Nuclear Stocks Fell Up to 88% From Their Highs. Most Still Sit Mid-Range on Their Own History.

"Below 52-week high" compares the latest weekly close with the highest weekly close of the past year. Percentiles rank today's trailing multiple against the same company's weekly readings over the last five years. n/m means not meaningful: trailing earnings are negative.

The equal-weight basket of these 11 companies fell 41% in the year to the week of October 19, and NLR fell 39%. On EV/Sales, though, the ranked members sit between the 24th and 70th percentiles of their own ranges. Most five-year windows start in late 2021, before the rally, so a large fall from a recent high can still leave a stock in the middle of its own range. GE Vernova ranks against 90 weeks of history since its 2024 spin-off.

Operators sit low on earnings

Constellation and Vistra sat in the bottom third of their own P/E ranges, at the 22nd and 31st percentiles, while their EV/Sales ranks sat mid-range, at the 52nd and 60th. When the earnings rank sits well below the sales rank, profit per dollar of revenue is usually higher than in most of the past five years. PSEG sat low on both: the 32nd percentile on P/E and the 24th on EV/Sales. A low P/E rank after profits rise can also mean investors doubt those profits will last; operator earnings depend on power prices and contracts.

What a rank cannot show

Oklo has only two quarters of reported revenue, too few for a weekly history, so it has no rank. Its enterprise value is more than 2,000 times trailing revenue. Cameco files annual reports as a Canadian company, so it shows a market snapshot without a rank.

NuScale is the case to read slowly. Its latest weekly close was 88% below its high of the past year, and its EV/Sales of 83.2x still ranked at the 68th percentile, because trailing revenue fell 81%. Uranium Energy, with revenue down 44%, ranked at the 65th. When price and revenue fall together, a sales multiple can stay high. The missing-percentile guide and how to value nuclear stocks explain both cases.

Policy moved faster than revenue

Executive Order 14300, signed in May 2025, set a goal of 400 gigawatts of US nuclear capacity by 2050, up from about 100 in 2024, and an 18-month limit for decisions on new reactor applications. The Energy Department said its goal of three test reactors reaching criticality by July 4, 2026 was met (DOE). On September 29 the Nuclear Regulatory Commission approved construction of what it called the nation's first commercial small modular reactor, a 300-megawatt GE Vernova Hitachi design at TVA's Clinch River site; TVA still needs a separate operating license (NRC).

None of that is revenue yet. Oklo's second-quarter 2026 filing showed about $1.2 million of revenue, from services rather than power sales. New reactors take years to license, finance and build.

Changes to AI Power

Oklo, NuScale and Cameco moved from AI Power to this theme on October 23, 2026, where developers and fuel suppliers have their own layers. Constellation, Vistra and GE Vernova stay on both pages: AI Power follows data-center demand for electricity, and this page follows the reactors.

The nuclear screen lists all 11 companies with a switch for each measure, and the uranium screen narrows to mining and enrichment.

Numbers are from StockResearch's SEC-filings and weekly-price dataset read Oct 23, 2026, with the latest weekly observation beginning Oct 19. Price comparisons use weekly closes. Multiples and percentiles are rounded for display. The Nuclear member table shows current readings. This is research, not investment advice.