with AI infrastructure supports public-interest purposes over the life of the asset. Depending on the project, its ownership, and the authority of the host jurisdiction, those commitments might support workforce development, community college partnerships, public research, civic innovation, emergency management, public-interest AI services, community technology programs, or digital opportunity initiatives.

The key principle is permanence. Rather than relying entirely on voluntary philanthropy or periodic grants, communities could negotiate durable and enforceable benefit structures. Community benefits agreements provide one existing mechanism. These agreements are generally contracts through which developers commit to defined local benefits, often in exchange for community support or non-opposition. Recent research on data center development emphasizes that effective agreements require measurable commitments, transparency, implementation responsibilities, and enforcement rather than broad promises.5 Universal-service programs, cooperative patronage structures, public-purpose funds, development agreements, and endowments offer other precedents, although each operates under a different legal and institutional framework.

The objective is not to discourage private investment or to impose one uniform model on every project. It is to establish the expectation that communities hosting infrastructure with substantial local demands should have a meaningful role in defining and receiving long-term benefits. The appropriate mechanism would vary by jurisdiction, project scale, public incentives, resource use, and community priorities.

Why Cooperatives Matter

Cooperatives are particularly relevant because they provide an established mechanism for members to collectively own and govern enterprises. Rural electric cooperatives, telecommunications cooperatives, agricultural cooperatives, and credit unions demonstrate that member ownership can coexist with public agencies and investor-owned firms. The historical case for rural electric cooperatives is especially pertinent: in the 1930s, most rural homes and farms lacked electricity, and federal financing helped locally organized cooperatives extend service into areas that established utilities had not served.4

That history should not be romanticized or treated as an automatic template for AI. Cooperative ventures still require viable business models, capable governance, technical expertise, capital, regulatory compliance, and sustained member participation. Yet the model remains relevant because it can align infrastructure investment with the interests of member-owners and return value through service, local control, or patronage rather than through investor returns alone.

The transition of some electric cooperatives into broadband offers a modern precedent for institutional adaptation. The National Rural Electric Cooperative Association reports that more than 250 electric cooperatives are deploying broadband or developing plans to do so.6 Its 2025 broadband annual report states that 128 participating cooperatives served nearly one million families and businesses across 31 states.7 Those figures do not mean that compute services are a natural or inevitable next step. They do show that some cooperatives have already extended their capabilities from electric distribution into fiber networks, telecommunications operations, cybersecurity, regulatory compliance, and digital services.

Many rural areas are unlikely to attract hyperscale data center campuses. Yet those regions may possess assets of increasing value to a more distributed digital economy, including fiber infrastructure, electric utility networks, rights-of-way, educational institutions, workforce training capacity, community relationships, local governance structures, and distinctive local or regional data. Cooperative structures may offer one way to organize those assets for long-term participation in the AI economy.

A Cooperative Compute Hypothesis

One possibility worth exploring is the emergence of Cooperative Compute Utilities. This is a hypothesis proposed here, not an established industry category or a prediction that electric cooperatives will broadly enter the computing market. The concept asks whether some member-owned or community-serving institutions could eventually operate regional computing capacity in much the same way that some expanded from electric service into broadband.

Electric cooperatives already operate complex systems involving power distribution, communications, operational technology, cybersecurity, and field operations. Some also own fiber networks and maintain long-standing relationships with their member-owners. As AI applications develop, selected workloads may benefit from regional processing because of latency, data-sovereignty requirements, resilience goals, connectivity costs, or institutional partnerships. Other workloads will remain concentrated in hyperscale facilities because large-scale model training and cloud platforms benefit from specialized hardware, enormous capital investment, and economies of scale. The opportunity, therefore, is not to assume that community-scale compute will displace hyperscale providers, but to identify use cases for which local or regional capacity creates a distinct advantage.

Potential applications could include public-sector analytics, education and workforce training, precision agriculture, smart energy systems, local research partnerships, small-business experimentation, or privacy-sensitive services. A credible pilot would need to establish actual demand, utilization targets, power and cooling requirements, cybersecurity controls, staffing, capital and operating costs, procurement rules, data governance, and a sustainable revenue model. Without those elements, the idea would remain an appealing metaphor rather than a workable utility.

A Cooperative Data Commons Hypothesis

A second hypothesis involves data governance. Data is a critical input to many AI systems, but it is important to distinguish between its potential social or economic value and the claim that every dataset is inherently valuable. Value depends on quality, provenance, representativeness, legal authority, interoperability, relevance, and the ability to use data responsibly. Rural communities and local institutions may generate useful information through agriculture, forestry, transportation, health care, energy systems, and environmental monitoring while retaining limited influence over how that information is aggregated, governed, and reused.

