An Analysis of Small Modular Reactors (SMRs) for Commercial Electricity Generation in the United States

Introduction 

The case for small modular reactors has shifted substantially since 2024. What was once a policy-driven conversation about decarbonization has become a commercially driven race to secure reliable, carbon-free baseload power for artificial intelligence infrastructure. With AI model training and inference demanding uninterrupted electricity at scales that intermittent renewables cannot guarantee, the world’s largest technology companies have made historic commitments to nuclear power. 

Google and Kairos Power signed the first U.S. corporate agreement to develop a fleet of SMRs, covering up to 500 MW across six to seven reactors with the first targeted for 2030 [2]. Amazon led a $500 million financing round for X-energy and announced a partnership with Energy Northwest to deploy up to 12 of X-energy's Xe-100 SMRs in Washington state [3]. Microsoft signed a landmark 20-year, $1.6 billion agreement with Constellation Energy to restart the Three Mile Island reactor in Pennsylvania, widely described as the largest corporate nuclear deal in history [4]. Meta issued a request for proposals targeting 1 to 4 GW of new nuclear generation for the early 2030s and entered an agreement with TerraPower for up to eight advanced Natrium nuclear plants [5]. Oracle announced plans for a gigawatt-scale data center powered by three SMRs, with CEO Larry Ellison stating that building permits were already secured [2]. 

The International Energy Agency (IEA) projects global data center electricity consumption will reach 1,100 TWh in 2026, equivalent to Japan's entire national electricity consumption, an 18% upward revision from December 2025 estimates [6]. Hyperscalers are expected to spend nearly $700 billion on data center projects in 2026 alone [6]. Against this backdrop, 2026 is shaping up to be a watershed year for U.S. SMR regulation, with the Nuclear Regulatory Commission (NRC) expected to issue licensing decisions on the first two commercial SMR construction permits, and the first experimental advanced reactor poised to achieve criticality at Idaho National Laboratory. 

Power Needs of AI Data Centers in the U.S. 

The energy demands of U.S. data centers are approximately three times higher than current capacity, driven by cloud migration, digitalization, high data usage, and the rapid scaling of AI [7]. According to McKinsey research, between 2024 and 2030, electricity demand for U.S. data centers is expected to increase by roughly 400 terawatt-hours at a compound annual growth rate of approximately 23% [7]. 

Recent figures underscore urgency. The IEA has revised its 2026 data center consumption estimate upward to 1,100 TWh, an 18% increase over its December 2025 projection [6]. Microsoft’s electricity demand for AI data centers alone is projected to surge over 600% by 2030 [8]. These projections have materially altered the economics and political urgency of SMR deployment: where governments once had to incentivize nuclear investment, technology companies are now actively competing to secure offtake agreements with SMR developers. Industry analyst UBS estimates that global company spending on AI infrastructure will reach $480 billion in 2026 [9]. 

Characteristics of Small Modular Reactors (SMRs) 

SMRs are defined as nuclear reactors with an electrical output generally below 300 MWe, designed with modular technology, factory fabrication, and serial production [10]. As of early 2026, over 127 modular reactor designs have been identified globally, with seven designs already operating or under construction, 51 in pre-licensing or licensing processes across 15 countries, and dozens more in early development [11]. 

Main advantages include: (a) small size enabling standardization of major components on assembly lines, reducing cost and simplifying deployment; (b) passive safety design with below-ground deployment and integration of major systems into one unit; and (c) a wider market, as SMRs can satisfy incremental demand without the enormous capital commitments of gigawatt-scale plants [12]. SMR technology has also advanced beyond the light-water reactor (LWR) designs that dominated early discussions. Active designs now include sodium-cooled fast reactors (TerraPower’s Natrium), molten fluoride salt-cooled reactors (Kairos Power), gas-cooled reactors (X-energy's Xe-100), and graphite-moderated sodium-cooled extra-modular reactors (Aalo Atomics’ Aalo-1), each offering distinct advantages in fuel efficiency, operational flexibility, and waste characteristics [13]. 

