Small modular reactor.
The boring reactor wins.

American nuclear just had its best year in half a century. The pure-play small modular reactor stocks are down by a third to a half.

Both halves are true. Five new reactors went critical this summer under the Energy Department's fast-track pilot program, including one Oklo built in Texas in 11 months, in an industry where projects usually run a decade late. In late September, the NRC issued the first US construction permit for a commercial small modular reactor, four months ahead of schedule. Meanwhile, Oklo's stock is down about half this year, X-energy trades well below its April IPO price, and Holtec launched an IPO in September and pulled it eight days later.

The promise hasn't changed. A small modular reactor (SMR) is a nuclear plant shrunk to 300 megawatts or less and built from factory-made pieces instead of poured on site over a decade. The idea is the one that made cars cheap: build the tenth unit for less than the first, and the hundredth for less than the tenth. Nuclear has never had a learning curve, and SMRs are an attempt to give it one. The customers are real, too. Data centers need power that runs around the clock without burning gas, and Google, Amazon, and Meta have all signed up with reactor developers.

So why the selloff? Because none of these reactors sells a kilowatt-hour for years, and until then the developers pay their bills by selling stock. Oklo filed to sell up to $1 billion of shares in September, and NuScale's share count is up about 50% in a year. When AI sentiment cooled, the market stopped wanting to fund 2030 power with 2026 dollars.

Nobody on this list of small modular reactor stocks gets paid for SMR power in 2027. What separates them is when the cash actually shows up:

• Paid now: the reactor vendor with a permit, plus the fuel and parts suppliers. They collect before any SMR starts up.
• Paid later: the developers, who fund themselves with stock until the power flows.
• Private: three big names still waiting on a ticker.

These companies take in real money for building, fueling and supplying reactors, whether or not any one developer's design ever sells power.

GE Vernova NYSE: GEV

If you want the SMR most likely to make commercial power first, it belongs to GE Vernova. Its BWRX-300 is under construction at Darlington, Ontario, the first of four units planned there. On September 29, TVA got the first US construction permit for a commercial SMR to build one at Clinch River, Tennessee. A week earlier, GE signed a memorandum with partners to market the same design in Central and Eastern Europe and the UK.

SMRs are a small piece of a giant that sells gas turbines and grid gear. That's the appeal. GE Vernova never has to sell stock to keep building.

BWX Technologies NYSE: BWXT

BWXT has built the reactors in the US Navy's submarines and carriers for decades, for a customer that doesn't cancel and doesn't miss a payment. The same shops make TRISO fuel and components for other companies' reactors. In August the Army picked BWXT to build a 20-megawatt microreactor at Fort Campbell. You get paid for nuclear work today, with SMRs as the upside.

Centrus Energy NYSE: LEU

Most of the advanced reactors on this list run on HALEU, uranium enriched to between 5% and 20%, well above the roughly 5% ordinary reactors burn. Centrus runs the only NRC-licensed HALEU plant in the country, in Piketon, Ohio.

X-energy signed a definitive enrichment contract in August. Oklo signed a letter of intent in June. Radiant and Antares, two microreactor startups, signed in September. Every reactor that goes critical is another customer in Centrus's line, and today there's no other line to stand in.

Each of these small modular reactor stocks is a bet on one reactor design getting built before the money runs out. Until then, the shareholders are the customers.

Oklo NYSE: OKLO

Oklo is the speed story. Its Groves test reactor in Texas went critical on August 5, eleven months after groundbreaking, and was the first pilot-program reactor to go critical on private land. The real product is Aurora, a fast reactor of up to 75 megawatts that can run on fresh fuel or recycled nuclear waste. The first one is going up at Idaho National Laboratory, targeting 2028.

The market loves the speed and hates the bill. Oklo filed to sell up to $1 billion of stock in September, and the shares are down about half this year. If Aurora starts on time, today's price will look like a gift. If it slips, expect more share sales.

