Reading · Compound Impact

As public opposition against data center developments grows, we are seeing residents take matters into their own hands.

The Ohio Attorney General certified a public petition proposing a state constitutional amendment that would permanently ban new data center construction in the state. The petition needs more than 400,000 signatures by July 1 for the measure to be on the ballot this November. It is a rare move, and one of the few efforts I have seen to try to make opposition to data center development state law. At least 11 other states have introduced legislation this session for temporary pauses, a more common and more cautious posture. Local communities that have the authority to do so are imposing moratoriums — 54 and counting.1

Denver’s City Council recently passed a yearlong moratorium, prompted not only by grid stress but by drought conditions. Denver Water called for a 20% reduction in water use by city residents. Water scarcity is now joining grid stress as a trigger for communities to push back. And while there is growing evidence that more efficient chip manufacturing and improved cooling systems will reduce water demand, efficiency gains are years away. The demand on water infrastructure in water-stressed and drought-prone communities is happening now. These communities are right to pause.

Yet, even in these communities, a pause is only as useful as what happens inside it. We need more transparency in development projects and we also need to take a closer look at the human cost of energy insecurity that data centers might only make worse, if left unaddressed.

Before the data center boom, energy insecurity was already rising. The 2024 Residential Energy Consumption Survey, released by the U.S. Energy Information Administration this month, found that 32.9% of U.S. households — about 43.56 million — reported at least one form of energy insecurity in 2024.2 That is up from 27.2% in 2020, a 5.7 percentage point increase. The National Consumer Law Center notes something important about the timing: this increase predates the recent sharp run-up in energy and utility prices.3 What the data already shows will almost certainly worsen with hyperscaler AI data centers.

Energy insecurity is not an isolated condition in low-and-moderate-income communities. It takes multiple forms — some visible, some quietly devastating — and the distribution is not random. The experiences are most acute in communities of color, at rates roughly double the national average.4

Utility analyses shared in last week’s post, Where the Grid Fails First, underscore the structural, not merely statistical, impact of data center siting decisions exacerbating socioeconomic and racial disparities.

Protecting ratepayers from rising electricity costs has been a consistent thread in this work and in public dialogue this year. But now the data tells us a different, more nuanced, story about who bears the burden of infrastructure costs. Policymakers on both sides of the aisle acknowledge that the affordability crisis has deepened in this country, and if we take time to understand energy insecurity more deeply, we must tell a fuller story.

Energy insecurity is a public health crisis. It is carried disproportionately by working families with the least capacity to absorb rising costs. The research literature links chronic energy insecurity to anxiety and depression — a burden that falls on families who are not responsible for the structural conditions that produced it.5

Data centers alone did not create energy insecurity. The causes are long-standing: systemic racial and economic inequality linked to redlining, substandard housing stock, inadequate utility consumer protections, chronically underfunded assistance programs. And AI data centers are arriving at a moment when the grid is already strained, with utilities seemingly choosing to bifurcate the system to support new loads. This is a market failure. Energy costs are rising due to failed public policies and powerful private sector interests seeking economic gains.

For a moratorium to be meaningful, we must grapple with the pre-existing conditions of our energy system. Pausing a data center project does not reduce a family’s energy burden. The issues that data centers exacerbate have long existed. And while AI adoption remains low among individuals, enterprise and government adoption is growing rapidly. These dynamics are converging at a moment when affordability is front and center.

Absent movement in Congress, states and utilities are deciding for themselves how to respond.

In Colorado, Xcel Energy filed with state regulators to create a separate rate class for large-load customers, preventing data center infrastructure costs from being distributed to residential ratepayers. It is worth noting this is a structural protection that may be achieved through other means. It is not a moratorium, and it is not legislation. The most durable protections emerging right now are being built through rate design. That is a signal worth tracking.

West Virginia is the only state to have passed a law explicitly barring local governments from blocking data center development. Moratoriums are not an option for local communities in the state, so advocates are now pushing for a state-level moratorium instead. Citizens of West Virginia are now engaged in a longer, more expensive campaign to make their voices heard. Community opposition is putting pressure on policymakers statewide. Preemption does not quiet resistance; it displaces it. It just makes the fight longer and harder for the parties with less capacity to sustain one.

Enterprise AI adoption is accelerating — and the companies leading the infrastructure buildout are simultaneously completing massive layoffs. Oracle recently eliminated tens of thousands of positions, a move driven less by operational efficiency than by a shift in capital allocation toward AI infrastructure. Read: more data centers. The race to build is funded, in part, by the people no longer on payroll. The communities being asked to host that infrastructure are often the same communities absorbing those job losses. Economic uncertainty does not reduce opposition to unwanted development — it deepens it. Without stronger policy responses and safety nets that match the scale of displacement, the conditions for opposition are not going away.

There’s another side of the story: governments are taking in AI infrastructure while simultaneously deploying AI systems to support public services. A new analysis finds that state and local governments are accelerating the adoption of AI tools in public services — procurement, workforce systems, and citizen-facing applications.6 New York City adopted a comprehensive plan for AI use in government support services like NYC311 back in 2023. And for a few years now, Los Angeles has integrated AI to help ease traffic, with improvements in travel times and air quality. If communities are bound to welcome data center infrastructure and AI-assisted public services, without fully experiencing the benefits, the capacity gap multiplies and trust erodes.

Moratoriums are meant to be temporary. Pausing development creates some space for negotiation. They are a tool that can give communities time to assess the environmental, health, and community impact of facilities drawing heavily on local land, water, and energy. Moratoriums could also create a window of opportunity for institutional capacity building to happen and for policymakers to engage in meaningful conversations with their constituents about how AI adoption and infrastructure might improve their lives.

Data centers are unpopular right now. Public opposition is not limited to the communities where projects land. A recent Emerson poll found that only 34% of Pennsylvanians support data centers being built in or near their communities — a majority opposition that has not translated into majority protection.7 Pennsylvania is one of many states where policymakers are attracting data center projects, while the economic, environmental, and public health impacts remain difficult to assess because the terms are opaque.

To be effective, a moratorium period should shift the focus from meeting the needs of the private sector to addressing the real, documented human needs of residents, taxpayers, and utility customers.

While useful to pause, a moratorium is not a plan. It is time to build one.

What Happens Inside the Pause?

Comments