Rise of AI Datacenters in Michigan

Paper originally written on Dec 5, 2025 as an assignment for Professor Cantor's Political Ecology class at PSU

The goal of this paper is to discuss data centers and their impact on their surroundings in terms of the environment and community from a political ecology lens. Specifically, a deeper dive into the Saline Michigan “Stargate”  data center project being pushed by Oracle & OpenAI.

There has been a boom in the technology industry surrounding the usage of AI in everything from social media to the workplace. AI isn’t an intangible object though, the engine of this new technology is based in physical locations which hold real-world impact on the environment and communities. The computing done by AI takes place in a location known as a data center. These are often warehouses or specially-built buildings where technology is stored for the purposes of computing, data storage and so forth. With 13 billion reported in annual profit from Microsoft’s AI programs alone, in my opinion there is a clear motivator for companies to start investing in their own data center infrastructure (Bishop). 

There are currently 5,427 data centers operating in the U.S. (Leichter). An article on decarbonization and urban sustainability of data centers proposed, based on current estimations, data centers can be expected to have a compounded annual estimated growth rate between 5-11% (Liu, et al.). Like most situations, the rapid rate at which the data centers are being introduced is being scrutinized by environmental advocates. Data centers utilize land, water and energy in large quantities and often without consulting the local community members for their expertise and opinions beforehand. 

Ground Coverage

Location selection is based largely on cost, natural disasters, utility availability and tax advantages. After these are taken into consideration, data center developers have two choices: greenfield or brownfield development. From purely a land-use perspective, brownfields are ideal. Preexisting industrial sites are often cheaper and quicker to develop making them ideal for tight timelines. They also have the bonus of minimizing the physical disturbance construction can wreck on native habitats, garnering LEED building points and typically existing within commercially-zoned areas already containing the necessary infrastructure required to support larger operations. However, greenfields can have their own advantages, especially for larger companies who have the capital to invest with. These areas can provide more space with the capability for full customization from the bottom up. This would allow developers greater opportunities to innovate new ways to design a more efficient data center. 

Water Utilization

Water plays an important role in data centers in keeping the equipment from overheating. According to Pew Research, in 2023, data centers in the U.S. directly consumed 16 billion gallons of water with the expectation of consuming 16 to 33 billion gallons annually (Leppert). That’s an average of 2,948,221 gallons of water per data center, equivalent to roughly 24 four-person households annual water consumption (EPA). When looking at the indirect water consumption the number becomes considerably larger.  In a study conducted on the total environmental footprint of data centers looking at water consumed directly and indirectly through electricity generation alongside water and wastewater utilities to total 136 billion gallons overall back in 2018 (Siddik, et al.). But what does this mean for the rest of us?  According to the 2021 Infrastructure Report Card on waste water systems, the U.S. currently only meets 37% of the total infrastructure capital needs with a D+ rating (Infrastructure Report Card 157). With such an abysmal rating on our water infrastructure currently it stands to reason the increased presence of data centers will exacerbate an already strained system. Some data center locations consume as much as 30% of the municipal water supply making them a primary stakeholder in the effects poor waste water management has on communities (Martin & Ling). This could affect treatment plants and their ability to safely process water, maintain adequate pipelines and ensure service to all local residents. This is a situation already taking place in Morrow County, Oregon. The lack of proper wastewater treatment practices in combination with heavy industrial utilization from Amazon’s data center and big agricultural groups has resulted in water with levels of nitrates far past state and federal limits leading to rampant health issues in the community according to a Rolling Stone Investigative article (Cooper). Clean water is a right, in my opinion. It would be ideal if prior to construction, companies with intent to build large scale operations invested in proactively fixing and improving the water and waste water systems of the townships. Collaboration with local community and indigenous researchers could allow a better understanding on existing hydrology and how to minimize catastrophes. This way companies could prevent disastrous health and environmental crises while building a positive rapport with the local community. 

