Nuclear Stocks Fall Even as Hyperscalers Commit $710 Billion to Data Centers in 2026
Uranium equity funds have retreated even as four AI hyperscalers commit $710 billion to 2026 capital spending, exposing a timing gap between nuclear demand and new supply.
Global X Uranium ETF (URA) closed Friday (2026-08-01) at $39.07, down 2.01% on the session. The broader uranium equity complex has struggled to hold gains through 2026 even as the demand case for nuclear power has strengthened. Over the twelve months through early June (2026-06-01), URA had returned 62%, finance.yahoo.com reported, but that run has masked a sharp reversal in recent months.4,3
The demand side is not in question. Microsoft, Meta, Amazon, and Alphabet are collectively guiding to more than $710 billion in combined 2026 capital expenditure, finance.yahoo.com reported in mid-June (2026-06-12). The Department of Energy projects data centers will account for up to 12% of U.S. electrical demand by 2028, with Lawrence Berkeley National Laboratory estimating a range of 6.7% to 12% of total annual consumption. The EIA's Annual Energy Outlook 2026 models data center server electricity use growing over the same horizon.5,4
Constellation Energy, the largest U.S. nuclear fleet operator, delivered first-quarter FY2026 results that reflect the business strength behind the narrative. Adjusted EPS came in at $2.74, ahead of the $2.60 consensus, with revenue of $11.122 billion beating expectations by 28% and growing 64% year over year. Management reaffirmed full-year adjusted operating EPS guidance at $11.00 to $12.00 and set a target of more than 20% base EPS growth through 2029.5
Those results have not steadied the stock. Constellation shares had fallen to $242.30 as reported in mid-June (2026-06-12), down 19% over the prior month and 31% year to date at that point, according to finance.yahoo.com. Cameco had dropped roughly 20% over the same stretch.5
Supply timing is part of the explanation. The IEA expects more than 70 gigawatts of new nuclear capacity to come online by the mid-2030s, described in late May (2026-05-28) Forbes reporting as one of the strongest pipelines in 30 years. Forbes noted explicitly that this pipeline will not power data centers in the near term or provide near-term grid backup capacity.2
Three ETF structures offer different routes into the trade. The Range Nuclear Renaissance Index ETF (NUKZ), targeting reactor operators and small modular reactor developers, had returned 42% over the prior year as of early June (2026-06-01), finance.yahoo.com reported. VanEck Uranium and Nuclear ETF (NLR) blends utilities, miners, and services companies; it had returned 133% over five years, though its diversified basket limits upside against high-conviction single-name plays. URA offers the deepest liquidity in pure uranium mining exposure among the three.4,3
The Uranium and Nuclear ETF (URAN) adds a fourth variation: direct exposure to Japanese and Korean reactor builders at lower cost than competitors, according to AOL reporting from late May (2026-05-28). That exposure carries distinct macro risk. The IMF cut South Korea's 2026 growth forecast to 1% on April 22nd (2026-04-22), down from 2% in January, and trimmed Japan's forecast by 0.5 percentage points in the same revision. The Trump administration had announced 25% reciprocal tariffs on Korean goods and 24% on Japanese goods before suspending them for 90 days, per mid-May (2026-05-19) reporting, introducing uncertainty around the economic backdrop in both markets.3,1
Back at Constellation, the balance sheet warrants attention. Long-term debt jumped to $17.5 billion following the Calpine acquisition, and the nuclear fleet's capacity factor slipped to 92% from 94% year over year. Integration execution remains an overhang.5
The concrete catalyst ahead is the PJM Reliability Backstop Procurement framework, scheduled to enable bilateral contracting between data centers and nuclear operators from March 2027. Constellation already holds long-term power purchase agreements with Microsoft, Meta, and CyrusOne, including a 380 megawatt deal at Freestone signed in February (2026-02), with exclusivity on an additional 380 megawatts. How broadly the PJM framework opens that market beyond a handful of existing counterparties sets the pace of any sector recovery — the ETF gains of the past twelve months still need contract volume, not just narrative, to find a floor.5