In December 2025, Howell Township residents led a successful effort to halt a proposed $1 billion data center. Over 200 people attended a community panel where FracTracker and other experts presented analysis on energy demand, grid capacity, and backup generators. Following the meeting, the developer withdrew the project and zoning amendment, and the township enacted a six-month moratorium on new proposals. The case demonstrates how organized community action, supported by technical data, can influence local decisions on large-scale energy projects.
Update (Jan. 6, 2025): This article was updated to incorporate more records that were obtained for 2015-2017. In addition to the updated figures, FracTracker Alliance would also like to make readers aware of the “Daily Accident Report” started by Save Ohio Parks as a result of this issue.
Overview
Introduction
December 2025 – Recently, FracTracker’s Midwest Program Coordinator, Gwen Klenke, was asked to speak about the energy impacts of a proposed $1 billion data center in Howell, MI. She spoke alongside Dr. Ben Green, an assistant professor in the University of Michigan School of Information and an assistant professor (by courtesy) in the Gerald R. Ford School of Public Policy, Prescott Balch, a retired technology executive, and Andrea Pierce, a water protector and Policy Director at Michigan Environmental Justice Coalition.
The panel discussion was hosted by Livingston County Residents for Responsible Development (LCRRD), a grassroots group working to protect their county and beyond from data centers. Following Gwen’s participation at the LCRRD panel, the developers proposing the data center, Randee LLC, with consultation from Stantec, withdrew their rezoning request and the zoning text amendment that would allow the data center to be sited in the township. Additionally, a six-month moratorium on data center proposals has been enacted by the township board. These actions were largely the result of community pushback. By asking questions on social media and bringing in experts to testify, LCRRD successfully rallied their community and beyond. We are extremely grateful that they invited us to speak and collaborate on the issue more broadly.
While we are thrilled about this success, community members continue to keep a close eye on new developments and feel that the six-month moratorium should be expanded to a township-wide ordinance preventing data centers. The discussion highlighted many of the broader energy, infrastructure, and accountability issues raised by large-scale data center proposals, which this article explores in more detail.
Data Centers and Energy Concerns
Powering data centers
Data centers have been a part of technology infrastructure for quite some time. Yet their energy impacts have only recently moved to the forefront of public discussion. This can largely be attributed to the rise in generative artificial intelligence (AI). Because generative AI requires more data and is more computationally intense than previous computing, data centers being permitted today require much more energy to function.
Tech companies actively promote these energy-intensive technologies as they invest heavily in AI and compete to develop leading systems. As companies race to train increasingly complex models, they seek out virtually any available source of power to gain a competitive advantage. This has led to aggressive efforts to connect new data centers to existing electrical grids, as is the case with the Howell Township data center, as well as investments in nuclear power plants, battery storage, and renewable energy generation. Where electricity is available, tech companies are seeking access.
But electricity isn’t the only thing needed to power data centers. These facilities also depend on transmission infrastructure, such as power lines owned by utility companies that move electricity from the grid to the facilities. In addition, most data centers rely on on-site backup generators to maintain operations during grid outages. These backup generators, typically owned and operated by the data center rather than the utility companies, are worth additional scrutiny.
Generators
Often diesel-fueled, when these generators turn on, they can be particularly noisy, smelly, and they can increase air pollution levels in the area. In Michigan, backup generators are primarily governed by federal Environmental Protection Agency (EPA) air quality standards for stationary engines, with additional oversight by the Michigan Department of Environment, Great Lakes, and Energy (EGLE) in cases where state air permits are required based on the generator’s size, fuel type, or usage. But data centers are installing generators with “emergency” designations that allow them to run continuously during power outages. This could have become a concern considering that the Howell Township data center would have utilized DTE’s grid, which has become less reliable in recent years.
How often do we expect tech companies to take advantage of the “emergency” designation and run for longer periods of time? How often can we anticipate backup generators running?
In the case of Elon Musk’s xAI data center in Memphis, Tennessee, environmental groups documented that at least 33 of 35 methane-gas turbines appeared to be operating at once, despite the company publicly reporting a permit application for only 15 generators. This issue is now the subject of legal challenge and concern over air pollution and permitting compliance.
In a similar manner, transmission company PJM has been considering a proposal to allow placing ratepayers under rolling blackouts to accommodate the data centers. However, the data centers themselves could potentially be the subject of PJM’s consideration of rolling blackouts, meaning there may be an increased need to run generators. Nationwide, these data centers need electricity, but don’t have enough.
The Howell Township Data Center Proposal
In order to understand the Howell data center fight, it may be helpful to draw a comparison to the proposed Saline data center, which is one of the state’s largest and most contested proposed projects. Below is a snapshot of both projects (Figure 1; Table 1).
