What's Happening?
The push for nuclear power to meet the energy demands of AI data centers is encountering significant cost and regulatory challenges, despite substantial venture funding. U.S. nuclear startups have raised $4.6 billion this year, an increase from $3.8 billion in 2025.
However, only two new commercial reactors are currently under construction in the U.S. and Canada, with no other projects having received regulatory approval. A June 2 study by the Carnegie Endowment for International Peace, authored by John Pendleton and Mackenzie Schuessler, indicates that hyperscalers spent $443 billion on data centers in 2025 and are projected to exceed $700 billion in 2026. This rapid demand for power contrasts sharply with the slow pace of nuclear power development, which can take a decade or more to deliver. The Carnegie study also flags high-assay low-enriched uranium (HALEU) as a nonproliferation and security concern, noting the immaturity of its broader fuel chain. Washington has allocated $2.7 billion for uranium enrichment in January 2026 and requested $30 billion in new loan authority for the 2026 budget to address these fuel supply issues.
Why It's Important?
The disparity between the rapid growth of AI data center power demand and the slow deployment of new nuclear capacity poses a critical challenge for the U.S. energy landscape and its economic stakeholders. The high upfront costs and lengthy regulatory processes for nuclear projects mean that the promised nuclear relief for AI data centers may not materialize until the mid-2030s. This timeline mismatch could force a reliance on other energy sources, such as natural gas, to bridge the gap, potentially impacting carbon emission goals and utility bills. The Carnegie Endowment's findings underscore that while AI demand has made nuclear power more fundable, it has not accelerated its development. This situation affects tech companies seeking reliable, carbon-free power, investors in nuclear startups facing long-term returns, and consumers who may experience rising electricity costs due to increased data center demand. The immaturity of the HALEU fuel chain also presents national security and nonproliferation concerns, requiring significant government investment and development.
What's Next?
In the immediate future, existing nuclear plants are expected to be the primary source of nuclear power for data centers, with more deals focusing on life extensions and restarts rather than new reactor construction. The Nuclear Regulatory Commission (NRC) has taken steps to streamline the licensing process, including Executive Order 14300 in May 2025, which set 18-month deadlines for new reactor applications, and the finalization of Part 53 in March 2026, a risk-informed framework for advanced reactors. However, these regulatory improvements do not address the fundamental challenges of financing and supply chain constraints. The White House’s March 2026 Ratepayer Protection Pledge aims to ensure that data center needs do not raise household bills, though its enforceability remains to be seen. Stakeholders will be watching for a third commercial reactor entering construction in the U.S. or Canada, updates on SMR costs, and the progress of Oklo’s combined license application under the new Part 53 framework. The pace of hyperscaler spending on data centers will also be a key indicator, as a slowdown could reduce the urgency for nuclear contracts, while continued growth will intensify pressure on regulators and developers.
Beyond the Headlines
The current situation highlights a fundamental tension between the 'software speed' of the AI industry and the 'infrastructure speed' of nuclear development. This gap raises ethical and policy questions about how to balance technological advancement with sustainable energy infrastructure. The reliance on existing plants for immediate power needs, while beneficial in the short term, could lead to a redirection of electricity from other customers, potentially exacerbating energy supply issues in certain regions. The high cost of new nuclear builds in the U.S. and Europe compared to countries like China and South Korea points to deeper issues in construction, labor, and financing, suggesting that the problem is not merely technological but also systemic. The proliferation concerns associated with HALEU also introduce a complex geopolitical dimension, requiring careful international cooperation and robust safeguards. Ultimately, the long-term success of AI's nuclear power push will depend on overcoming these multifaceted challenges, which extend beyond mere capital investment to encompass regulatory reform, supply chain resilience, and public acceptance.
