Beyond Tax Revenue: What Does a Data Center Leave Behind?

The proposal appeared to have many of the features communities are told to seek. It would reuse former industrial property rather than convert undeveloped land. Local officials wanted it. Maine’s governor publicly supported it. Project supporters projected roughly $550 million in investment, more than 800 construction jobs and at least 100 permanent positions.

Then the company expected to operate the facility withdrew.

By June 2026, the developer said the project was on hold. The jobs, tax revenue and industrial revival that had shaped the state debate remained projections rather than delivered benefits.

Jay demonstrates the central weakness of measuring a data-center project by its announcement.

Capital investment is not the same as durable local value.

Construction employment ends. Operators change. Financing fails. Technology companies reorganize. Utility forecasts shift. A community can spend years planning around a project that never becomes operational.

The better question is not how much money a developer has announced. It is what the community will still own, control or be able to enforce if the original developer, operator or customer changes course.

That standard changes how projects should be compared.

A university-centered technology project may be smaller than a hyperscale campus but leave behind academic infrastructure and institutional control. A former industrial site may offer valuable land and transmission assets but remain vulnerable to the loss of one operator. A multi-gigawatt project may include sophisticated utility protections while offering community benefits that remain difficult to verify. A state may attach workforce and environmental conditions to public support without applying the same requirements to every privately financed development.

Durable local value comes from ownership, binding agreements, reusable infrastructure, workforce institutions and exit protections. It does not come automatically from the size of a press release.

Jay: a project can shape public policy before it exists

Jay lost a major industrial anchor when the Androscoggin Mill closed. The town had understandable reasons to support a new use for the site.

Former industrial properties can be attractive locations for data centers because they may already include industrial land, utility corridors, transportation access, large buildings and a workforce familiar with complex facilities. They also offer the possibility of restoring a damaged local tax base without converting undeveloped land.

The proposed Jay data center became central to Maine’s statewide policy debate. Lawmakers passed legislation that would have temporarily restricted permitting for large data centers. Governor Janet Mills vetoed the measure because it did not exempt the former mill proposal, and lawmakers sustained the veto.

The project therefore influenced public policy before its operator, financing and long-term operation were secure.

That sequence should concern communities evaluating major infrastructure proposals.

A town may begin planning roads, utilities, workforce programs and future tax revenue around benefits that depend on one operating company, one development partner, one financing structure and one anchor customer.

When the expected operator withdrew, Jay still possessed the former mill site. It did not possess the promised data-center economy.

The Jay lesson

Brownfield redevelopment can be preferable to greenfield development, but a brownfield location does not make the proposed benefits durable.

Communities should ask before granting approvals or restructuring public policy:

• Is the operating company contractually committed?
• Has financing closed?
• Are deposits or performance guarantees in place?
• Who pays for public predevelopment work if the project stops?
• Can permits and engineering studies transfer to another user?
• Are utility improvements useful to other industries?
• Does the site retain a viable fallback use?
• Has the community delayed alternative redevelopment opportunities?

A project should not become the foundation of a local recovery strategy until the community knows what survives if the first operator leaves.

Fisk University: ownership changes the negotiating position

Fisk University’s proposed Quantum Leap Innovation and Technology Center offers a fundamentally different model.

Fisk describes Quantum Leap as a $900 million campus transformation that includes a planned $400 million Innovation and Technology Center. The proposed 100,000-square-foot facility would occupy five undeveloped campus acres, with approximately 70,000 square feet devoted to data and technology space and 30,000 square feet devoted to academic uses.

Fisk describes the project as an HBCU-owned facility built on university land and integrated with its academic mission. The university says it is intended to give students direct access to advanced technology, support new curricula and research, expand workforce development, reduce the digital divide and create opportunity in the surrounding community.

The project is smaller than a hyperscale campus. That is part of its significance.

Fisk is not presenting the facility primarily as a large industrial load located near a university. It is presenting advanced computing and technology infrastructure as part of the university’s academic and civic mission.

Institutional ownership can give Fisk leverage that a conventional host municipality may not possess.

If the university retains meaningful control over the land, buildings and operating structure, it may be able to protect academic access, research capacity, curriculum integration, workforce training, recurring revenue and community programming.

That does not mean the model is already proven.

The public materials reviewed for this article do not fully disclose the commercial operator, ownership of the computing equipment, financing structure, portion of capacity reserved for Fisk, revenue-sharing terms, default protections or obligations that would bind a future operating partner.

Fisk has established the intended mission. The operating structure will determine whether the university retains practical control.

The Fisk lesson

Ownership matters because it changes the starting point of the negotiation.

A community receiving only tax revenue must ask a private owner for access, reporting and benefits. An institution that owns the land and shapes the development can potentially define those conditions before selecting an operator.

