In recent years, the United States has been experiencing a genuine renewed interest in nuclear energy.
Utilities restarting long-closed plants, a Trump administration determined to push for the construction of new reactors, dozens of startups developing innovative technologies, and large tech companies already ready to sign contracts for the energy that will be produced one day: these are the elements of a rapidly evolving scenario described in a new report by Bloomberg.
The transformation of the American power grid will not be immediate, however. Nuclear development requires very long times, enormous investments, and careful management of risks and complexities.
Yet the current conjuncture, marked especially by the explosion of energy consumption linked to AI, is creating the conditions for what many call a “nuclear renaissance.”
Changing the scenario
For decades, nuclear power has been a cornerstone of US electricity production, providing about one-fifth of the country’s total energy needs.
Currently, 94 nuclear plants are operational across 28 states, the vast majority of which came online between the 1970s and 1980s.
After the shale gas revolution and a long period of stagnant demand, enthusiasm for new construction had almost completely faded. The few projects launched, such as units 3 and 4 of Vogtle in Georgia, accumulated massive delays and more than doubled costs, casting a shadow over the entire sector.
Today, however, the context has radically changed. The rapid growth of AI data centers has caused energy demand to skyrocket, making nuclear power attractive again.
According to estimates by the Electric Power Research Institute, in a strong expansion scenario, data centers could consume up to 17% of national electricity by 2030, compared to the current 5%. This is a constant demand, 24 hours a day, which perfectly matches the characteristics of nuclear power.
The limits of other sources
Alternatives show weaknesses. Turbines for new gas plants have multi-year delivery times and rising costs. Coal struggles to compete economically and appears a risky investment in light of possible future shifts in environmental policies. Solar and wind suffer from intermittency and require costly battery storage systems to ensure continuity.
In this context, American public opinion has decidedly softened towards nuclear power: according to a Pew Research Center survey, 59% of adults now support greater use of nuclear energy, compared to 43% just ten years ago. Several states are also reconsidering their historic bans, a sign of a profound cultural change.
Restarting the past
Faced with the long timelines of new constructions, many utilities have chosen a more pragmatic path: bringing existing plants back to life.
Holtec International aims to restart the Palisades plant in Michigan by the end of the year, thanks to support from state and federal funds. At Three Mile Island in Pennsylvania, Constellation Energy targets 2027 for the restart of a reactor after signing a twenty-year agreement with Microsoft. Similarly, NextEra Energy plans to reactivate the Duane Arnold plant in Iowa by 2029 thanks to a 25-year contract with Google.
These projects demonstrate that, when possible, reactivating existing plants is faster, less expensive, and less risky than building new ones from scratch.
The ambitions of the Trump administration
The Trump administration has set very ambitious goals: aiming to quadruple national nuclear capacity to 400 GW by 2040 and to start construction of at least ten large reactors by the end of the decade.
Over $80 billion has been allocated to support Westinghouse technology, with the goal of replicating standardized projects and achieving economies of scale.
On the bureaucratic front, efforts are underway to speed up processes: an executive order requires the Nuclear Regulatory Commission to evaluate licenses within 18 months.
However, this deregulatory push raises legitimate safety concerns, especially after indications to review radiation exposure limits.
The bet on Small Modular Reactors
The real innovation comes from small modular reactors (SMRs).
These reactors, more compact and assembled on-site, promise greater safety, cost-effectiveness, and faster construction compared to traditional plants. Many adopt passive safety systems based on gravity and coolants alternative to water.
Several startups including Oklo, TerraPower, NuScale, and others are making concrete progress, with some reactors already reaching criticality in the Department of Energy’s pilot program. Large tech companies like Meta have already signed agreements to purchase energy from these projects.
Nevertheless, the first commercial plants likely will not be operational before the early 2030s, and initial costs will remain high. Success will depend on the ability to truly industrialize production.
Nuclear fusion
Alongside traditional fission, big tech is also investing in fusion, which in theory would offer clean energy without long-lived radioactive waste.
The United States is attracting billions in private capital in this field. Google, for example, has invested in Commonwealth Fusion Systems and committed to purchasing energy from its first commercial plant.
However, timelines remain long and uncertain: it will take at least another ten years before truly competitive fusion plants are seen.
The Achilles’ heel
A often underestimated problem concerns uranium. The United States heavily depends on imports of enriched material, with Russia until recently covering a significant share.
The ban introduced after the invasion of Ukraine has made the construction of a domestic supply chain urgent, both for traditional uranium and for HALEU required by many SMRs.
The Trump administration has allocated significant funds for this goal, but the path to independence is still long and will require intensive capital use.