A Cooperative Data Commons could provide a structure through which participating organizations share selected data under mutually agreed rules while preserving accountability, privacy, security, access controls, and local decision-making. This proposed term combines ideas from data commons and data-cooperative models, but those concepts should not be treated as interchangeable. Emerging research describes data cooperatives as member-governed organizations that enable participants to collectively manage and potentially benefit from data, while also warning that coordination, scalability, technical capacity, legal compliance, and member engagement remain significant challenges.8

Such a model could help communities develop local AI capabilities while aligning data use with community priorities and maintaining trust. In this scenario, the cooperative becomes more than an infrastructure provider. It becomes a steward of shared knowledge assets. Any practical implementation, however, would require clear rules for consent, ownership or custodianship, permitted uses, withdrawal, retention, licensing, auditability, benefit sharing, and responsibility when data produces harm or error.

Implications for Rural Regions

For regions such as Michigan’s Upper Peninsula, the opportunity may not lie primarily in attracting multibillion-dollar hyperscale developments. The more realistic opportunity may be participation through infrastructure ownership, workforce development, applied research, energy integration, data governance, and carefully targeted computing capacity. Electric cooperatives, municipal utilities, tribal nations, community colleges, universities, economic development organizations, and local governments already possess pieces of the physical infrastructure, institutional capacity, knowledge, and trust required for collaborative development.

The emerging Cloverland Training & Technology Center is especially interesting in this context. Cloverland has publicly described the center as a regional hub for hands-on learning, safety education, apprenticeships, and industry-recognized training spanning electrical systems, clean energy, broadband infrastructure, and other technology-driven fields.9 My own regional work has increasingly treated the center not simply as a workforce facility, but as a potential living laboratory where the EUP could connect training to intelligent infrastructure, applied research, energy systems, broadband, cybersecurity, and eventually carefully scoped regional compute or AI experimentation. That does not mean the center should become a commercial data center. Its more distinctive opportunity may be to give students, utilities, researchers, employers, and communities a place to learn how these converging systems actually work together.

There is a parallel institutional question for the EUPConnect Collaborative. EUPConnect began as a consortium and has grown into a cross-sector regional network spanning schools, local governments, tribal partners, higher education, healthcare, economic development, research, and network operators.10 Earlier in my own research, I described the collaborative itself as a form of civic infrastructure: relationships, shared knowledge, institutional memory, and the capacity to act together can become regional assets just as surely as fiber or facilities.11 The cooperative model explored in this article suggests a possible next step. Rather than remaining indefinitely a grant-oriented consortium, EUPConnect could explore whether a formal cooperative or cooperative-like membership organization would provide a durable structure for shared governance, resource pooling, regional data stewardship, workforce initiatives, research, and eventually community-owned digital services. This should be treated as an organizational hypothesis, not a predetermined restructuring. The important question is whether cooperative principles such as member governance, shared investment, mutual benefit, and long-term stewardship fit the institution EUPConnect has already begun to become.

That pairing creates an intriguing regional architecture: the Cloverland Training & Technology Center could provide a physical place for learning, demonstration, and applied experimentation, while an evolved EUPConnect could provide the regional connective tissue for governance, partnerships, shared knowledge, and cooperative initiatives. The two organizations would remain institutionally distinct, but could reinforce one another around the same objective: helping more of the value created by infrastructure investment remain in and circulate through the Eastern Upper Peninsula.

That observation should be treated as a regional strategy proposition rather than an empirical claim that the necessary coalition or market already exists. The next step would be to inventory regional assets and demand: available middle-mile and last-mile fiber, substation and feeder capacity, prospective compute users, educational programs, research partners, public-sector applications, relevant datasets, cybersecurity capacity, and organizations able to operate shared infrastructure. That inventory would help distinguish an economically grounded opportunity from a narrative built only around technological possibility.

The strategic question, therefore, is not simply, How do we attract a data center? It is How do we organize regional assets so that communities can participate in the value created by the AI economy? Regions that organize early may be better positioned to capture a modest but meaningful share of that value. Regions that do not may remain primarily purchasers of services developed and governed elsewhere.

From Infrastructure to Shared Prosperity

The debate should not be reduced to public versus private data centers. Private investment will remain indispensable because modern AI infrastructure requires substantial capital, specialized hardware, technical expertise, energy, connectivity, and operational scale. The more important institutional challenge is determining how communities, investors, educational institutions, utilities, cooperatives, and public organizations can participate together in creating long-term value while assigning costs and risks fairly.