Energy Consumption

Data Centers consume massive amounts of energy. According to A Pew Research investigation, they consumed 183 terawatt-hours (TWh) of electricity in 2024 and are expected to increase to 426 TWh by 2030, both numbers surpassing the energy demand of some nations (Leppert). The verdict on whether consumption is good or bad varies by situation, metric and personal opinion. The energy itself can originate from a variety of sources, with natural gas supplying over 40% of electricity for data centers, and wind and solar combined constituting 24% of electricity while nuclear power supplied 20% and coal 15% overall (Leppert). While it would be ideal if every data center could operate utilizing a direct source of renewable energy, we are still limited in some of the capabilities the sector is able to handle. There is no greater boogeyman to the data center as a power outage. What this means though, is either investment mutually with local power providers which is beneficial to all or improvements and energy utilization costs are passed off to the local community without prior community involvement. According to a study from Carnegie Mellon University this could cause an average 8% spike in electricity bills to over 25% increase by 2030 (Wade, et al.). As mentioned, power outages are to be avoided by data centers, this is part of the reason they build in areas with minimal to no natural disaster presence at current time. However, most mission-critical and hyperscale data centers come equipped with numerous diesel generators to power operations during outage events (Cochran). Companies are keen to mitigate the damage fumes from circulating inside their centers as they can cause corrosion to the servers during run times however I was unable to locate information regarding mitigation of fumes outside of the buildings. At this time, Carnegie Mellon estimates 270.65 tons of CO2 will be released annually by 2030  (Wade, et al.). The practice of diesel generators long term could cause lasting harm to the local communities and land they are built on unless an alternative can be devised. 

While further investment into renewable energy sources directly supporting data centers would be ideal for future construction as mentioned earlier, the question stands on what the overall environmental and physical health impact will be as we wait for new innovations. While areas may be less prone to natural disasters and outages, the impact of these diesel generators can cause system wide issues from neuroinflammation to cardiovascular issues (Levesque, et al.). 

Saline Michigan 

In October, Related Digital in partnership with OpenAI and Oracle's Stargate program announced they would be constructing a $7 billion, 1.65 million sqft data center in Saline, Michigan on 575 acres of property (Bridge Michigan). This construction comes despite the pushback from the citizens of Saline, who originally voted against rezoning land for the company’s use but were forced to accept the development after a lawsuit was threatened by the developers for exclusionary zoning (Bridge Michigan). Compared to most situations throughout history of large institutions encroaching on a community, our usual landmarks aren’t here. The township is primarily white, above average income at a high rate of educational attainment according to the census report (“Census Reporter”). Related Digital claims part of their decision comes from the 345kV transmission lines running through the property as well as the site’s topography in their August proposal (Related Digital 1). 

Like most big construction, there are no environmental positives from construction. The area is a greenfield with active wetlands and 127 acres of solar currently on the site and yet, related digital claims the construction is “a superior alternative to other developments that will otherwise occur on the land” (Related Digital 1). This is a common theme throughout the project's proposal where the authors utilize persuasive writing techniques which focus on logos and pathos, frequently diminishing some of the potential impacts this project may have. The company also makes an appeal to the community’s “rural character” subjects. They make promises of land conservation, leasing to farming and restoration of the site to a natural area if the project is decommissioned (Related Digital 2). While these promises seem objectively favorable there are still concerns surrounding the site and its operations. 

Once construction starts on the site, given the large scale and the disturbance of wetland, the company is creating what's known as a novel ecology in political ecology. Even if they do shutter operations the restoration to a natural area will be similar to what it started out as will be an incredibly difficult and expensive endeavor due to factors such as soil compaction and change in chemical composition. 

Water usage is a huge concern with data centers and this project is no different. The project requires 20,000 gallons of water daily to maintain operations, more than double the average data center (Related Digital 3). The project is being built on a closed loop infrastructure where the water is originating from a 118’ deep well on the property and private septic system (Related Digital 3). In my opinion this leads to some deeper questions as to what their methods with the waste water will be. The site is primarily wetland and above grade next to the Saline River according to topographical maps of the area. Mismanagement of the water could lead to a similar situation from what Morrow County, Oregon is experiencing as this is a similarly agrarian community. This location also has the added issue of the hard water common to the midwest which is harsh on untreated pipes. The township itself has been fighting against corrosion caused by the minerals for years and holds bi-annual flushing to remove build up - when they have the water to spare (DPW). 

While Saline’s wells are far away from the building site, unfortunately they do appear to be at the same depth level as the data center based on the water well map. If these wells share the same aquifer, this could leave the township and data center competing for water usage.  In places such as my hometown in California we experienced issues with shallower wells drying up quicker once big agricultural companies drilled deeper, leaving many homeowners without water for around three to five months of the year. While neighbors are known to be more conscientious of their water usage to ensure the wells don’t run dry as quickly. We can look to Common Property Theory in political ecology as a framework for this to try and find a solution before some of the work goes forward. One where the disenfranchised community members are able to conduct hydrological surveys to determine what well depth and water policies would be best for the data center. 

