In Jay, Maine, the story of industrial decline and reinvention is playing out on a scale that would have been hard to imagine even a decade ago. The Androscoggin paper mill—once a major employer in a rural town roughly 67 miles northwest of Portland—was built around a simple promise: steady work, predictable output, and a local economy that could count on the rhythms of manufacturing. At its peak, the mill employed about 1,500 people. Then came the moment that changed everything. In 2020, a pulp digester exploded, and the facility was forced to close permanently.
For many rural communities, that kind of closure doesn’t just end a job—it ends a whole ecosystem. Suppliers lose customers. Local businesses lose foot traffic. Tax bases shrink. Families move away, and the remaining residents are left with a question that’s both practical and emotional: what happens to the buildings, the infrastructure, and the identity that once centered daily life?
Now, a different kind of industry is stepping into that same physical footprint.
Earlier this year, a resale agreement was finalized for the former Androscoggin mill site, a 1.4 million-square-foot facility that had been purchased in 2023 through a joint venture led by JGT2 Redevelopment and other holding and capital companies. The project is being led by developer Tony McDonald. Over the next three years, the team dismantled the mill’s machinery, shipped it to Pakistan, and worked to clean up the industrial site for reuse. The transformation is part of a broader pattern: large infrastructure needs tied to the data economy are increasingly pushing development into rural areas, where land, existing industrial space, and sometimes utility arrangements can make projects feasible.
This is not simply a story about “data centers replacing mills.” It’s a story about how the geography of computing is changing—and how rural America is being asked to absorb the consequences, opportunities, and tradeoffs of that shift.
The Androscoggin mill conversion is one example, but it also highlights a deeper dynamic: the data economy is no longer content to live only near major coastal hubs or dense urban power grids. As demand for cloud services, AI workloads, and always-on digital infrastructure accelerates, developers are searching for locations that can support massive electrical loads, provide reliable connectivity, and offer enough space to build at scale. Rural areas, particularly those with legacy industrial sites, can look attractive on paper. They may have large parcels, existing road access, and buildings that—after significant renovation—can be repurposed rather than built from scratch.
But the real question is what “repurposed” means in practice. A paper mill and a data center are both industrial facilities, yet they operate under very different constraints. Paper production depends on process water, chemical handling, and heavy mechanical systems designed for continuous throughput. Data centers depend on power delivery, cooling efficiency, redundancy, and the ability to maintain stable operating conditions for servers that cannot tolerate downtime.
That difference matters because it changes what communities will experience during construction and after operations. It also changes what kinds of jobs are likely to appear—and which ones disappear.
When a mill closes, the job losses are immediate and visible. When a data center arrives, the job picture is more complicated. Construction work can bring a surge of employment—often for electricians, HVAC technicians, civil contractors, and specialized trades. But long-term staffing for a data center is typically smaller than what a manufacturing plant might employ at peak. Many roles are technical and operational, and some are outsourced or filled by workers who commute from larger regions. That can leave communities with a familiar feeling: the building is new, but the economic benefits don’t always distribute evenly.
Still, the conversion of a 1.4 million-square-foot facility suggests something important: this isn’t a small pilot project. It’s an attempt to turn a large industrial asset into a platform for the next phase of infrastructure demand. That scale implies that the developer expects meaningful utilization—enough to justify the costs of dismantling, cleanup, and redevelopment.
And that brings us to the most overlooked part of these transformations: the logistics of making old industrial space usable again.
In the Androscoggin case, the project involved breaking down the mill’s machinery and shipping it to Pakistan. That detail matters because it shows the work wasn’t limited to cosmetic cleanup. Dismantling industrial equipment is time-consuming, expensive, and operationally complex. It also requires careful handling of materials and compliance with environmental and safety requirements. The fact that the team spent three years on dismantling and cleanup indicates that the site needed substantial preparation before it could be considered for reuse.
This is where rural redevelopment often diverges from the way people imagine “new tech” arriving. Data centers are frequently discussed as if they simply appear—like a box dropped onto a map. In reality, they are built through a chain of decisions: acquisition, permitting, engineering, utility negotiations, remediation, and construction sequencing. When the site is a former mill, the chain includes additional steps: removing legacy systems, addressing contamination risks, and reconfiguring the space to meet entirely different operational needs.