The Fisk model should therefore be judged by a different set of questions than a conventional hyperscale campus:

• How much capacity is reserved for students and faculty?
• Who decides which research projects receive access?
• Does commercial revenue subsidize academic programs?
• Are neighborhood programs guaranteed?
• Do operating obligations survive a change in partner?
• Does Fisk retain ownership of buildings, fiber and energy infrastructure?
• Can the facility continue serving the university if commercial demand changes?

Fisk is the strongest ownership-oriented concept in this comparison, but it remains a proposal whose operating and financing structure requires public clarification.

Project Camellia: ratepayer protection is not the same as community benefit

OpenAI says Project Camellia in Effingham County, Georgia, is being designed around a 3.2-gigawatt Georgia Power service plan, with capacity expected in phases between 2028 and 2032.

OpenAI’s announcement emphasizes jobs, community investment, responsible water design, AI training and the expectation that project-specific infrastructure costs will not be shifted onto existing electricity customers.

Georgia Power has described the service arrangement as potentially including up to 1,000 megawatts of demand response. The utility is also planning a new 500/230-kilovolt substation and related transmission infrastructure intended to serve the large load and strengthen the surrounding grid.

Those commitments are important, particularly at a scale large enough to influence generation and transmission planning.

They also illustrate a distinction that is often lost in public debate.

Ratepayer protection

Ratepayer protection asks whether the large customer pays for its interconnection, dedicated infrastructure, minimum service obligations, stranded assets and contract termination exposure.

Georgia Power’s large-load framework includes measures such as customer-specific infrastructure payments, long-term contracts, collateral, minimum monthly bills and termination protections.

Community benefit

Community benefit asks whether local residents receive permanent jobs, defined wages, apprenticeships, local procurement, school and college partnerships, water protections, emergency-service support, public reporting and reusable local assets.

A project can satisfy utility cost-allocation rules without delivering a complete civic-benefit package.

OpenAI’s public announcement describes broad community intentions. The public record reviewed for this article does not contain a complete community benefit agreement specifying permanent-job minimums, wage requirements, local-hiring boundaries, annual community investment, workforce-training numbers, water-use limits, reporting obligations, remedies for missed commitments, successor liability or decommissioning security.

The utility contract and regulatory process may provide meaningful protection against direct cost shifting. They do not automatically establish a complete civic bargain.

Demand response must also be defined

The proposed demand-response capability could provide substantial grid flexibility. Its value depends on contractual details:

• Is participation mandatory?
• How quickly must load be reduced?
• How often can the utility call it?
• Are there penalties for failure?
• Is reduction achieved by curtailing computing work?
• Does the site switch to backup generation?
• Are emissions created elsewhere when grid consumption falls?

A number in a public announcement is not the same as a demonstrated operating result.

The Camellia lesson

Hyperscale projects should be evaluated through multiple contracts, not one corporate statement.

The relevant documents may include utility service agreements, development agreements, tax incentive agreements, water and wastewater agreements, workforce agreements, community benefit agreements, emergency-response agreements and decommissioning security.

The project’s durable value will depend on how those documents allocate costs, assets and obligations over several decades.

Pennsylvania: public support can create enforceable leverage

Pennsylvania has taken a different approach by connecting state support to defined development standards.

Pennsylvania and Amazon announced a planned investment of at least $20 billion and projected at least 1,250 technology jobs across multiple future campuses. Those figures are announced commitments, not completed outcomes.

In 2026, Pennsylvania published Governor’s Responsible Infrastructure Development standards, known as GRID. The standards establish conditions for data-center projects seeking Commonwealth support, including coordinated state assistance, faster permitting and access to certain public incentives.

The framework addresses energy affordability, grid responsibility, workforce development, job quality, environmental protection, transparency and community participation.

The standards matter because public support creates leverage.

A state may not impose every requirement on every privately financed project, but it can condition valuable benefits on compliance. Those benefits may include tax incentives, expedited permitting, agency coordination, infrastructure support, grants and site-development assistance.

A developer can decline public support. It should not be able to receive public benefits without satisfying the conditions attached to them.

Pennsylvania lawmakers have also advanced legislation that would codify portions of the GRID framework for projects seeking state tax benefits. The legislative process was not complete in the sources reviewed, so the administrative standards should not be presented as universal statewide law.

Falls Township: brownfield reuse with continuing obligations

Amazon’s proposed Falls Township campus would occupy approximately 250 acres at Keystone Trade Center, a former U.S. Steel industrial property.

The location strengthens the land-use case. The property was previously industrial, and the project may reuse portions of existing regional infrastructure while avoiding some greenfield conversion.

Brownfield location does not eliminate environmental or infrastructure scrutiny.

Pennsylvania’s Department of Environmental Protection held a July 2026 community meeting on a proposed air-quality approval for the development, showing that backup generation and other emissions remained active regulatory questions after the major investment announcement.

