A new study by Georgia Tech researchers Brian An, Daein Kang, John Kim, and Moe Kyaw Thu analyzes how national governments describe Small Modular Reactors (SMRs) in official energy policy documents. Using natural language processing (NLP) on more than 800,000 words extracted from 66 national and international energy plans, the authors assess whether SMRs are framed as narrowly technical innovations or as contributors to broader urban energy transitions. Their findings show that SMR discourse remains dominated by references to reactor design, regulation, and safety, while themes central to modern energy planning—such as resilience, urban–rural equity, cogeneration, and diversified energy services—appear inconsistently and with low prominence.
Perhaps most notably, governance‑related concepts such as community engagement, siting justice, and public trust are largely absent from the dominant keyword clusters revealed through TF‑IDF and LDA analysis. This pattern contrasts with long‑standing evidence that nuclear deployment outcomes hinge on procedural fairness, transparency, and risk communication. As cities face rising electricity demand, climate‑driven outages, growing data center loads, and new siting pressures, the lack of urban‑relevant framing in national SMR strategies may limit the technology’s ability to support equitable and resilient energy systems.
The authors conclude that viewing SMRs chiefly as engineering solutions risks missing their potential contributions to multi‑service energy portfolios and resilience planning. They argue that meaningful integration of SMRs into smart energy cities will require a broader policy architecture—one that explicitly addresses governance, cross‑sectoral applications, spatial justice, and local participation. Expanding future analyses to include state, provincial, and municipal policies will also be essential, given that these levels of government oversee land use, community engagement, and emergency management—factors central to nuclear siting and energy justice.
To learn more you can read the paper (How National Nuclear Energy Plans Frame Small Modular Reactors (SMRs): Evidence from Natural Language Processing of Global Policy Documents by Brian An, Daein Kang, John Kim, Moe Kyaw Thu :: SSRN), listen to the podcast, or contact us (epicenter@gatech.edu – which we can then forward to the faculty).
Podcast:
KEY TAKEAWAYS
1. National SMR Policy Discourse Is Predominantly Technical
Across all documents studied, references to SMRs focus heavily on reactor design, safety, regulation, and development, with limited attention to socio‑economic or urban applications.
2. Urban and Societal Priorities Receive Minimal Emphasis
Themes such as geographic equity, climate resilience, and cogeneration appear only weakly in the NLP results, suggesting a disconnect between emerging SMR narratives and actual policy language.
3. Governance and Public Trust Are Underrepresented
Despite their central role in nuclear deployment, concepts like community engagement, siting fairness, and risk communication receive little attention in national strategies.
4. Current Framing Risks Reinforcing Spatial Inequities
Without explicit policies addressing risk distribution, SMR deployment may repeat historical patterns where rural or marginalized communities host facilities while urban centers benefit.
5. Technological Innovation Alone Is Insufficient for Deployment
Although SMRs offer modularity, passive safety, and potential multi‑service value, national policy documents do not articulate how these features integrate into broader planning frameworks.
6. Broader Policy Design Is Needed for Urban Integration
Effective SMR deployment will require governance mechanisms, community benefit structures, cross‑sectoral service planning, and links to municipal resilience strategies.
7. Sub‑National Policies Will Be Crucial
Cities and states manage land use, emergency planning, and public engagement. Future research should examine how these levels frame SMRs, as they will shape real‑world implementation.
This summary was written with the assistance of Microsoft Copilot on January 15, 2026. Its content was edited and verified by EPIcenter staff and affiliates. The podcast was created with Google NotebookLM, December 10, 2025. Published January 27, 2026.