Saline township has the fortune of running a reverse osmosis water treatment plant according to the public works department but it won’t protect the community members living off of private wells from potential increased toxins in their water supply. Even then, the city’s urban metabolism can only support so much strain. An increased demand from the water aquifer plus the potential for accelerated degradation of their only treatment plant if heavy metals from corroded pipes at the data center could wreck havoc on city budget and safety.

Luckily, in the state of Michigan, residents don’t need to worry about an increase to their electricity bills as the recent legislature prevents electricity ratepayers from footing the bill for improvements stemming from data center construction and implementation (Related Digital 5). Out of all the current issues with this data center the energy situation is lower on the list due to the pre-existing infrastructure on the site and commitment from the energy company to improve the grid. The project will still require the implementation of generators and a battery center which means there is a risk for environmental contamination not only locally from diesel fumes but also impact globally as the metals utilized are often sourced using marginalized labor from areas in African countries and China. 

The project is not fully underway yet, the citizens of this town have been advocates for their own environmental justice and emboldened by the backing of state representatives in their outcry. Their most recent win blocked DTE’s request to bypass public hearings and formal legal scrutiny (Eberbach). A crucial step in ensuring community involvement in the energy aspect of this project going forward . The next step will be the virtual hearing on December 18th on the permit application for Saline data centers proposed wetland, stream and flood plain impacts (Allnutt). 

Community involvement at this next stage could be crucial in helping minimize potential habitat damage while also calling attention to the lack of a clear water plan and how the data center will affect the overall water quality for the community. We can hope going forward Related Digital will be more mindful of their community and environmental impact and start to include community voices in the project to innovate solutions that work for everyone.   

Video made using Canva and Canva-based assets

Bibliography

Andrews, Joey and Hertel, Kevin. “Legislative Analysis House ENTERPRISE DATA CENTERS” Michigan Legislature, 20 May 2024, https://legislature.mi.gov/documents/2023-2024/billanalysis/House/pdf/2023-HLA-4906-4D5B41D0.pdf. Accessed 28 November 2025.

Allnut, Brian. “Michigan Regulators Blow Past DTE’s Data Center Deadline” Planet Detroit, 5 December 2025, https://planetdetroit.org/2025/12/dte-data-center-contract-decision/. Accessed 8 December 2025.

Bailey, Terell. “Michigan Public Service Commission hears concerns from residents over proposed data center” CBS Detroit, 4 December 2025, https://www.cbsnews.com/detroit/news/residents-concerned-data-center-saline-township-mpsc/. Accessed 7 December 2025.

Bishop, Todd. “Microsoft’s AI revenue run rate reaches $13B annually as tech giant tops earnings expectations” GeekWire, 29 January 2025, https://www.geekwire.com/2025/microsoft-earnings-2/. Accessed 7 December 2025.

Bridge Michigan. “https://planetdetroit.org/2025/10/saline-hyperscale-data-center/” Planet Detroit, 31 October 2025, https://planetdetroit.org/2025/10/saline-hyperscale-data-center/. Accessed 26 November 2025.

“Census Reporter: Saline, MI” Census Reporter, 2023, https://censusreporter.org/profiles/16000US2671140-saline-mi/. Accessed 4 December 2025.

Cochran, Brad C. “Avoiding Adverse Air Quality in Hyperscale Data Centers Due to Re-Entrainment of Diesel Exhaust.” ASHRAE Transactions, vol. 127, no. 1, 2021, pp. 259–66. Portland State University Library, https://go-gale-com.proxy.lib.pdx.edu/ps/i.do?p=ITBC&u=s1185784&id=GALE%7CA666206434&v=2.1&it=r&aty=ip.

Cooper, Sean. “The Precedent Is Flint’: How Oregon’s Data Center Boom Is Supercharging a Water Crisis” RollingStone, 24 November  2025, https://www.geekwire.com/2025/microsoft-earnings-2/. Accessed 3 December 2025.

DPW. Water Quality Updates. City of Saline, 2025, https://www.geekwire.com/2025/microsoft-earnings-2/. Accessed 5 December 2025.

Eberbach, Jennifer. “Protesters rally against DTE’s attempt to fast-track OpenAI, Oracle data center” Michigan Live, 2 December 2025, https://www.mlive.com/news/ann-arbor/2025/12/protesters-rally-against-dtes-attempt-to-fast-track-openai-oracle-data-center.html. Accessed 7 December 2025.

“Energy supply for AI” IEA, April 2025, https://www.iea.org/reports/energy-and-ai/energy-supply-for-ai. Accessed 4 December 2025.