X-energy NASDAQ: XE

X-energy is the developer furthest along on making its own fuel. It has contracted enriched uranium from Centrus, and its TRISO-X subsidiary will turn that into finished fuel under a first-of-its-kind NRC license, at what X-energy calls the first new US fuel plant in more than 50 years. The plant's structure was finished in September. When fuel is the bottleneck, owning the last step matters.

Its customers are industrial, not just tech: Dow's chemical site in Seadrift, Texas, plus Amazon and Britain's Centrica, whose chosen reactor design was cleared to enter UK design review in September. Q2 revenue and grant income was $54.6 million, up 154%, with $1.9 billion of liquidity. The stock still trades more than a third below its $23 IPO price, which is what a hot IPO usually looks like five months later.

NuScale Power NYSE: SMR

NuScale has the only SMR design certified by the NRC, and still no firm US order. Everything hangs on one deal. Partner ENTRA1 is negotiating to build up to 6 gigawatts for the Tennessee Valley Authority, which would be the largest nuclear program in US history, and NuScale has said it hopes for a deal by year-end.

Until then NuScale funds itself with stock. Its total share count is up about 50% in a year, it set up another $750 million share sale in August, and UBS cut it to Sell in September. A signed TVA deal flips all of that overnight. Without it, you own a certified design nobody has bought.

NANO Nuclear Energy NASDAQ: NNE

NANO Nuclear is the lottery ticket. It's designing Kronos, a microreactor small enough to truck to a mine, a military base or a remote data center. There's no licensed design and nothing under construction. Its first construction permit application, at the University of Illinois, is still with the NRC, and its only revenue is a rounding error from a fuel-transport business it bought in May. In August the Army handed out its microreactor contracts to five companies, and NANO wasn't one of them.

The biggest checks in SMRs are still going to private companies. Each of these would reset the comps the day it lists.

TerraPower

Bill Gates' reactor company is building Natrium in Kemmerer, Wyoming: a 345-megawatt sodium-cooled reactor with a molten-salt battery that can push output to 500 megawatts when the grid needs it. Meta agreed in January to support up to eight of them. In September, TerraPower and Bechtel failed to agree on terms for the next phase, and TerraPower is rebidding the main construction contract. It says the 2030 schedule hasn't moved. The biggest hyperscaler nuclear deal now has to survive its first contractor.

Kairos Power

Kairos is Google's reactor company, with a deal for up to 500 megawatts by 2035. It starts with Hermes 2 in Oak Ridge, Tennessee, which will sell about 50 megawatts into TVA's grid. In September, Samsung C&T agreed to invest up to $100 million in cash and engineering work and join the team building Kairos's first commercial plant. Kairos has never confirmed IPO plans, but it's the private name the market most wants to own.

Holtec

Holtec is the grown-up of the group. It already makes money storing spent fuel and decommissioning old plants, and it's restarting Palisades in Michigan, set to be the first US plant to come back from decommissioning. In September it launched an IPO at a valuation of up to about $10 billion, then pulled it eight days later, citing market conditions.

When a profitable nuclear company can't get a deal done, that tells you where sentiment is. When Holtec relaunches, that will tell you it's turned.

The first US construction permit for a commercial SMR went to the most boring design on this list. The BWRX-300 is a boiling-water reactor, the same basic physics regulators have licensed for 60 years, shrunk down and simplified. Meanwhile the market has paid up for the exotic designs: sodium, molten salt, fast neutrons.

That's backwards from how this industry gets cheaper. The whole SMR pitch is a learning curve, and a learning curve only starts on the second unit. A design built once is a science project. A design built four times at one site, then in Tennessee, then in Europe, is a product: the suppliers have done it, the crews have poured that concrete, and the regulator has already read the paperwork.

So the 2027 signal for small modular reactor stocks isn't the next criticality headline or the next hyperscaler letter of intent. It's the repeat order. The first design to sign its second, third and fourth customers gets the cost curve, and everyone else ends up bidding against it.

Right now, that's the boring one.