The United States is considered to be in a second nuclear era, enabled by: (1) growing electricity demand and declining generation capacity margins; (2) concern over large-scale fossil fuel combustion exacerbating global warming; and (3) awareness of energy supply’s impact on national and energy security [14]. SMRs are positioned to complement larger base-load plants and to serve locations unsuitable for utility-scale generation. 

Key challenges remain. Building the massive assembly lines needed for serial production requires substantial upfront capital and large order books, creating a startup problem [12]. SMR designs also raise issues with waste retrieval, reactor repair, and decommissioning. The industry’s rush to market, driven by the urgent need to address climate change, risks insufficient testing and may lead to cost overruns or cancellations, as demonstrated by NuScale’s cancelled Idaho project in late 2023 [12]. 

Policy Support and Regulatory Frameworks 

The regulatory environment for SMRs in the United States has evolved rapidly and accelerated meaningfully in 2025 and 2026. The NRC is expected to issue licensing decisions on the first two commercial SMR construction permits during 2026, a major milestone for the industry [5]. 

NuScale Power remains the only SMR developer to have received NRC design approval, initially for its 50 MWe module in 2023, and subsequently for its uprated 77 MWe design in May 2025, a milestone reached ahead of schedule [15]. NuScale’s commercialization partner ENTRA1 reached a nonbinding collaborative agreement with the Tennessee Valley Authority (TVA) to deploy up to 6 GW of NuScale capacity across TVA’s seven-state service region [15]. 

TerraPower’s Natrium reactor, a liquid-sodium-cooled design capable of supplying up to 500 MWe and backed by Bill Gates, has begun non-nuclear construction at a retiring coal plant in Kemmerer, Wyoming. The NRC completed its environmental review in October 2025 and issued a final safety evaluation in December 2025; a construction permit decision is expected in the first half of 2026. Meta has entered an agreement with TerraPower for up to eight Natrium plants, and NVIDIA’s investment arm NVentures has also backed the company [5]. 

In December 2025, the U.S. Department of Energy made one of its most significant SMR investments to date, selecting TVA and Holtec Government Services to each receive $400 million in federal cost-shared funding, totaling $800 million, to support the early deployment of advanced light-water SMRs in Tennessee and Michigan. TVA will advance deployment of a GE Vernova Hitachi BWRX-300 at its Clinch River Nuclear site in Oak Ridge, Tennessee, targeting operations in the early 2030s. Holtec plans to deploy two SMR-300 reactors at the Palisades Nuclear Generating Station site in Michigan, pursuing an innovative one-stop-shop model covering construction, operation, and electricity sales in partnership with Hyundai Engineering & Construction [16]. 

One of the most striking near-term milestones is taking shape at Idaho National Laboratory. In March 2026, Aalo Atomics unveiled its completed Critical Test Reactor, a 10 MWe sodium-cooled experimental reactor, constructed in just 40 days at INL. Aalo expects the reactor to achieve criticality well before the July 4, 2026 deadline set under President Trump’s Nuclear Reactor Pilot Program. Eleven advanced reactor projects were selected by the DOE under this program with the goal of at least three achieving criticality by Independence Day 2026. Aalo is also planning an experimental data center adjacent to the reactor, which would mark the first time a nuclear plant and a data center have been co-located, offering a concrete model for the AI-nuclear convergence [17]. 

At the federal policy level, the White House issued nuclear-focused executive orders designed to accelerate domestic deployment, including the Reforming Nuclear Reactor Testing at the Department of Energy executive order (EO 14301) signed in May 2025. The DOE issued a Categorical Exclusion finding confirming no significant environmental impacts from the construction, operation, and decommissioning of DOE-regulated advanced nuclear reactors, removing a key procedural barrier [5]. The DOE’s Office of Nuclear Energy continues to support SMR commercialization through its GAIN program, the NRIC, and Funding Opportunity Announcements for advanced SMRs [1]. 

Comparison with Other Energy Sources 

The key differentiator of SMRs relative to other low-carbon energy sources remains in their capacity for continuous, reliable generation of electricity, addressing the fundamental limitation of intermittent solar and wind power [18]. For AI data centers requiring uninterrupted, 24/7 power, this characteristic is not merely an advantage; it has become a prerequisite for large-scale nuclear offtake agreements. Industry analysis suggests nuclear energy could meet up to 10% of data center electricity demand by 2035 [19]. 