Hanus, N., Newkirk, A., & Stratton, H.. “Organizational and psychological measures for data center energy efficiency: barriers and mitigation strategies”. Energy Efficiency, 16(1), Article 1. Portland State University Library, https://doi.org/10.1007/s12053-022-10078-1.

“Interactive Water Well Map for Saline, MI” WaterWellMap, 2025, https://waterwellmap.com/map/saline-mi. Accessed 5 December 2025.

Jackson, Colin. “Saline data center fight to arrive at Michigan Public Service Commission” WKAR News, 2 December  2025, https://www.wkar.org/wkar-news/2025-12-02/saline-data-center-fight-to-arrive-at-michigan-public-service-commission/. Accessed 7 December 2025.

Leichter, Rasmus. “Saline data center fight to arrive at Michigan Public Service Commission” Cargoson, November  2025, https://www.cargoson.com/en/blog/number-of-data-centers-by-country. Accessed 7 December 2025.

Leppert, Rebecca. “What we know about energy use at U.S. data centers amid the AI boom” Pew Reseach Center, 24 October 2025, https://www.pewresearch.org/short-reads/2025/10/24/what-we-know-about-energy-use-at-us-data-centers-amid-the-ai-boom/. Accessed 23 November 2025.

Levesque, S., Surace, M. J., McDonald, J., & Block, M. L. “Air pollution & the brain: Subchronic diesel exhaust exposure causes neuroinflammation and elevates early markers of neurodegenerative disease.” Journal of Neuroinflammation, 8(1), Article 105 2011. Portland State University Library, https://doi.org/10.1186/1742-2094-8-105.

Liu, F.H.M., Lai, K.P.Y., Seah, B. et al. “Decarbonising digital infrastructure and urban sustainability in the case of data centres.” npj Urban Sustain 5, 15 2025. Portland State University Library, https://doi.org/10.1038/s42949-025-00203-1.

Martin, Offutt and Ling, Zhu. “Data Centers and Their Energy Consumption: Frequently Asked Questions” Congress.gov, 26 August 2025, https://www.congress.gov/crs-product/R48646. Accessed 27 November 2025.

Related Digital. “Washtenaw County Data Center Proposal” Saline Township, August 2025, https://salinetownship.org/uploads/notices/QA%20Data%20Center%20Proposal.pdf. Accessed 7 December 2025.

Ren, Shaolei and Wierman, Adam. “Mitigating the Public Health Impacts of AI Data Centers” Harvard Business Review, 5 November 2025, https://hbr.org/2025/11/mitigating-the-public-health-impacts-of-ai-data-centers. Accessed 28 November 2025.

Roberts, Billy. Data Center Infrastructure in the United States. 2025. National Renewable Energy Laboratory, https://www.nrel.gov/docs/gen/fy25/94502.jpg

Siddik, M. A. B., Shehabi, A., & Marston, L. “The environmental footprint of data centers in the United States.” Environmental Research Letters, 16(6), Article 064017 2021. Portland State University Library, https://doi.org/10.1088/1748-9326/abfba1.

“Tracking Data Centers: Energy Demand, Pollution, and Public Impact” FracTracker Alliance, 29 July 2025, https://www.fractracker.org/2025/07/national-data-centers-tracker/. Accessed 2 December 2025.

Unosson, J., Kabéle, M., Boman, C. et al. “Acute cardiovascular effects of controlled exposure to dilute Petrodiesel and biodiesel exhaust in healthy volunteers: a crossover study.” Part Fibre Toxicol 18, 22 2021. Portland State University Library, https://doi.org/10.1186/s12989-021-00412-3.

“U.S. Data Centers and Power Demand” Aterio, 29 November 2025, https://www.aterio.io/insights/us-data-centers. Accessed 8 December 2025.

Velte-Schäfer, Andreas, et al. “Utilizing Waste Heat from Data Centers with Adsorptive Heat Transformation – Heat Exchanger Design and Choice of Adsorbent.” Energy Conversion and Management, vol. 310, 118500, 2024. Portland State University Library,  https://doi.org/10.1016/j.enconman.2024.118500.

Wade, C., Blackhurst, M., DeCarolis, J., et al. “Electricity Grid Impacts of Rising Demand from Data Centers and Cryptocurrency Mining Operations.” Open Energy Outlook for the United States, June 2025. https://energy.cmu.edu/_files/documents/electricity-grid-impacts-of-rising-demand-from-data-centers-and-cryptocurrency-mining-operations.pdf 

“Wastewater” Infrastructurereportcard, 2021, https://infrastructurereportcard.org/wp-content/uploads/2020/12/Wastewater-2021.pdf. Accessed 4 December 2025.