By 2030, global data centers will consume as much electricity as Japan, according to a recent report by the International Energy Agency. Feeding that new energy demand with fossil fuels will exacerbate the climate change crisis. AI infrastructure will also put increasing strain on the electricity grid and lead to rising energy costs for communities.
Incorporating solar and wind energy in hybrid systems could provide an attractive solution to both problems, but the Trump administration has declared all-out war on renewable energy, taking a range of actions to stymie its development. The administration has instead embraced nuclear—particularly new-generation small modular nuclear reactors (SMRs) situated near data centers—as the optimal means to bolster energy generation for our expanding AI universe. This move has considerable appeal, particularly for those concerned about the climate change crisis and interested in cleaner alternatives to fossil fuels. But the manner and speed with which it is being developed raises serious legal and policy concerns.
These concerns arise from a coordinated loosening of regulatory oversight on three fronts: (1) AI development, (2) nuclear deployment, and (3) the environmental review that would ordinarily govern both. We have written elsewhere about the risks raised by the Trump administration’s deregulation of AI development. Our concern here is with the less-examined legs of this triad: the dismantling of nuclear safety oversight and the gutting of environmental review in order to accelerate AI development. The regulatory architecture now being dismantled is the hard-earned product of lessons learned from earlier experiences in developing and operating technologies that pose high risks to both the public and the environment.
Removing the Regulatory Guardrails
A series of executive orders issued in 2025 and 2026 signaled a dramatic shift in federal policy. On one track, the Trump administration moved to accelerate AI development, framing regulatory oversight as an obstacle to innovation and national competitiveness. On another, it targeted nuclear energy, specifically SMRs and microreactors, as the solution to AI’s insatiable demand for electricity. A third set of actions sought to streamline or bypass environmental review requirements that might slow either effort. Taken together, the message is unmistakable: remove friction, move fast, and let Big Tech build the future.
Source: International Energy Agency via the Pew Research Center.
That future increasingly looks like privately operated nuclear reactors situated next to hyperscale data centers. The logic is seductive. AI systems require enormous amounts of strictly uninterrupted power. Diverting grid capacity from residential and commercial users is actively opposed by residents. Nuclear power is dense and reliable, and although new-generation SMRs still pose risks (including new risks of unproven technology, increased waste, and security for widely distributed units), they are considered far safer than older nuclear technology. Pairing reactors directly with data centers promises efficiency, resilience, and strategic advantage. What is more, it is also a low-carbon solution (although this benefit appears to be a positive byproduct rather than a particular objective for this administration). Consequently, pressure from the corporate sector to implement this plan quickly is huge.
But law and regulation exist to temper such haste and carefully assess and moderate the risks associated with rolling out new technologies. For decades, the federal government promoted nuclear energy but also regulated its power through an independent body, the Nuclear Regulatory Commission (NRC), designed to prioritize safety over speed. That separation was not incidental: Congress created the NRC in 1975 precisely because its predecessor, the Atomic Energy Commission, had combined promotion and regulation in a way that was widely seen as a disqualifying conflict of interest. The Trump administration’s executive orders, which shift significant licensing authority to the Department of Energy (DOE), risk recreating the very conflict the NRC was established to cure. Indeed, the DOE has already reportedly secretly revised a swath of security and safety directives to facilitate the construction of new nuclear reactors.
Source: National Laboratory of the Rockies.
Similarly, environmental laws like the National Environmental Policy Act (NEPA) require agencies to pause, study, and disclose risks before developing new infrastructure and industrial facilities. These regulatory frameworks are the result of hard lessons learned from failures to adequately govern technological and environmental risks in the past—from the nuclear accident at Three Mile Island to the petrochemical pollution of “Cancer Alley” in Louisiana. NEPA’s environmental impact assessments have long been criticized as inordinately slow, and there is certainly an irony in environmental protection slowing down efforts to transition to carbon-free energy. But Congress began to address these issues in 2023 amendments imposing page and time limits. The core purpose of NEPA assessments remains entirely relevant: understand how much the environment will be damaged before the damage is done.
The Trump administration’s nuclear directives, including the recent overhaul of nuclear safety regulations without public notice and comment, impose tighter timelines on licensing, encourage the use of federal land for rapid deployment of nuclear reactors, and open the door to alternative approval pathways outside of the traditional regulatory framework. Its environmental orders, meanwhile, aim to narrow the scope of review, reduce procedural hurdles, and limit the ability of agencies and courts to delay projects deemed strategically important. In February 2025, the White House Council on Environmental Quality went further still, rescinding its NEPA regulations entirely and leaving agencies to draft their own procedures from a non-binding template.
In isolation, each move can be defended as reform. Indeed, given our deep concern regarding climate change, we note the tension here—from a climate change perspective, more rapid transition to carbon-free energy generation is crucial. But these moves, taken together, are troubling. Both the AI and nuclear industries pose the potential for significant societal harms. The regulatory frameworks overseeing them, together with environmental review, should provide mutually reinforcing regimes to govern and moderate those risks, minimizing the likelihood of such harm. The current drive towards deregulation across all three regimes— representing a shift from precaution to heedless acceleration—converts that mutually reinforcing constraint into a risk multiplier.
This is all the more so because of who will benefit from deregulation. The companies leading the AI revolution are not utilities. They are technology firms whose core competencies are speed, scale, and disruption. Their incentive is to build infrastructure as quickly as possible to maintain competitive advantage. When those incentives align with a federal policy that prioritizes rapid deployment, the result is not cautious experimentation. It is industrialization at startup velocity. That may work for software. It is a far riskier proposition for nuclear systems.