Cooperatives offer one proven institutional framework worth examining, but not a universal answer. They occupy a space between conventional government ownership and investor ownership, and their member-owned structure can align infrastructure investment with community interests. Their relevance to AI will depend on whether they can develop viable technical and economic roles rather than merely borrow the language of earlier infrastructure movements.

The history of electricity, telephone service, and broadband supports a narrower and more defensible conclusion than the claim that every foundational technology inevitably produces the same institutional response. As infrastructure becomes essential, societies repeatedly confront questions of access, affordability, ownership, regulation, and public benefit. AI infrastructure is now bringing those questions into view. The issue is not simply whether communities should benefit from the AI economy, but whether we will build practical institutions that allow them to participate in shaping and governing it.

Notes and Sources

  1. U.S. Government Accountability Office, “Artificial Intelligence: Generative AI’s Environmental and Human Effects,” GAO-25-107172, April 22, 2025. https://www.gao.gov/products/gao-25-107172 ↩

  2. Congressional Research Service, “Data Centers and Water: Frequently Asked Questions,” R49057, July 31, 2026. https://www.congress.gov/crs_external_products/R/PDF/R49057/R49057.1.pdf ↩

  3. Federal Communications Commission, “Universal Service Fund”; Congressional Research Service, “The Universal Service Fund and Related FCC Broadband Programs: Overview and Considerations for Congress,” R47621, updated January 21, 2026. https://www.fcc.gov/general/universal-service-fund; https://www.congress.gov/crs_external_products/R/PDF/R47621/R47621.9.pdf ↩

  4. U.S. Department of Agriculture Rural Development, “USDA Celebrates 90th Anniversary of Rural Electrification Act,” May 27, 2026; National Rural Electric Cooperative Association, “History: America’s Electric Cooperatives.” https://www.rd.usda.gov/newsroom/news-release/usda-celebrates-90th-anniversary-rural-electrification-act-billions-invested-electric-infrastructure; https://www.electric.coop/our-organization/history ↩ ↩2

  5. Nicol Turner Lee and Darrell M. West, “Why Community Benefit Agreements Are Necessary for Data Centers,” Brookings Institution, January 29, 2026; Devashree Saha, Jolie Villegas, and Carla D. Walker, “Powering Progress: How Community Benefits Agreements Can Drive Responsible Data Center Development,” World Resources Institute, September 8, 2026. https://www.brookings.edu/articles/why-community-benefit-agreements-are-necessary-for-data-centers/; https://www.wri.org/technical-perspectives/community-benefits-agreements-data-centers ↩

  6. National Rural Electric Cooperative Association, “Broadband.” https://www.electric.coop/issues-and-policy/broadband ↩

  7. National Rural Electric Cooperative Association, “NRECA Broadband 2025 Annual Report.” https://www.cooperative.com/nreca/governance-reporting/Documents/2025_Broadband_Annual_Report.pdf ↩

  8. Francisco Mendonça, Giovanna Di Marzo, and Nabil Abdennadher, “Data Cooperatives: Democratic Models for Ethical Data Stewardship,” arXiv preprint, April 2025; Project Liberty Institute and Decentralization Research Center, “How Can Data Cooperatives Help Build a Fair Data Economy?” July 2025. The first source is a preprint and should be treated as an emerging conceptual review rather than conclusive evidence that data cooperatives work at scale. https://arxiv.org/pdf/2504.10058; https://www.projectliberty.io/wp-content/uploads/2025/07/How_can_data_coops_help_build_a_fair_data_economy.pdf ↩

  9. Cloverland Electric Cooperative, “Building Skills, Powering Progress,” June 24, 2026; Cloverland Electric Cooperative, “Building Skills & Powering Progress with New Training Center,” August 28, 2025. Cloverland describes the proposed center as a regional workforce hub supporting hands-on training, apprenticeships, safety education, electrical and energy systems, broadband infrastructure, and industry-recognized credentials. ↩

  10. EUPConnect Collaborative materials describe a regional partnership encompassing K–12 schools, local governments, tribal partners, higher education, healthcare, economic-development organizations, community organizations, and research and network partners. The cooperative organizational direction discussed here is the author’s proposed next-stage hypothesis, not a current legal description of EUPConnect. ↩

  11. Jason F. Kronemeyer, “The Real Infrastructure We Are Building,” August 20, 2026. The essay argues that EUPConnect’s accumulated relationships, governance practices, shared knowledge, institutional memory, and collective capacity constitute a form of civic infrastructure. ↩