Figure 1. Map showing the locations of Michigan data center proposals (FracTracker, December 2025).
| Saline | Howell | |
| Status | Proposed | Cancelled |
| Estimated data center load | Hyperscale; 1.4 GW | Unconfirmed; likely hyperscale |
| Buildings | 3 large, 1 small | Unconfirmed; 10 reported |
| Investment | $7-10 billion | Unconfirmed; $1 billion reported |
| Initial reported campus footprint (acres) | 575 | 1,077 |
| Operator | Oracle, OpenAI, Related Digital | Unconfirmed; Alleged by a township trustee to be Meta |
Table 1. Snapshot of Howell Township and Saline data center proposals (FracTracker, December 2025).
While larger campuses often contain more facilities that could consume more power, the amount of land a data center occupies does not by itself determine how much electricity it uses. Those concerned about data center energy use should note that the Howell data center was projected to have roughly twice the land impact of a development already reported as potentially increasing the grid’s peak capacity by 25%. The Howell township project would have connected to the same grid. And, for readers unfamiliar with DTE’s grid mix, you can find last year’s information on the grid below (Figure 2).
![]()
Figure 2. DTE’s grid mix, composed of primarily coal and natural gas generated power (DTE, 2024).
At this point, readers may be wondering, “How big is 1.4 GW and what does hyperscale mean?” It is useful to pause here and clarify. Several basic comparisons to these power usage amounts are as follows:
- 1 MW is enough to power around 400 homes and a school
- 100 MW is generally considered “hyperscale,” which is a name for larger data centers
- 1 GW is comprised of 1,000 MW
- 1 GW is enough to power a city approximately the size of Orlando, Florida
The proposed 1.4-GW data center in Saline would use enough electricity to power a city roughly the size of Philadelphia. This project, along with the now-withdrawn Howell proposal, is part of a broader wave of data centers seeking to connect within DTE’s service area. Had it moved forward, the Howell Township data center would have added a substantial new load, compounding existing pressures on an already constrained grid.
Grid Capacity
While giving consideration to data center proposals, DTE is reportedly looking to add more capacity to its grid. In July, DTE let reporters at Reuters know they were in advanced talks to supply at least 3 GW of capacity to data centers. But by August, only a month later, DTE let Inside Climate News journalists know that it was in discussion for an additional 4 GW of potential load, bringing the total to around 7 GW of proposed data center demand. This trend is not unique to DTE; neighboring Consumers Energy has also stated an intention to increase its grid capacity by 15 GW. These corporate utilities are looking to add a substantial increase in power capacity to the grid for these data center projects.
It’s important to be clear: the reason DTE is looking to add this grid capacity is that, at present, existing capacity is insufficient for data center projects in the pipeline. Claims to the contrary do not reference actual capacity and demand. In Table 2, we can see that currently, on the most energy-intensive day, demand reaches roughly 9.5 GW, while installed capacity is approximately 11 GW. However, this capacity would be insufficient to accommodate the additional multi-gigawatt load being requested by proposed data centers. A similar pattern is evident in Consumers Energy.
| DTE | Consumers Energy | |
| Peak Demand | 9.5 GW | 7 GW |
| Grid Capacity | 11 GW | 7.6 GW |
| Estimated data center demand | 7 GW | 15 GW |
Table 2. DTE and Consumers Energy capacity and demand as reported (Inside Climate News, August 2025; MPSC 2025).
Seven GW of power doesn’t materialize without significant investment. In this case, DTE plans to add capacity to its grid by building more battery storage and adding more natural-gas generated power. The projects are large infrastructure projects that need up-front investments, which raises the question of who bears these costs.
Rate Hikes and How We Pay for Data Centers
In Michigan, most utilities are regulated monopolies. This means that DTE is the sole utility for those in the eastern part of the state. In practice, DTE provides ratepayers with electricity, and when they do, they notify the Michigan Public Service Commission (MPSC) of how much it costs. The MPSC considers if the costs are appropriate given the service provided, and if they are, passes the bills to ratepayers. If DTE improves service or “does more,” they can ask the MPSC to approve a rate increase, which can result in higher bills. One of DTE’s most recent requested rate hikes in November would raise residential ratepayers’ electric bills by up to 9%. It’s important to note that DTE is looking to add more capacity to the grid, specifically to power data centers, yet it has outstanding reliability issues.
When ratepayers pay their electric bill, they cover two primary categories of costs (Figure 3). The first is operating costs, which cover things such as company salaries, utility truck maintenance, and software. When ratepayers pay for this on their bill, it is paid back at cost. Meaning, if DTE employs 100 employees, all ratepayers split the cost of those 100 employees, but nothing more. The second kind of cost, capital expenses, includes large-scale physical assets, such as grid infrastructure. When ratepayers pay for this on their bill, they pay for the cost of the infrastructure plus an additional return. To build or install large-scale capital expenditures, corporate-owned utilities like DTE get loans from banks and investors. No one loans money without interest, so it can be expected that ratepayers are asked to pay a little more on their bills to pay back the investors’ initial costs in funding the infrastructure. However, this model can incentivize utilities to favor capital-intensive infrastructure investments, since returns are earned on capital expenditures rather than operating costs. Thus, DTE and corporate utilities are strongly incentivized by data center proposals that allow them to justify infrastructure investments that will ultimately be passed on to residential ratepayers.
