Nuclear goes modular: AI power surge sparks manufacturing race and regulatory shakeup

The gist
A new breed of nuclear startups and regulatory reforms is supercharging America’s race to power AI with modular, factory-built reactors.
What to know
- Valor Atomics and Radiant are shifting nuclear from lab to assembly line, using modular designs and rapid hardware iteration to meet surging AI data center demand.
- Regulatory shakeups—like TerraPower’s Natrium plant earning the first new reactor permit in a decade and 18-month federal licensing mandates—are breaking decades-old bottlenecks.
- Despite higher early costs (up to $400/MWh), small modular reactors are drawing billions in investment from Google, Microsoft, and the DOE, positioning nuclear as the digital era’s baseload backbone.
Nuclear’s Manufacturing Revolution
Startups like Valor Atomics and Radiant are treating nuclear as a manufacturing challenge, rapidly scaling modular reactors with assembly-line speed to serve AI’s insatiable energy appetite.
Valor Atomics exemplifies a manufacturing-centric approach to advanced nuclear reactors by prioritizing rapid hardware iteration and modular design to meet the urgent energy demands driven by AI infrastructure growth. Founder Isaiah Taylor emphasizes treating nuclear development as a manufacturing challenge rather than a regulatory or design puzzle, enabling the company to move swiftly from building a non-operational reactor during their seed round to preparing for startup within a single funding cycle. Their vision integrates concrete short-term milestones with a long-term goal of making energy ten times cheaper, reflecting a pragmatic, scalable strategy aligned with market needs.
Radiant is advancing manufacturing-centric innovation by transitioning from design to active production of their one-megawatt modular reactors, with all components ordered and assembly underway at their new 80-acre Tennessee facility on Department of Energy land. This vertical integration and modularity enable scalable mass production tailored for diverse applications including military bases, oil and gas sites, and AI-driven data centers—highlighted by a significant order from Equinix for 20 units. Their recent $300 million funding round is focused on capital expenditure and operational scale-up rather than R&D, underscoring a strategic shift to manufacturing and deployment acceleration.
The broader nuclear innovation landscape underscores the strategic advantage of small modular reactors (SMRs) as a manufacturing-centric solution that balances cost, scalability, and operational flexibility. SMRs, typically sized between 75 and 350 megawatts, offer reduced capital risk through staggered construction and refueling, making them ideal for replacing retiring coal plants and meeting incremental capacity needs. This modular approach complements emerging advanced coolant technologies and diverse reactor sizes, from microreactors to gigawatt-scale plants, signaling a future where manufacturing efficiency and modularity drive nuclear’s role in powering AI and other high-demand sectors.
Valor Atomics’ commitment to vertical integration—ranging from inventing novel high-density, rebar-free concrete to in-house manufacturing of critical components—serves as a 'secret weapon' enabling rapid iteration and cost control in nuclear reactor production. This hands-on, hardware-first philosophy contrasts sharply with the traditional nuclear industry’s reliance on modeling and simulation, allowing Valor to drastically shorten the time between reactor startups, or 'tick rate,' with ambitions to reduce intervals from years to minutes. Their approach not only accelerates deployment but also attracts top young engineering talent eager to build tangible nuclear hardware, marking a transformative shift in nuclear manufacturing culture.
Regulators Hit Fast-Forward
Landmark policy shifts and expedited licensing are dismantling decades-old nuclear bottlenecks, paving the way for a new wave of advanced reactors to meet AI-driven power demand.
By early 2026, regulatory progress marked a pivotal shift in U.S. nuclear deployment, exemplified by TerraPower’s Natrium plant receiving federal approval—the first new commercial reactor permit in nearly a decade and the first for a non-light water reactor in over 40 years. This milestone underscores the urgency driven by surging electricity demand from AI infrastructure, which alone may require upwards of 30.5 gigawatts of new power, equivalent to roughly 90 Natrium-sized reactors, highlighting the critical need for streamlined licensing and accelerated construction to meet future energy needs.
The Trump administration’s 2025 executive orders catalyzed a regulatory overhaul aimed at quadrupling the U.S. nuclear fleet by 2050, mandating the Nuclear Regulatory Commission (NRC) to license new reactors within 18 months and designating nuclear plants powering AI facilities as 'defense-critical infrastructure' to bypass lengthy reviews. Additionally, the NRC’s reconsideration of the restrictive ALARA radiation regulation—long criticized as a 'zero banana rule' that effectively stalled new construction—signals a willingness to dismantle historic barriers that have hindered nuclear expansion for decades.
The Department of Energy’s $17.5 billion commitment to fast-track 10 Westinghouse AP1000 reactors exemplifies robust federal policy support designed to overcome entrenched regulatory and financial obstacles. By requiring utility partners to co-invest $1 billion per project and standardizing procurement with fixed pricing, the initiative aims to slash construction timelines by up to three years and jumpstart the domestic nuclear supply chain, directly aligning with national decarbonization goals and the urgent electricity demands of data centers, as emphasized by DOE Secretary Chris Wright.