Solar and wind retain significant advantages in speed of deployment and scalability, with construction timelines of months to a few years compared to nuclear’s multi-year construction cycle and substantial upfront investment [18]. The levelized cost of energy (LCOE) of solar and wind continues to fall rapidly, making them cost-competitive for most grid applications. However, the inability to guarantee baseload capacity, particularly during low-wind or low-sunlight periods, limits their suitability as primary power sources for energy-intensive AI workloads. 

Natural Gas Combined Cycle (NGCC) plants remain the dominant low-cost, reliable generation technology in the U.S. However, their carbon footprint conflicts with technology companies’ net-zero commitments and with federal and state decarbonization targets. A carbon tax or equivalent carbon pricing mechanism would substantially improve the relative economics of SMRs versus fossil fuel alternatives [20]. Without carbon pricing, SMRs are not currently cost-competitive with gas in liberalized markets, though this gap narrows significantly as SMR projects move from first-of-a-kind to nth-of-a-kind production [1]. 

The combined system value of SMRs with renewables and storage is increasingly recognized. A key challenge associated with high renewable penetration, particularly wind, is intermittency, which introduces volatility and increases integration costs. While storage can mitigate short-term fluctuations, it remains costly at scale. In this context, SMRs provide a complementary source of stable, dispatchable, and carbon-free power. Baker et al. (2018) model an SMR–wind–battery hybrid system and show that its flexibility helps stabilize the levelized cost of electricity (LCOE) as wind penetration increases, thereby enabling higher shares of variable renewables on the grid [28]. As the value of grid flexibility continues to rise, SMRs’ role in providing reliable, on-demand clean power is likely to become increasingly important. 

Structure and Advantages of SMRs 

SMRs’ design features on-site assembly of pre-fabricated modules, with reactor sizes enabling transportation by truck or rail. The integral configuration, placing primary system components inside the reactor vessel and eliminating external piping, increases vessel size while reducing overall plant containment size, creating positive impacts on both safety and economics [21]. 

Multiple advanced designs are now under concurrent development in the U.S., a significant change from the earlier period when NuScale’s LWR design dominated. Kairos Power’s fluoride salt-cooled design uses TRISO fuel (tristructural isotropic pebble fuel), which can withstand extremely high temperatures without melting, providing an inherent safety advantage. TerraPower’s Natrium reactor uniquely combines a sodium-cooled fast reactor with a molten salt energy storage system, allowing it to dispatch up to 500 MWe for several hours on demand, effectively functioning as a dispatchable asset in ways that most other clean energy sources cannot [5]. 

Aalo Atomics represents a new category of developer in this space. Its Aalo-1 extra-modular reactor (XMR) is designed to be manufactured in a factory in Austin, Texas, and assembled at a deployment site in a matter of weeks. The Aalo Pod concept groups five 10 MW factory-built modules around a single turbine to deliver 50 MWe per pod, targeting AI data center applications. Critically, Aalo uses standard 5% low-enriched uranium (LEU) fuel rather than high-assay low-enriched uranium (HALEU), avoiding the supply chain bottlenecks that face many competing advanced designs [17]. 

NuScale’s uprated 77 MWe Power Module remains the only design with NRC Standard Design Approval. Its VOYGR plant design scales to six modules for approximately 462 MWe. The passive safety systems rely on natural convection, gravity, and conduction for cooling, eliminating dependence on active emergency cooling systems and reducing both operational complexity and safety risk [15]. 

Economic Analysis of SMRs 

Traditional economies of scale, where larger plants produce cheaper electricity per unit, do not transfer directly to SMR investment analysis. SMRs are fundamentally different from large reactors in design, modularity, and deployment model [1]. Where traditional economies of scale benefit large plants, SMRs benefit from economies of multiples (efficiencies gained from operating multiple co-sited units) and the learning effect, whereby successive units become progressively cheaper to build and operate [21]. 

A Nth-of-a-Kind (NOAK) SMR unit could provide a levelized cost of energy roughly half that of a First-of-a-Kind (FOAK) unit, making it broadly comparable to a Natural Gas Combined Cycle plant [1]. Achieving this cost trajectory requires sustained order volumes and a mature supply chain. Big Tech offtake agreements are beginning to provide exactly this: X-energy's $700 million funding round in 2025 and Amazon’s $500 million investment demonstrates the scale of private capital now flowing into the sector [22]. 