While current thinking suggests that SMRs will be developed and operated by separate utilities, AI companies will be their main customer, exercising massive influence over their operations and policies. And both will be positioned to exercise outsized sway over a compliant regulatory agency. A cautionary analogue is Boeing’s 737 MAX. There, gradual regulatory capture that permitted industry self-certification led to catastrophic, fatal outcomes within a supposedly mature safety culture. Nuclear deployment at startup velocity, under a compressed and permissive review process, courts a similar dynamic in a domain with far less margin for error.
Undermining the Current Regulatory Frameworks
The fueling of AI by nuclear power raises questions current frameworks are only beginning to confront. For example, liability regimes for nuclear accidents, centered on the Price-Anderson Act, were designed with centralized plants and regulated utilities in mind—not with reactors owned or operated by technology companies whose corporate structures and capital reserves bear little resemblance to the investor-owned utilities that the Act assumes. Under the existing framework, a DOE-licensed SMR operated by a contractor would be indemnified by the federal government under a separate and less-tested provision of Price-Anderson, while the hyperscaler that leases the land and consumes the power would fall outside the federal nuclear liability regime altogether. This would leave any allocation of loss between federal indemnification, contractor liability, and state tort claims against the tech company to be worked out in litigation after the fact.
Similarly, environmental review processes assume discrete projects, not the current plans for a proliferation of small reactors embedded across a broad network of private facilities. Each individual reactor might receive a narrow site-specific review, or exclusion, but the cumulative environmental impact of the entire network and program—in the form of aggregate effects on water resources, spent fuel storage and transport, thermal discharges, land use, and community exposure—would remain entirely unexamined. This is the kind of systemic risk that programmatic review was designed to capture, and which the Trump administration is undermining.
Co-location compounds these problems. When a reactor sits on private land leased to an AI company, operated by a contractor under partial DOE rather than NRC authority, and is dedicated to a single corporate customer rather than a public energy grid, traditional lines of regulatory supervision, community input, and liability allocation become blurred. If a reactor is co-located with a data center, operated through a partnership between a tech company and a contractor, and approved under expedited federal authority, who is accountable when something goes wrong? The operator? The landowner? The federal government that fast-tracked the project or the agency that approved it? These are precisely the kinds of legal ambiguities that, in other industries, have led to elevated risk, protracted litigation, and regulatory backlash—the U.S. offshore oil and gas sector leading up to and after the Deepwater Horizon disaster exemplifies the problem. In the nuclear context, the stakes are simply higher.
Moreover, administrative law depends on agencies exercising independent judgment, not executing preordained policy outcomes under political pressure to facilitate and approve private action, which appears to be the modus operandi being developed by the Trump administration. The administration’s defenders argue that the risks created by these policies are manageable, and, more importantly, that the greater risk lies in falling behind in the AI race with China. Beijing is investing heavily in both AI and advanced nuclear technology. If the United States clings to outdated and ponderous regulatory models, the argument goes, it will cede leadership in both domains.
There may be some truth in that. But the comparison cuts both ways. China’s nuclear build-out proceeds under an authoritarian regulatory state whose speed depends on the absence of public participation, independent regulatory oversight, and possible judicial review of agency decisions that define U.S. administrative law. Competing with China does not require an abandonment of the core features of the American democratic administrative state, any more than competition with the Soviet Union did. Indeed, excessive deregulation in the interest of speed can be counterproductive, as it may result in backlash and litigation that will ultimately slow development over the longer term. Some regulatory reforms—to expedite environmental assessment, for instance—are certainly required, but wholesale rejection of the precautionary structure of the governing regulatory regimes, all in the name of national security and competition, is not.
Federal deregulation, moreover, does not operate in a vacuum. SMR siting will still trigger state public utility commission proceedings, state water quality certification under Section 401 of the Clean Water Act, and local zoning review. But these processes were not designed to assess nuclear safety or the adequacy of liability frameworks. They cannot substitute for the federal oversight being dismantled. But they will likely generate legal challenges and delays that the compressed federal process was supposed to eliminate, creating pressure to override state and local authority as well. The risk is of a deregulatory ratchet: each layer of independent review that fails to deliver the promised speed becomes the next target for removal.
The United States has navigated transformative technologies before. Railroads, aviation, telecommunications, early nuclear energy, nuclear weapons, and the internet all required new regulatory frameworks. Each of these challenges required balancing the benefits of innovation with public protection. The most successful approaches did not eliminate oversight; they modernized it while preserving its independence and core purpose of minimizing risk. What is unfolding now threatens to upend that balance altogether. By simultaneously accelerating AI development, deregulating nuclear deployment, and weakening environmental review, the federal government is not merely updating the rules. It is rewriting the relationship between innovation, risk, and accountability.
Speed Without Safeguards
The United States was the first to harness two of the greatest technologies of the last century: nuclear power and AI. The successful convergence of AI and nuclear power could indeed define the next era of U.S. strength. Nuclear power could provide the energy behind a new, AI-driven technological revolution that reinforces U.S. leadership in a competitive world. But power, whether computational or atomic, has always demanded restraint.
The United States may need to move quickly to maintain its dominance in the global AI race, but there is significant risk that it may be discarding the legal architecture that made the use of nuclear energy sustainable in the first place. We acknowledge that the deployment of carbon-free energy, including next-generation SMRs, is essential to help fuel the AI expansion, and that this may also require expedited environmental review. However, the risks demand the mutually reinforcing governance of established independent regulatory regimes grounded in the rule of law. Congress and the courts must act to restore the independence and integrity of the three interlocking regulatory frameworks before the reactors are in place, not after the first accidents demand it.
