Nuclear Energy Renaissance: Powering America's Carbon-Free Future

Migration note

This issue is a faithful port of the 2024 platform's "Nuclear Energy Renaissance" plank, not a fresh policy pass. It hasn't been independently re-verified against current 2026 nuclear policy, reactor licensing activity, or, most importantly given how much this specific plank has moved since 2024, the AI data-center electricity demand story. See the Research tier for what that gap means concretely.

The original argument

The document positions nuclear specifically as a complement to solar, wind, hydro, and geothermal, not a replacement. Its stated case is that nuclear provides "continuous, reliable, carbon-free energy" and grid stability that intermittent renewables alone don't supply, making a diversified portfolio "resilient" in a way any single source wouldn't be. Its regulatory critique targets speed rather than rigor: "outdated regulations and bureaucratic delays continue to stifle nuclear innovation," with the proposed fix being expedited approval processes for advanced designs, not loosened safety requirements. The nonproliferation argument is offered specifically to preempt an anticipated objection: that expanding domestic nuclear power raises weapons-proliferation risk. It does this by drawing a bright technical line between civilian-grade and weapons-grade enrichment.

What's happened since 2024: validated, and more complicated than the original text

The plank's basic bet turned out to be correct, and by a wider margin than the 2024 authors likely expected: the federal ADVANCE Act, the legislative vehicle for exactly the streamlined-approval proposal this plank calls for, passed the Senate 88-2 and the House 393-13, and was signed into law in July 2024. That's about as close to political consensus as any issue in this platform gets. But the near-unanimity conceals something the original plank didn't anticipate: the two "no" votes (Sens. Markey and Sanders) weren't a nuclear-safety objection. They were an environmental-justice one. The House version of the bill stripped $225 million in community-cleanup funding and $100 million earmarked for tribal uranium-contamination cleanup (an ongoing legacy issue on Navajo Nation land) that the Senate version had included. That specific, addressable cut, not opposition to nuclear power itself, is what cost the bill its last two votes, which is why Proposal 4 above restores it rather than treating the ADVANCE Act as a finished success story.

The live disagreement since then has shifted from "should we build" to "how fast, under what oversight." 2025 executive orders pushed to grow US nuclear capacity roughly fourfold by 2050 and rewrote the NRC's regulatory posture, drawing the same environmental/safety objection Markey raised in 2024, now aimed at the pace of deregulation specifically. Waste storage siting is where this plank's original "the waste fits on a football field" argument runs into a problem the volume comparison doesn't solve: the Supreme Court upheld federal authority to site interim storage in Texas and New Mexico in June 2025, but the Republican governor of Texas and the Democratic governor of New Mexico both oppose it in their own states regardless. That's a bipartisan not-in-my-backyard dynamic, not a left-right fight, and no amount of waste-volume statistics resolves it on its own.

The AI-data-center electricity-demand connection flagged as unverified in the prior pass is now confirmed. It's a more complicated story than "more demand strengthens the case for nuclear." That's true on a decade-long timeline; it's not what's happening on the ground right now, and this plank would be dishonest to imply otherwise.

What AI's electricity demand is being met with today, and at what cost

Nuclear capacity takes 10-15 years to bring online even under this plank's own streamlined- approval proposal. AI data-center demand is not waiting. It's being met right now, overwhelmingly, with new or reactivated gas- and diesel-fired generation, because grid-interconnection queues for cleaner power routinely run 4-7 years while a gas turbine can be stood up in 18-24 months. Industry-tracked estimates put the resulting behind-the-meter gas buildout in the tens of gigawatts nationally as of 2026, a directional, analyst-sourced figure rather than a government statistic, but consistent with specific, well-documented individual projects: Meta's Hyperion data-center campus in Louisiana is tied to a proposal for up to ten new gas plants; OpenAI's first Stargate site in Abilene, Texas is running roughly a gigawatt of on-site gas turbines.

The clearest, most legally consequential example is xAI's "Colossus" data center, which spans a site in South Memphis, Tennessee and an adjacent site in Southaven, Mississippi. Independent thermal-imaging investigations and public records show the company operated dozens of gas turbines beyond its permitted count at both sites. As of one 2026 filing, the original 15-turbine plan had grown past 50 at the Mississippi site alone, and the EPA has explicitly rejected the specific "mobile source" legal theory the company and Mississippi's permit board used to justify treating the turbines as exempt from standard permitting. The Memphis site sits adjacent to Boxtown, a nearly all-Black neighborhood founded by formerly enslaved people that was denied basic city water and sewer service into the 1980s and already hosts a refinery, a steel mill, and a former coal plant. The community's documented life expectancy was roughly eight years below the national average before this facility arrived. The NAACP, the Southern Environmental Law Center, and Earthjustice filed a federal Clean Air Act lawsuit in April 2026 and sought an emergency injunction in May, alleging both a straightforward permitting violation and a pattern of environmental racism; the Trump Justice Department intervened in June 2026 to seek dismissal on national-security grounds tied to the same company's military AI contracts. No court had ruled on either the injunction or the dismissal motion as of this writing. This is an active, unresolved legal fight, not a settled outcome, and this plank does not treat it as one.

This isn't only a story about air pollution near one facility. It's also, separately, a documented cost to ordinary electricity ratepayers. PJM Interconnection's own capacity-auction prices rose from under $29 per megawatt-day in the 2024/25 auction to over $329 in the 2026/27 auction, and PJM's own independent, FERC-mandated market monitor, not an advocacy group, attributed a majority of the most recent increase specifically to data-center load growth. Household bills in PJM territory have risen as a measurable result. The degree of exclusive attribution to data centers specifically is contested (PJM and the utilities also point to generator retirements and interconnection delays), but that AI-driven demand growth is a major contributor to price increases hitting ordinary households today is not seriously disputed.