The Social Costs of an AI-Led Nuclear Energy Renaissance
Summary
This research paper argues that the AI-driven push for nuclear energy in the U.S. risks shifting costs and environmental burdens onto host communities while benefiting corporations, especially as federal regulatory capacity declines.
Key insights
- Generative AI is significantly more energy-intensive than previous AI models, contributing to a projected increase in data center electricity use from 4.4 percent of total U.S. production (2014–2024) to at least 12 percent by 2028.
- The U.S. government is actively promoting nuclear expansion through executive orders from President Donald Trump in May 2025 to quadruple domestic nuclear electricity production within 25 years, building on the 2024 ADVANCE Act and the 2022 Inflation Reduction Act.
- There is a risk that nuclear energy becomes a 'private good' where electricity from new or reopened reactors powers data centers instead of homes, potentially increasing electricity prices for utility ratepayers through offtake agreements.
- Increased nuclear power production will exacerbate the lack of a permanent repository or consolidated interim storage for spent nuclear fuel, increasing the risk of groundwater contamination from materials like tritium in host communities.
- The co-location of nuclear power plants and data centers may cause severe water shortages, as both require large amounts of water for cooling, particularly in water-stressed regions like California, Texas, and Arizona.
- Small modular reactors (SMRs) may provide fewer direct economic benefits to host communities compared to traditional large reactors because their factory-based mass production requires less on-site labor.
- The Trump administration's efforts to reduce the regulatory capacity and authority of the Nuclear Regulatory Commission (NRC) may prevent host communities from accessing information or remediation for health and environmental risks.
- The urgency of tech entrepreneurs to deploy nuclear power within two to five years to meet AI needs could lead to 'cutting corners on safety' or the creation of unfinished projects.
Cite the original document
- APA
- Sarkar, J. (2025). The Social Costs of an AI-Led Nuclear Energy Renaissance. Carnegie Endowment for International Peace. https://carnegieendowment.org/research/2025/10/the-social-costs-of-an-ai-led-nuclear-energy-renaissance
- Chicago
- Sarkar, Jayita. The Social Costs of an AI-Led Nuclear Energy Renaissance. Carnegie Endowment for International Peace, 2025. https://carnegieendowment.org/research/2025/10/the-social-costs-of-an-ai-led-nuclear-energy-renaissance.
- Wikipedia
- {{cite report |last1=Sarkar |first1=Jayita |title=The Social Costs of an AI-Led Nuclear Energy Renaissance |publisher=Carnegie Endowment for International Peace |date=16 October 2025 |url=https://carnegieendowment.org/research/2025/10/the-social-costs-of-an-ai-led-nuclear-energy-renaissance |access-date=5 November 2026 |via=Climate Insights Directory (document carnend-002402)}}
- BibTeX
- @techreport{sarkar2025social, author = {Sarkar, Jayita}, title = {{The Social Costs of an AI-Led Nuclear Energy Renaissance}}, institution = {Carnegie Endowment for International Peace}, year = {2025}, month = oct, url = {https://carnegieendowment.org/research/2025/10/the-social-costs-of-an-ai-led-nuclear-energy-renaissance}, urldate = {2026-11-05}, note = {Indexed by Climate Insights Directory, document carnend-002402} }
Full text
Collected · Record updated