![]()
Figure 3. Graphic from NRDC detailing capital vs. operating expenses for utilities.
Transparency
Having regulator-operated utilities is further complicated by issues of transparency. MPSC may face structural limitations in trying to set fair rates for residential customers, including a lack of access to utility-specific knowledge. DTE has precise knowledge of how much energy data center proposals on its grid are seeking. Additionally, DTE has precise knowledge of the capacity its residential customers need. But the structure is still set such that the more DTE builds infrastructure, the more profit they make. To further the limitations MPSC may face, DTE employs lawyers and lobbyists who push hard for “necessary” grid additions and rate increases. Both the tech companies and the utilities benefit from limiting disclosure on their energy usage and how it affects ratepayer bills. Tech companies often limit public disclosure about training efficiency due to competitive pressures, while utility companies provide limited transparency in ways that make it difficult for ratepayers to determine which infrastructure investments are necessary for residential customers versus data centers.
Relatedly, it’s important to consider the regional implications of the transparency issues. Folks fighting the nearby Saline data center were seeking a contested case–a formal regulatory proceeding that would have allowed for independent review and public challenge of the project’s energy impacts. DTE came out in strong support of the now-withdrawn Howell township data center proposal, but the Saline fight for reliable energy information indicates DTE’s support of any new data centers warrants concern. Ratepayers in Michigan and beyond should be concerned with bearing the financial burden to power technology that may not last.
Conclusion
The withdrawn Howell Township data center proposal was not merely a local land-use issue, but rather part of a much larger, regional energy and accountability challenge driven by the rapid expansion of hyperscale data centers. As utilities and tech companies compete to secure power, communities are being asked to absorb increased pollution risks, infrastructure buildout, and rising electricity bills, often with limited transparency or public oversight. The experience in Howell demonstrates that informed, organized community engagement can meaningfully slow or stop projects that raise serious questions about grid reliability, environmental impacts, and who ultimately bears the cost. As data center proposals continue to proliferate across Michigan and the United States more broadly, residents, regulators, and local governments must demand clearer information, stronger safeguards, and decision-making processes that prioritize public interest over corporate convenience.
For those interested in learning more about the Howell township win, energy burdens associated with data centers, or anything else discussed in this article, visit the following pages:
- Tracking Data Centers: Energy Demand, Pollution, and Public Impact
- Howell Data Center | LCRRD
- What Happens When Data Centers Come To Town?
- What we know about energy use at U.S. data centers amid the AI boom
- Energy and AI – Analysis – IEA
- Extracting Profits from the Public: How Utility Ratepayers Are Paying for Big Tech’s Power
- Utility Accountability: How Do Utilities Make Money?
- North Star Data Center Policy Toolkit
References & Where to Learn More
[1] Our data update, conducted in the fall of 2024, was meant to incorporate older data from 2014-2017. The 2014 data was not included in this analysis because it was the beginning of online tracking for oil and gas incidents, and as such, it contained no location data.
Additionally, recording only started in 2016, so incidents noted prior to then were backlogged by ODNR.
[2] This could be a direct mention of the pad name, an indication that a given piece of equipment that we know resides on the well pads was leaking, etc.
[3] You can find and download all data from this piece at https://app.box.com/s/r0sv3rkb7b0cl49wmdh1cqgnoe2em7c0
[4] The “Brief Description” field has what I would consider to be some of the most useful information ODNR provided on each record. They can vary widely in terms of completeness though.
[5] Notably, the authorized owners of this well could not pay for the costs associated with cleaning the accident and plugging the well, so the state (using taxpayer money) plugged the well and cleaned the site at a cost of over $1 million. You can find more info on this incident outcomes here.
Topics in This Article:
Join the Conversation
Stay Informed
Support Our Work
FracTracker Alliance helps communicate the risks of oil and gas and petrochemical development to advance just energy alternatives that protect public health, natural resources, and the climate.
By contributing to FracTracker, you are helping to make tangible changes, such as decreasing the number of oil and gas wells in the US, protecting the public from toxic and radioactive chemicals, and stopping petrochemical expansion into vulnerable communities.
Your donations help fund the sourcing and analysis of new data so that we can keep you informed and continually update our resources.
Please donate to FracTracker today as a way to advocate for clean water, clean air, and healthy communities.
Leave a Reply
Want to join the discussion?Feel free to contribute!