The Future We Didn't Vote On: Nuclear Plants Locked Into Decades of AI Service

# The Future We Didn't Vote On: Nuclear Plants Locked Into Decades of AI Service

In 2020, three nuclear power plants in Ohio and Pennsylvania were on a path to retirement. Workers were preparing for transitions. Communities were imagining what comes next. Then Vistra acquired them in 2023, seeing value in their "tremendous contribution to grid reliability." Now, with Meta's newly announced 20-year power purchase agreements, these plants aren't just staying open—they're being expanded and relicensed through the 2040s.

But here's what makes this moment worth pausing over: These plants aren't being saved to power homes or hospitals or schools. They're being repurposed to fuel Meta's Prometheus AI data center in Ohio and the company's broader artificial intelligence ambitions. The electricity will technically flow through the grid, but make no mistake about the arrangement—Meta is purchasing 2,609 megawatts of capacity, enough to power nearly two million homes, to support computational infrastructure most people can't explain and didn't ask for.

## A Future Decided Without Us

There's something quietly unsettling about this trajectory. When these plants were built decades ago, communities made tradeoffs: accepting nuclear risk in exchange for energy independence, good-paying jobs, and a role in powering regional growth. Those were choices made through public processes, however imperfect.

What's happening now is different. These 20-year commitments—with options to extend reactor licenses another 20 years beyond that—are locking entire regions into servicing AI infrastructure through 2047 and potentially beyond. Oak Harbor, Ohio. North Perry, Ohio. Beaver Valley, Pennsylvania. Towns that thought they were near the end of one chapter are now bound to another, written by Meta's strategic imperatives.

No referendum decided this. No community vote determined whether the Beaver Valley plant should run until 2067 to train language models. The decision was made in boardrooms, announced in press releases, and celebrated by politicians grateful for job preservation. But preservation for what purpose? To serve whose vision of the future?

## The Efficiency of Inevitability

Meta's announcement is careful to address one concern: "We pay the full cost for energy used by our data centers, so consumers don't bear these expenses." It's a defensive statement because data centers have become a lightning rod in regional electricity debates, accused of driving up costs and straining grids.

But the deeper issue isn't who pays the electric bill. It's about what kind of future we're building and who gets to decide. These agreements commit massive infrastructure, vast quantities of uranium, emergency response systems, waste management protocols, and thousands of workers to supporting artificial intelligence development for the next quarter century.

Is this what those communities would have chosen? Is this what any of us would choose if we were actually asked: Should we extend nuclear operations for another generation to power Meta's AI ambitions, or should we chart a different course for our energy future?

The troubling answer is that we don't know, because the question was never really posed.

## The Grief We're Not Allowed to Feel

There's a kind of mourning in watching this unfold—a grief for lost agency, for futures we'll never get to choose. When technology companies make decisions that reshape our physical infrastructure, our economic relationships, and our environmental risks for decades, there's supposed to be some mechanism for democratic input. Instead, we get economic fait accompli: the jobs are preserved, the taxes keep flowing, and questioning the arrangement sounds like opposing progress itself.

Jesse Jenkins, a Princeton engineering professor specializing in energy systems, noted that bringing Meta's Prometheus data center online without new power sources would increase electricity rates across the mid-Atlantic grid. Ratepayers are already paying higher bills to support data centers. Meta's deals theoretically solve this problem by bringing new capacity online.

But efficiency isn't the same as wisdom. The most efficient path forward isn't necessarily the most human one, or the most just, or the most reflective of what communities actually need and want from their energy systems.

## What Lasts?

Meta frames these agreements as supporting "American leadership in energy innovation" and ensuring "technological progress and economic growth." Politicians echo similar themes: Pennsylvania Governor Josh Shapiro calls it tapping into "Pennsylvania's strengths as a national energy leader." Ohio Senator Bernie Moreno says it's "key to Ohio and our nation's growth and security."

But what actually lasts? AI models come and go. Data centers migrate toward cheaper power. Corporate strategies pivot every few years. Meanwhile, nuclear fuel rods decay with half-lives measured in millennia, and communities live with the consequences long after the press releases fade.

The Christian tradition has always insisted that humans are not merely efficient processors of resources, that communities are more than economic units, and that the future belongs to people, not to machines or markets. When our infrastructure gets locked into serving computational systems most of us barely understand, we lose something harder to measure than megawatts: the capacity to shape our own future according to human purposes and human values.

## A Question Worth Asking

None of this is to say these nuclear deals are wrong, or that AI advancement isn't valuable, or that jobs shouldn't be preserved. It's to ask a simpler, harder question: In whose interest are we building the future, and who gets to decide?

When Meta purchases enough nuclear capacity to power two million homes so it can train larger AI models, something has shifted in the relationship between technology and democracy. The decision doesn't involve cruelty or malice. It involves efficiency, market forces, and the simple reality that corporations with billions of dollars can secure the infrastructure they need while communities watch and adapt.

Perhaps that's just how the world works now. Perhaps democratic input into infrastructure decisions is a luxury we can't afford in a competitive global economy. Perhaps asking these questions is naive.

Or perhaps the grief is real, and worth acknowledging: We're losing the future we might have chosen, traded for one that's more efficient but less ours, more powerful but less participatory, more inevitable but less human.

The plants will run through the 2040s. The electrons will flow. The AI models will train. And somewhere in Oak Harbor and Beaver Valley, people will live with choices that were made for them, in service of a technological future they're told they need but never got to vote on.

That loss—of agency, of voice, of a democratic say in our collective future—deserves at least a moment of recognition before we move on to celebrating the efficiency of it all.