Despite these advances, the nuclear industry still grapples with a 'chicken and egg' dilemma where regulatory approval depends on operational data, but generating such data requires reactor operation—a catch-22 that has historically favored simulation over hardware iteration. Reviving the original dual regulatory pathways envisioned in U.S. nuclear legislation—one for commercial deployment and another for testing—could enable faster, less costly licensing and iterative hardware development. Valar Atomics’ success in leveraging a DOE pathway revived by executive order, combined with their vertically integrated, venture-backed approach, illustrates how regulatory reform paired with innovative business models can accelerate nuclear innovation and scale production rapidly.
Investors Bet on Nuclear’s Future
Tech giants and policymakers are pouring billions into nuclear as public sentiment shifts and market dynamics evolve, positioning the sector for a comeback despite persistent cost hurdles.
By early 2026, the surging electricity demand driven by AI infrastructure—estimated to require about 30.5 gigawatts, or roughly 90 TerraPower-sized reactors—has spotlighted nuclear energy as a critical solution despite economic challenges. While small modular reactors (SMRs) promise adaptability, their current cost estimates ranging from $150 to $400 per megawatt-hour remain significantly higher than modern gas plants, underscoring the need for cost reductions through mass manufacturing and regulatory support to achieve competitiveness.
Investor confidence in nuclear innovation is gaining momentum, exemplified by TerraPower’s federal approval to build the Natrium plant in Wyoming—the first new U.S. commercial non-light water reactor approved in over 40 years—signaling a renewed belief in nuclear’s potential role in the energy mix. This optimism is tempered by market realities where solar and natural gas currently dominate new capacity additions, yet nuclear’s reliable, carbon-free power is increasingly valued for complementing intermittent renewables and powering energy-intensive AI and industrial applications.
The market dynamics are further energized by a multifaceted investment landscape where tech giants like Google and Microsoft are directly purchasing power from recommissioned nuclear plants, reflecting diversified funding strategies and corporate commitment to nuclear innovation. This trend coincides with shifting public sentiment—particularly among millennials concerned about climate change—and policy momentum, including renewable portfolio standards in 32 states that mandate fossil fuel phase-outs and boost clean energy investments, indirectly benefiting nuclear development as part of a broader clean energy strategy.
Policy and investor focus on SMRs is crystallizing into targeted strategies such as the SMR Fund (SMRF), which aims to capture long-term growth by investing in companies actively implementing SMR technology designed for scalability and adaptability, especially for AI data centers. This approach reflects a strategic recognition that unlocking SMRs’ potential both domestically and internationally hinges on deploying technology layers that enable mass application, marking a pivotal shift in how nuclear innovation is financed and scaled.
SMRs: The AI Energy Backbone?
Advanced modular reactors are drawing massive investment and policy focus as the only scalable, carbon-free power source capable of keeping pace with AI’s explosive energy needs—if costs can come down.
By early 2026, nuclear power, especially through advanced small modular reactors (SMRs) like TerraPower's Natrium plant, emerged as a promising solution to meet the surging electricity demands of AI data centers and digital infrastructure, with estimates suggesting AI alone could require up to 30.5 gigawatts—equivalent to about 90 TerraPower-sized reactors. SMRs offer reliable, carbon-free baseload power that complements intermittent renewables, making them strategically vital for powering critical AI infrastructure that demands continuous electricity. JPMorgan highlights that with sufficient investment and regulatory support, SMRs can be built to serve early markets including data centers and retiring coal plant sites.
Despite the strategic promise of SMRs, economic viability remains a significant hurdle, as TerraPower’s first unit is projected to cost around $4 billion with electricity costs estimated between $200 and $400 per megawatt-hour—substantially higher than modern gas plants priced at $55–$85. This cost disparity underscores the challenge of achieving economies of scale and highlights the importance of scaling beyond initial projects; as TerraPower’s CEO candidly noted, the true test will come 'when the tenth is built,' emphasizing that early deployments are experimental rather than definitive solutions.
A March 2026 KeyLogic report underscores nuclear power’s critical role in meeting America’s AI and data center energy demands, projecting that data center electricity use could increase five to fifteen times by 2050, potentially consuming nearly 40% of today’s total U.S. electricity. This surge has already attracted over $20 billion in commitments from major tech companies, while the Trump Administration has launched policies targeting 300 GW of new nuclear capacity by mid-century. Achieving these ambitious goals will require unprecedented coordination across financial, regulatory, and industrial sectors, alongside further research into electricity rates, permitting, and grid reliability to ensure nuclear’s backbone role in AI infrastructure energy supply.
Complementing these strategic and economic analyses, an Argonne National Laboratory study highlights practical pathways to integrate nuclear power with data centers, including restarting retired reactors and deploying SMRs to provide reliable 24/7 electricity critical for U.S. digital infrastructure growth. While pairing nuclear plants with data centers promises significant economic benefits such as job creation and increased tax revenue, the integration faces notable challenges including regulatory hurdles, lengthy construction timelines, and high upfront costs. Future research aims to develop a comprehensive framework focusing on feasibility, siting, and case studies to support sustainable, scalable nuclear-powered digital infrastructure.








