You didn’t have to know the difference between yellowcake and poundcake to see that nuclear energy was a hot trade last year. Spencer Jakab’s Heard on the Street column in Friday’s Journal opens with that line and then makes the point that separates a good idea from a good investment: the future remains bright, and investors who chased the later stages of the rally are nursing losses. Even with some stocks down well over half from their peaks, valuations mostly remain too hot.
How the trade was supposed to work
The logic was sound and it still is. The AI build-out stokes power demand from hyperscalers looking for as much reliable low-carbon electricity as they can find, and solar and wind cannot match nuclear for baseload. So the hyperscalers signed long-term contracts — Microsoft and Amazon among those doing “behind the meter” deals with operators of existing plants or ones soon to come online.
That arrangement technically bypasses the grid that ordinary customers use. In the case of nuclear, it mostly diverts power that would otherwise have been sold by a utility, which is worth noticing the next time a regional electricity bill goes up.
The Holy Grail was the next step: small modular reactors, non-utility scale, mass-producible. Along with technological progress and government support, that thesis drove enormous investor interest into unprofitable reactor companies like NuScale and fuel providers like Centrus.
What went wrong
The road has been longer and bumpier than investors assumed. Holtec postponed an IPO that had been expected to price Thursday. UBS downgraded NuScale to “sell,” questioning its path to profitability — and NuScale’s design is closer to commercialization than several peers. X-Energy, which has financial backing from Amazon, surged 23% on its first day of trading in April to a market value near $12 billion and has dropped steadily ever since, shedding more than half.
The suggested alternative is the boring one: large, profitable utilities that already own reactor fleets, such as Constellation Energy (CEG) and NRG Energy (NRG). And even there Jakab attaches a warning — Citigroup analysts put both on a list of the most crowded utility stocks.
Our read
This is the clearest case study of the year in the difference between being right about the world and making money on it.
The demand is not in doubt. Data centers are on track to take something like 12% of U.S. electricity by 2030. The physics of baseload power is not in doubt either. What was always in doubt was the timeline — reactors take years merely to get off the drawing board, let alone to generate a kilowatt-hour — and a stock price is a claim on cash flows arriving at a particular time. Push the cash flows out five years and the same correct thesis supports a far lower price.
Jakab’s closing advice is that the smart move might be to wait for nuclear to go out of fashion again, which it seems to do about once a decade. That is a hard thing to act on and an easy thing to respect.
So this desk is adding Constellation Energy (CEG) and NRG Energy (NRG) to the watch list and not to any model. The reasoning is deliberately narrow: they are the profitable way to own the thesis, they are also the crowded way, and the September letter’s conditions for putting new money to work failed two of three tests this week. Watching something is what you do with a good idea that has not yet met its own entry rule.