Boarin et al. (2012) evaluated two business cases for large reactors versus SMRs under "merchant" (market-rate financing) and "supported" (government-subsidized financing) scenarios. Under liberalized electricity markets with high financing costs, SMRs achieve lower production costs than large reactors. However, simple reliance on government subsidies makes SMRs less attractive relative to large reactors, underscoring the importance of market-based demand [1]. The entry of technology companies as guaranteed offtakers represents a structural shift that the original economic models did not anticipate: long-term power purchase agreements reduce financing risk, lower the cost of capital, and bring SMR economics substantially closer to the merchant scenario. 

The key economic risk remains the FOAK problem. NuScale’s cancellation of its Idaho project in 2023, driven by rising cost estimates and insufficient customer commitments, illustrates how vulnerable the first commercial deployments are to shift economics. The U.S. government’s willingness to absorb FOAK risk through loan guarantees and cost-sharing arrangements, such as the $1 billion DOE loan for Three Mile Island’s restart and the $800 million cost-shared funding for TVA and Holtec, is therefore critical to unlocking the NOAK cost curve [16]. 

Public-Private Partnership 

The U.S. public-private model for SMR deployment has evolved substantially beyond initial cost-sharing arrangements. The DOE’s original approach focused primarily on sharing costs to support NRC design certification. More recent partnerships involve technical support, testing infrastructure, demonstration reactors, and federal deployment commitments that serve simultaneously as market signals and safety demonstrations [1]. 

The December 2025 DOE awards of $800 million to TVA and Holtec exemplify this evolved model. Rather than simple subsidies, the awards are structured as cost-shared funding tied to deployment milestones, fleet-scale replication plans, and domestic supply chain development. Holtec’s approach illustrates the depth of public-private integration: the company fulfills the roles of technology vendor, supply chain manager, constructor, plant operator, and electricity merchant within a single project, partnering with Hyundai Engineering & Construction for construction execution [16]. 

The DOE Nuclear Reactor Pilot Program, authorized by executive order in May 2025, takes the model further still. Rather than capital, the program provides selected developers with a DOE concierge team to eliminate bureaucratic friction and an authorization pathway that allows testing on DOE land without NRC licensing. Aalo Atomics used this framework to go from ground-breaking to a completed reactor building in under six months, a pace that would have been unimaginable under traditional nuclear development timelines [17]. 

The structure of corporate offtake agreements with technology companies introduces a further dimension to U.S. public-private partnership. Unlike traditional utilities, hyperscalers bring long-term, creditworthy demand commitments, large upfront capital contributions, and strong incentives for on-time delivery. This changes the risk profile of SMR projects: with a guaranteed buyer, project completion risk falls, making government loan guarantees and co-investment more attractive. 

Internationally, collaboration under the Generation-IV International Forum (GIF) continues, with 13 countries co-developing six advanced reactor concepts [1]. In March 2026, the European Commission published an SMR strategy projecting 17 to 53 GW of EU SMR capacity by 2050, backed by the European Industrial Alliance on SMRs and a target to launch first projects by the early 2030s [23]. The UK also advanced its SMR program in March 2026, with the Secretary of State providing regulatory justification for the Rolls-Royce SMR design, the first such approval for a small modular reactor in the country, and appointing Arup to provide engineering support for the first SMR at Wylfa in Wales [24]. 

Obstacles of SMR Deployment 

Despite the improved momentum, substantial obstacles to SMR deployment remain. Climate vulnerability is an underappreciated risk: the number of climate-linked nuclear power outages has risen from approximately 0.2 per reactor-year in the 1990s to 1.5 per reactor-year in the 2010s, driven by extreme weather events, elevated ambient temperatures, and algae growth in cooling systems [25]. As climate change intensifies, siting and cooling requirements for SMRs must incorporate these long-term risks. 

