The Fuel Isn’t Going Anywhere

When I was at Lawrence Livermore, the nuclear waste question was already old. Everyone in the building knew the United States hadn’t solved it. I came back to it this week, decades later, expecting to find the answer I’d missed. There isn’t one.

That would be a footnote if nuclear were standing still. It isn’t. The datacenter buildout has turned nuclear into the power source everyone wants, and the federal goal is to roughly quadruple U.S. nuclear generation over the next 25 years. Every one of those reactors makes spent fuel on day one. None of them has a place to send it.

I’m not writing this as a skeptic of nuclear power. I started my career in Navy nuclear operations and I want the expansion to work. That’s exactly why the back end can’t keep riding along as an afterthought.

The number that doesn’t move

Start with the inventory. About 86,000 metric tons of commercial spent fuel sit at 75 sites across the country, mostly at current and former plant locations. The operating fleet adds roughly 2,000 tons a year.

The federal government was contractually obligated to start taking that fuel in 1998. It never did. Yucca Mountain, the only site Congress ever designated, stopped in licensing in 2010 after billions spent, and it’s still the only site federal law allows for permanent disposal.

This year brought real activity. DOE invited states to host “Nuclear Lifecycle Innovation Campuses,” received 28 applications from 26 states, and named five finalists in July: Idaho, Louisiana, Oklahoma, Tennessee and Utah. That’s the headline number, and it’s genuinely encouraging.

Here’s the number that matters more. Tons of commercial spent fuel moved to permanent disposal: zero. States legally committed to host a permanent repository: zero. Amendments to the Nuclear Waste Policy Act signed into law: zero. Applications are a signal. Movement is fuel leaving a reactor site for somewhere it can stay.

How expansion got permission anyway

If there’s no place for the fuel, how is anyone licensing new reactors? The answer is a regulatory decision, not an engineering one.

In 2014 the Nuclear Regulatory Commission adopted what’s now called the Continued Storage Rule. It concluded that spent fuel can be stored safely in pools and dry casks well past a reactor’s licensed life, and indefinitely if it comes to that. Courts upheld it. From that point on, a reactor license no longer depended on a repository existing.

To be fair, the rule rests on a real record. Dry cask storage has operated for decades without a significant release. The physical volume is small. Even all the commercial fuel the country has ever produced would cover something like a football field stacked a few meters high. This is a hard political problem, not an impossible technical one.

There’s also a financial piece most people miss. Because DOE breached its contracts, utilities recover their on-site storage costs through lawsuits paid from the federal Judgment Fund. GAO notes that bill grows every year. The cost of not deciding has been moved off the utility and onto the taxpayer, which removes the pressure on the people building plants to demand a solution.

So the system works the way it was designed to. It just wasn’t designed to end.

The part that did get solved

My memory of the problem included low-level waste, too, and that half is a useful contrast. In the 1980s low-level waste was a genuine crisis, with only a few disposal sites and the threat that they’d close. Congress made states responsible, and while the regional compact system mostly failed to build new sites, a handful of facilities, including Clive, Utah and Andrews, Texas, ended up carrying the load. Routine low-level waste from operating plants has somewhere to go today.

The lesson isn’t that the system was elegant. It’s that a back end works once someone with authority owns it and a host community says yes. The remaining gap on that side, Greater-Than-Class-C waste like activated reactor internals, still lacks a disposal route, which tells you the same thing from the other direction.

Give the new plan its due

The campus approach is the most serious attempt to break the impasse in fifteen years, and it’s smarter than the ones before it.

Previous efforts asked a state to accept the waste and nothing else. The campus model bundles the front end of the fuel cycle (enrichment, fuel fabrication, advanced reactors, even datacenters) with the back end, and lets a state build an industrial base around it. One DOE official described the front-end pieces as the appetizers and waste disposition as the main course. Volunteer states are new. In the early 1990s, when the government last asked, nobody raised a hand.

That’s real progress, and the people working on it deserve credit for it.

Where it still stops short

Read the program carefully and the word is “disposition,” not “disposal.” DOE has said plainly it isn’t asking states to host a deep geologic repository. States can propose interim storage or recycling instead. Reprocessing changes the volume and composition of the waste, but it still leaves high-level waste that needs a permanent home. Interim storage is, by definition, interim.

The pieces that would make it permanent all sit with Congress, and none of them has passed:

• Legal authority. The Nuclear Waste Policy Act still points at Yucca. Until it’s amended, no other site can legally take fuel for permanent disposal. Legal analysts also warn that recent court limits on agency discretion may block even private interim storage without new legislation.
• Durable funding. Roughly $50 billion sits in the Nuclear Waste Fund. The administration wants to restart the fee utilities once paid into it. Others argue the existing money should be unlocked first, with any new fee tied to milestones so it can’t be diverted again.
• An institution built to outlast administrations. GAO’s experts, state regulators and former DOE officials have converged on the same idea: a dedicated organization with its own funding, focused on nothing but this.
• Terms for new reactors. The contracts DOE signs for new plants give utilities and states little assurance the government will ever take that fuel. Expansion is adding liability on the same terms that already failed.

None of this is a secret. GAO, the Nuclear Waste Strategy Coalition, state energy offices and policy analysts have all said versions of it this year. What’s missing isn’t the diagnosis. It’s movement.

Why the datacenter buildout raises the stakes

I’ve written for months that you don’t remove a bottleneck, you relocate it. The AI buildout ran into power, and the industry’s answer increasingly runs through nuclear. That relocates the constraint again, from the grid to the fuel cycle, and the fuel cycle’s hardest rung is the back end.

That back end is the consent rung in its purest form. Every repository effort in U.S. history has died there, under administrations of both parties. Money can buy a reactor, a fuel contract and a transmission line. It can’t buy a community’s permanent yes, which is why the campus plan’s bundling approach is the right instinct.

The new piece is who now sits at the table. The hyperscalers and datacenter developers signing long-term nuclear deals are the demand that justifies the expansion. They have more leverage over the back end than any buyer in the industry’s history, and right now almost none of them are using it. I’ve been asking a related question about whether safety oversight is keeping pace with these deals. The waste question is the same pattern one step further down the fuel cycle.

What has to happen

A few strong voices have made the case. Strong voices without movement leave us where we started. Here’s what movement would look like, and who can push it:

• Congress: Amend the Nuclear Waste Policy Act to authorize consent-based siting of both interim storage and a permanent repository. Unlock the Nuclear Waste Fund with milestone-based controls. Create a dedicated waste organization with direct access to that funding.
• DOE: Make a permanent disposal pathway a defined deliverable of the campus program, not an optional future chapter.
• Datacenter buyers and developers: Put the back end in the term sheet. If you’re underwriting new nuclear capacity, ask for the plan for its fuel and support the legislation that makes one possible.
• Investors and boards: Add one question to nuclear diligence. Who takes this fuel, when, and at whose cost? Price the answer.
• Everyone who works in or near this industry: Tell your representatives this is part of the build, not a separate issue. It’s one of the few nuclear questions that hasn’t split cleanly along party lines, which makes it winnable.

The fuel isn’t going anywhere. The question is whether the policy will.