The result is that rural communities aren’t just hosting a new business. They’re becoming part of a long redevelopment process that can reshape local labor markets, land use patterns, and municipal planning priorities.
There’s also a political and cultural dimension to this shift. Rural America has long been asked to accept external investment—sometimes with promises of jobs and tax revenue, sometimes with concerns about environmental impacts, traffic, and strain on local services. Data centers add a new layer to that debate because they are tied to the digital economy, which can feel abstract compared to manufacturing. A paper mill produces something tangible. A data center produces nothing you can hold. Its value is measured in capacity, uptime, and the ability to run workloads that may be invisible to the people living nearby.
That invisibility can create skepticism. Residents may wonder: who benefits, exactly? Where does the power come from? How much water is used for cooling? What happens if demand changes? Will the facility remain stable for decades, or will it become another abandoned industrial shell?
Developers, for their part, often argue that data centers can be a stabilizing force. They can provide long-term leases, predictable revenue streams, and infrastructure upgrades that benefit the region. They may also bring improvements to roads, communications networks, and sometimes local energy infrastructure. But the credibility of those claims depends on transparency and on whether the community sees measurable outcomes beyond the initial construction phase.
The Androscoggin story also points to a broader trend: the reuse of industrial sites is becoming a strategy for meeting the data economy’s needs while avoiding the cost and disruption of building entirely new campuses. Reusing existing structures can reduce land consumption and potentially shorten timelines. It can also allow developers to leverage existing industrial zoning and infrastructure corridors.
Yet reuse is not a free lunch. Old industrial buildings come with structural constraints, layout limitations, and hidden costs. Retrofitting for data center operations may require significant electrical upgrades, new cooling systems, and careful airflow management. Even if the building shell is usable, the internal transformation can be extensive. That’s why the timeline matters. In Jay, the project took years to dismantle and prepare the site, culminating in a resale agreement finalized earlier this year.
That timeline suggests a level of commitment that goes beyond quick conversion. It implies that the developer is preparing for a long-term operational plan, not just a short-term real estate play.
Still, the “rural data center boom” narrative can obscure the unevenness of outcomes. Not every rural community will experience the same benefits. Some places may gain construction jobs and improved infrastructure. Others may see limited local hiring, higher costs for public services, or environmental concerns that are difficult to quantify until operations begin.
Even within the same state or region, the details can vary widely depending on utility capacity, grid interconnection timelines, and the availability of fiber and network routes. Data centers are power-hungry and connectivity-dependent. If a rural area lacks sufficient grid capacity, the project may require costly upgrades or may be delayed. If the fiber backbone is far away, connectivity costs can rise quickly. These factors influence whether a conversion becomes a thriving hub or a stalled development.
That’s why the Androscoggin case is worth watching closely. It’s not just a headline about a mill turning into a data center. It’s a test case for how rural redevelopment can align with the data economy’s infrastructure demands.
There’s also a subtle but important point about what these conversions say about the future of industrial America. For decades, the dominant story was that manufacturing jobs were leaving. Rural towns built around mills, plants, and factories faced closures that seemed irreversible. Now, a different kind of industrial activity is moving in—one that is still industrial in its footprint and infrastructure requirements, but whose purpose is to support computation rather than physical goods.
That shift doesn’t necessarily mean rural communities are returning to prosperity. But it does suggest that the industrial landscape is evolving rather than disappearing. The question becomes whether the new industry can deliver economic stability in ways that match the scale of what was lost.
In Jay, the mill’s closure in 2020 ended employment for about 1,500 people at peak. A data center may not replicate that number of jobs. However, it could create a different kind of employment mix: fewer positions, but potentially higher wages for certain technical roles; more contract work during construction; and ongoing demand for maintenance, security, and operations.
Whether that translates into broad community benefit depends on hiring practices, training pipelines, and the willingness of developers to invest locally. It also depends on how local governments manage permitting, infrastructure planning, and community engagement.
Another factor is the environmental and resource conversation. Data centers require cooling, and cooling strategies can involve air, water, or hybrid systems depending on design and local conditions. Power generation and transmission also carry environmental implications, especially if the electricity supply is not fully decarbonized. Communities that have already experienced industrial pollution may be particularly sensitive to new developments