Cost and schedule overruns remain the industry’s greatest vulnerability. The failure of large-scale nuclear renaissance projects, such as the Vogtle expansion in Georgia, cast a long shadow over nuclear economics, and investors remain cautious. The NuScale Idaho cancellation in 2023 demonstrated that even SMR projects with regulatory approval can collapse due to commercial factors. Managing FOAK execution risk through stronger project oversight, fixed-price contracting, and modular construction discipline will be essential to maintaining investor confidence. 

Path dependency and carbon lock-in continue to constrain deployment. Investment in existing gas and coal infrastructure, established regulatory frameworks, and workforce skills all favor incumbents over new nuclear entrants. Public opposition, often shaped by high-profile historical accidents and reinforced by negative media coverage, remains a significant barrier despite SMRs’ substantially improved safety profiles [25]. 

Fuel supply chains present a critical near-term constraint. Many advanced SMR designs, including those from TerraPower and Kairos, require High-Assay Low-Enriched Uranium (HALEU), a fuel type currently produced in limited quantities and requiring new or modified transport, storage, and regulatory infrastructure. Establishing a domestic HALEU supply chain will require significant government investment and coordination [10]. Aalo Atomics explicitly designed its reactor around standard 5% LEU to avoid this constraint, an approach that may influence other developers. Uranium prices have risen sharply and remain elevated relative to historical norms, adding to operating cost uncertainty across all designs. 

Nuclear waste management remains a long-term challenge. Comprehensive frameworks for spent fuel from advanced SMR designs are not yet finalized, and a March 2026 study identified significant backend nuclear fuel cycle gaps across 141 SMR designs currently under development [26]. Addressing these issues early in the design phase, rather than repeating the trial-and-error approach of previous decades, is critical to long-term sustainability. 

Limitations of Approach and Analysis 

Since SMRs are recent technological developments, comprehensive data on ongoing projects remain limited, and commercially sensitive information is proprietary. The analysis in this paper involves projections relying on assumptions about future market conditions that are highly uncertain. The FOAK cost trajectory for advanced designs such as molten salt and sodium-cooled reactors is particularly uncertain, with limited empirical data. There may also be a time lag in data used, as rapidly evolving regulatory and commercial conditions mean some information may be superseded shortly after publication. 

Future research should examine hybrid system efficiency integrating SMRs with renewables and storage; lifecycle assessments of advanced SMR designs; the geopolitical implications of HALEU supply chain development; comparative analysis of SMR regulatory timelines across jurisdictions; the novel governance questions raised by co-located nuclear reactors and AI data centers on federal land; and evaluation of innovative financing mechanisms such as green bonds and long-term corporate power purchase agreements for SMR projects. 

Conclusion 

The landscape for small modular reactor deployment in the United States has been transformed between 2024 and 2026. The convergence of surging AI-driven electricity demand, over 10 GW in Big Tech nuclear commitments, $800 million in new federal cost-shared funding for TVA and Holtec, NRC regulatory progress toward the first commercial SMR construction permits, and a completed experimental reactor at Idaho National Laboratory has created conditions without precedent in the history of civil nuclear power. 

Public-private partnership remains the cornerstone of successful SMR commercialization, but its character has evolved from cost-sharing toward a richer model encompassing demand-side anchoring by corporate offtakers, streamlined testing frameworks on federal land, and co-located demonstration projects that directly serve AI infrastructure. NuScale’s TVA agreement, TerraPower’s NRC progress, Kairos Power’s Google partnership, Amazon’s X-energy investment, the DOE’s awards to TVA and Holtec, and Aalo Atomics’ imminent reactor criticality at INL collectively signal a transition from proof-of-concept to active commercialization. 

The early mover advantage in SMRs carries real strategic and economic weight. Domestic SMR competitiveness strengthens the U.S. position in the global nuclear value chain, creates high-quality employment, and reinforces energy security at a moment when AI infrastructure has become a core dimension of national competitiveness. Achieving net zero by 2050 will require nuclear power alongside rapid renewable deployment, and SMRs are uniquely positioned to supply the dispatchable, carbon-free baseload that AI-era electricity demand requires. 

The critical challenge is now execution. Cost discipline, supply chain development, regulatory efficiency without compromising safety standards, fuel availability, and waste management frameworks must all be addressed concurrently. If the momentum of 2025 and 2026 can be sustained through construction and into commercial operation, the United States may be entering a new chapter for nuclear energy, one supported not only by government policy, but increasingly by the demands of the digital economy. 