The public should still ask:

• How many permanent jobs belong to this site?
• What tax incentives apply?
• Which existing assets are actually being reused?
• Which new substations and lines are required?
• How many backup generators are proposed?
• What emissions limits apply?
• What happens if the campus is never fully built?
• Can the site support another industrial user?

The Pennsylvania lesson

Conditions attached to public support can be one of government’s strongest tools.

The effectiveness of those conditions depends on whether they are binding, which projects they cover, how compliance is measured, whether reports are public, whether benefits can be revoked and whether obligations bind future owners.

A standard without monitoring is a request. A standard tied to money, approvals and enforceable remedies can become durable leverage.

Homer City: what survives when the customer changes?

The former Homer City Generating Station offers another form of infrastructure reuse.

Pennsylvania identifies the property as a decommissioned two-gigawatt coal-fired power plant that stopped operating in 2023. Homer City Redevelopment proposes converting the site into a large natural-gas-powered data-center and generation campus.

The site may offer major advantages:

• Existing transmission connections
• Large industrial acreage
• An established power-sector workforce
• Transportation infrastructure
• A community accustomed to industrial operations
• Potential continued use of a strategically important energy site

But the project is not a clean-energy conversion. It would substitute new natural-gas generation and computing demand for the former coal operation.

That creates several tests of durable value.

Workforce transition

• How many former coal workers can obtain jobs?
• What training is funded?
• Are construction jobs converted into permanent careers?
• Are local unions and community colleges involved?

Infrastructure ownership

• Who owns the new generation?
• Who owns the transmission connection?
• Can the power infrastructure serve other customers?
• Do useful assets remain if the initial data-center tenant leaves?

Customer concentration

• Is the generation dependent on one computing customer?
• What happens if that customer delays construction?
• Can power be sold to the wider grid?
• Are minimum payments guaranteed?

Environmental transition

• How do lifecycle emissions compare with the former coal plant?
• What methane assumptions are used?
• How much water is required?
• Who remains liable for former and future contamination?
• Is decommissioning security required?

The Homer City lesson

A former power plant can provide an unusually valuable platform for new development. The redevelopment is durable only if its assets retain value without the first computing customer.

If generation, transmission and site improvements are usable by other industries or the broader grid, the community retains options. If every asset is designed around one private customer, the project may recreate the dependency that made the original plant closure so damaging.

The missing document: a public community benefit agreement

Across the projects reviewed, complete publicly accessible community benefit agreements were difficult to locate.

The public sees press releases, project fact sheets, political announcements, utility filings, job projections and statements of intent.

What it often does not see is one consolidated agreement answering:

• What is legally required?
• When must it be delivered?
• Who measures performance?
• What reports become public?
• What happens when targets are missed?
• Do obligations bind a successor?
• What happens when the project closes?

The absence of a publicly accessible agreement does not prove no private agreement exists. It means the accessible public record does not allow independent verification of many promised benefits.

A serious community benefit agreement should identify jobs, education, infrastructure, environmental obligations, financial security and successor responsibilities.

The exit test

Jay shows why every major project should be evaluated under an exit test before construction begins.

If the project stops before construction

• Does the community retain deposits?
• Are public planning costs reimbursed?
• Can permits and studies be reused?
• Can another operator take over?

If construction stops halfway

• Who secures the site?
• Who pays contractors?
• Who removes unfinished equipment?
• Who restores roads and drainage?

If the operator leaves after opening

• Are taxes or lease payments guaranteed?
• Can another user occupy the facility?
• Who owns the substation and fiber?
• Do community programs continue?
• Are employees protected?

If the facility closes permanently

• Is there a decommissioning bond?
• Who removes batteries, generators and cooling systems?
• Who remediates contamination?
• Can the building be reused?
• Does the host inherit a specialized structure with no market?

A community does not possess durable value when all benefits disappear with the original customer.

A durable-value scorecard

Future projects should be scored using a common set of questions.

The largest announcement is not necessarily the best project

Fisk may create less computing capacity than a hyperscale campus, but it could retain academic ownership and access.

Jay promised industrial revival but showed how quickly projected value can disappear when one operator leaves.

Project Camellia may include unusually sophisticated ratepayer and demand-response protections, but broader community commitments still require enforceable detail.

Pennsylvania’s GRID standards show how public support can be exchanged for accountability, while its brownfield projects demonstrate that industrial reuse still requires environmental and fiscal review.

Homer City may preserve strategically valuable power infrastructure, but its long-term value depends on ownership, emissions, workforce transition and whether the assets remain useful without one data-center customer.

The central lesson is straightforward:

The most valuable data-center project is not the one with the largest capital announcement. It is the one that leaves the host with assets, skills, institutions and enforceable rights that remain useful after the original developer’s plans change.

Tax revenue matters.

Construction jobs matter.

Investment matters.

But durable local value begins with what the community keeps.

Sources