Goldman Flags 2.3 Billion Pound Uranium Supply Gap as Natural Reactor Programs Expand
Goldman Sachs projects a 2.3 billion pound uranium supply shortfall through 2045, with natural-uranium reactor programs from India to the United States adding to demand pressure.
The URA uranium exchange-traded fund stood at $44.91 as of August 8, 2026, up 3.64%, a gain that aligns with a demand forecast Goldman Sachs formalized in May (2026-05-19) — the bank's first explicit incorporation of small modular reactors into its uranium supply-and-demand model. Analyst Brian Lee projected cumulative SMR deployments of nearly 46 gigawatts by 2045. That addition lifted Goldman's 2045 nuclear generation forecast by about 6% and layered 62 million pounds of new uranium demand on top of prior long-term estimates, a 17% upside to the bank's earlier figures.3,2,4
The supply picture offers no matching response. Lee warned of a cumulative uranium supply deficit of approximately 2.3 billion pounds between 2025 and 2045, driven by simultaneous growth across the conventional reactor fleet and the emerging SMR pipeline. At 2.3 billion pounds, the projected shortfall cannot be resolved by optimizing existing mine output. It would require years of new mine development; most of those mines are not yet sanctioned.3,2
India sits at the natural uranium end of this demand picture. The country's three-stage nuclear program opens with natural uranium in pressurized heavy water reactors, recovers plutonium from spent fuel for use in fast breeder reactors as the second stage, and ultimately aims to produce uranium-233 from India's substantial thorium reserves. State-owned Nuclear Power Corporation of India is expected to account for around 54 gigawatts of the country's 100-gigawatt nuclear target by 2047; the remaining 46 gigawatts would need to come from other public bodies, state governments, private companies, and joint ventures, according to a report published on July 29, 2026 (2026-07-29).7
A fresh uranium arrangement between India and Australia drew attention in that context. India's heavy water reactors, unlike light-water designs prevalent in the West, run on natural uranium rather than enriched fuel. Enrichment capacity is not the bottleneck. Natural uranium feed volumes are. Any disruption to India's sourcing hits the first stage of its breeder cycle and constrains the entire downstream program.7
Elsewhere, the US regulatory environment has shifted toward advanced designs. Natura Resources secured a key safety approval for its liquid-fueled reactor design on July 22, 2026 (2026-07-22), the first such clearance for an advanced liquid-fueled system in the United States. A separate federal decision allowed Duke Energy's 759-megawatt Robinson Unit 2 plant in South Carolina to operate until 2050 under accelerated timelines. Life extensions, in aggregate, add to near-term uranium demand as reliably as new builds.6,3
EIA data set the size parameters for competing designs. Large-scale reactors typically produce between 550 megawatts and 1,500 megawatts per unit. SMRs top out at 300 megawatts; microreactors, a subset of SMRs, generate 20 megawatts or less and can operate independently of the main grid. Many advanced designs require high-assay low-enriched uranium enriched to between 5% and under 20% uranium-235, against sub-5% enrichment for conventional fuel. Natural uranium reactor designs skip the enrichment step entirely, giving them a potential cost and logistics edge where enrichment infrastructure is limited or politically constrained.1
Eagle Nuclear Energy Corp (NASDAQ: NUCL) disclosed in mid-July 2026 (2026-07-16) that it had engaged an artificial intelligence firm founded by a former Los Alamos National Laboratory theoretical physicist to support reactor simulation and optimization for its SMR program. The firm specializes in AI tensor network mathematics and physics-based simulations. The use of advanced computational methods to model reactor behavior is a pattern emerging across the sector, aimed at compressing development timelines that have historically taken decades.5
US electric utilities currently operate about 98 gigawatts of nuclear generating capacity, most of it built in prior decades. Very little new capacity has come online in recent years. High capital costs and lengthy licensing processes have held back new builds; Goldman's demand projections, NRC approvals, and a series of life-extension decisions are applying pressure from multiple directions simultaneously.1
But how fast the supply side can respond is what Goldman's deficit number ultimately turns on. If India's natural uranium program ramps faster than projected, or if SMR deployments run ahead of the 46-gigawatt base case, the 2.3 billion pound gap widens before new supply can reach it. New mine sanctions, fuel cycle investments, and HALEU enrichment capacity across multiple countries all need to move in parallel. As of mid-2026, progress on none of those fronts has matched the demand trajectory Goldman's model now describes.2,3