References 

[1] Mays, G. (2020). Small Modular Reactors (SMRs): The Case of the United States of America. Handbook of Small Modular Nuclear Reactors (Second Edition). ScienceDirect. 

[2] Introl (2026, January 8). Nuclear power for AI: inside the data center energy deals. introl.com. 

[3] IEEE Spectrum (2024, December 12). Big Tech Embraces Nuclear Power to Fuel AI and Data Centers. 

[4] Yahoo Finance (2025, December 19). 3 Nuclear Power Stocks Set to Flourish in 2026 on AI Data Center Boom. 

[5] American Bar Association (2026, March). Pressure to succeed: Small modular (nuclear) reactor approvals on the horizon? americanbar.org. 

[6] Tech Insider (2026, April 2). AI Data Centers Now Use More Power Than 30 Countries. 

[7] Green, A., Tai, H., Noffsinger, J., & Sachdeva, P. (2024, September). How data centers and the Energy Sector Can Sate AI's hunger for power. McKinsey & Company. 

[8] TradingView (2025, December). Looking ahead to 2026: why hyperscalers can't slow spending without losing the AI war. 

[9] Jama Software (2025, October). Tech Giants Turn to Nuclear Power for AI's Energy Appetite. 

[10] World Nuclear Association (2026, March). Small Modular Reactors. world-nuclear.org. 

[11] Canary Media (2025, July). Small modular reactors are gaining steam globally. Will any get built? 

[12] Cooper, M. (2014). Small modular reactors and the future of nuclear power in the United States. Energy Research & Social Science. 

[13] Nuclear Engineering International (2026, March). Aalo unveils critical test reactor. neimagazine.com. 

[14] Ingersoll, D.T. (2012). Deliberating on the deployment of SMRs. Nuclear Technology, 165(1), 94-101. 

[15] NuScale Power (2025, June 3). NuScale Power's SMR Achieves Standard Design Approval from U.S. NRC for 77 MWe. nuscalepower.com. 

[16] U.S. Department of Energy (2025, December 2). Energy Department Selects TVA and Holtec to Advance Deployment of U.S. Small Modular Reactors. energy.gov. 

[17] Nuclear Engineering International (2026, March 25). Aalo unveils critical test reactor; Aalo Atomics company blog (2025, August). Aalo Selected by DOE to Test Aalo-X. 

[18] Lovering, J.R., Nordhaus, T., & Yip, A. (2016). Historical construction costs of global nuclear power reactors. Energy Policy, 91, 371-382. 

[19] Trellis (2025). Amazon, Google, Meta and Microsoft go nuclear. trellis.net. 

[20] Locatelli, G., & Mancini, M. (2010). Small-medium sized nuclear, coal and gas power plant: A probabilistic analysis of their financial performances. Energy Policy. 

[21] Locatelli, G., Bingham, C., & Mancini, M. (2014). Small modular reactors: A comprehensive overview of their economics and strategic aspects. Progress in Nuclear Energy. 

[22] IPO Club (2025, December). SMR Stock Report. ipo.club. 

[23] European Commission (2026, March 10). Commission unveils strategy to bring Europe's first SMRs online by the early 2030s. energy.ec.europa.eu. 

[24] World Nuclear News (2026, March 13-16). Arup to support UK's first SMR at Wylfa; Regulatory justification for Rolls-Royce SMR. 

[25] Fedchenko, V. (2024, January). Small modular reactors may have climate benefits, but they can also be climate-vulnerable. SIPRI. 

[26] Kang, J.S., Chang, I.G., & Cheong, J.H. (2026, March). Comprehensive review of small modular reactor development focusing on challenges in the backend nuclear fuel cycle. Nuclear Engineering and Technology, 58(3). 

[27] Crownhart, C. (2025, May). Can nuclear power really fuel the rise of AI? MIT Technology Review. 

[28] Baker, T. E., Epiney, A. S., Rabiti, C., & Shittu, E. (2018). Optimal sizing of flexible nuclear hybrid energy system components considering wind volatility. Applied Energy, 212, 498–508.