Disclaimer: This is not financial advice. Do your own research before making any investment decision. SMR stocks are largely pre-revenue and speculative.
Small modular reactors, or SMRs, are compact nuclear power plants designed to generate up to 300 megawatts of electricity per unit, far less than a conventional gigawatt-scale reactor but far more than any solar or wind farm can deliver on demand. That combination of density, reliability, and zero-carbon output is exactly what AI data centers need, and it is why the largest cloud operators in the world have started signing letters of intent, power purchase agreements, and direct equity stakes with SMR developers. The investment thesis is real. So is the risk.
If you are tracking AI energy stocks, SMR plays are the highest-conviction, longest-duration bet in that category. This guide covers the companies building these reactors, the hyperscaler deals backing them, the regulatory clock you need to understand, and the honest case for why most of these stocks could stay speculative for years before a single unit goes online.
What Are SMR Stocks?
SMR stocks are publicly traded shares in companies whose primary business involves designing, licensing, constructing, or operating small modular reactors. Unlike conventional nuclear power companies, which sell electricity from existing large-scale plants, most pure-play SMR developers are pre-revenue businesses. They generate no electricity today. Revenue, if it comes, depends on clearing a multi-year regulatory approval process, securing construction financing, and then building and commissioning plants that have never been built at commercial scale.
The investor bet is forward-looking: that demand for always-on, carbon-free baseload power from AI data centers, semiconductor fabs, and industrial decarbonization will create a large market for SMR output in the 2030s, and that the companies which get their designs licensed and their order books filled today will capture that market. Some of these companies are pure plays. Others are subsidiaries or divisions of larger conglomerates. A handful are development-stage startups that accessed public markets through SPAC mergers. Understanding which category you are buying matters enormously for risk calibration.
Why AI Data Centers Changed the Nuclear Calculus
Data centers running large language models operate at a power density that renewable intermittency cannot reliably serve. A single AI training cluster can draw hundreds of megawatts continuously, around the clock, regardless of weather. Grid-scale batteries do not yet exist at the cost and duration needed to fill renewable gaps at that scale. Natural gas peakers work but add carbon and fuel-price volatility. Nuclear, specifically always-on baseload nuclear, solves the reliability problem without the emissions problem.
Conventional large nuclear plants take 10 to 15 years from license to first power and cost tens of billions of dollars per unit. SMRs are designed to cut both figures materially: smaller physical footprint, factory-manufactured components, standardized designs that simplify licensing, and modular construction that lets operators add capacity incrementally. The theory is that what took 12 years and substantially more capital for a 1.1-gigawatt plant could eventually be achieved in 5 to 7 years and a fraction of the cost for a 300-megawatt unit.
The hyperscalers moved on this thesis fast. Microsoft signed a power purchase agreement to restart a unit at Three Mile Island to serve its data center load, an existing reactor, not an SMR, but a clear signal of direction. Amazon Web Services invested in X-energy, an SMR developer working on the Xe-100 high-temperature gas-cooled reactor design, and also agreed to purchase power from a proposed Energy Northwest SMR project in Washington State. Google signed a deal with Kairos Power to purchase power from its fluoride salt-cooled reactor once plants come online. Oracle announced it is designing a data center campus built around three SMR units. These are not concept papers. They are signed agreements with counterparties who have real capital at stake. Each deal structures differently: equity investments, power purchase agreements contingent on commissioning, and offtake contracts. The common thread is that none of them deliver a single kilowatt-hour today.
For a broader picture of the companies building and powering AI infrastructure, see S4Tips coverage of AI data center stocks.
The Key SMR Developers You Should Know
NuScale Power (ticker: SMR) is the only SMR company to have received a final design approval from the U.S. Nuclear Regulatory Commission for its light water SMR design, a milestone that took years and tens of millions of dollars to achieve. That regulatory milestone is real and meaningful. What is also real is that NuScale lost its anchor customer, the Carbon Free Power Project in Idaho, after rising cost estimates made the project economically unworkable for the rural utility cooperatives that were supposed to buy the power. The company continues to pursue international markets and data center customers, but it has no confirmed construction project underway in the United States as of mid-2026.
Oklo (ticker: OKLO) is a fast fission company that went public via SPAC in 2024. Its Aurora powerhouse design is a microreactor, targeting outputs in the 15 to 50 megawatt range, smaller than conventional SMR definitions but positioned for the same data center market. Sam Altman, the CEO of OpenAI, is the chairman of Oklo. That connection attracted significant investor attention and media coverage, though it has no operational bearing on when Oklo will receive its combined license from the NRC or when it will generate its first kilowatt-hour commercially.
Kairos Power is a private company. It is not publicly traded. Its Hermes demonstration reactor in Tennessee received a construction permit from the NRC in late 2023, making it the first advanced reactor construction permit issued in more than 50 years in the United States. The Google deal is contingent on Kairos successfully commissioning commercial units, which it targets for the early 2030s.
X-energy, backed by Amazon, is also private. It is developing the Xe-100 pebble-bed high-temperature gas reactor, a design type with a long international development history but no completed commercial plant in the U.S.
Westinghouse Electric, which is privately held after being acquired by private equity, is developing the AP300, a scaled-down version of its proven AP1000 reactor design. The AP1000 has actually been built: units at Vogtle in Georgia are operating. That construction heritage is a meaningful differentiator. Westinghouse is not a public stock play directly, but its parent companies and partners may offer indirect exposure.
GE Vernova (ticker: GEV), the energy technology spinoff from General Electric, develops the BWRX-300, a boiling water SMR design. GE Vernova is a large diversified energy company, so SMR development is one segment of a much broader business that includes wind turbines, gas turbines, and grid technology. Investors in GEV are not making a pure SMR bet.
Constellation Energy (ticker: CEG) operates the largest fleet of existing nuclear power plants in the United States. It is not an SMR developer, but it is the company behind the Three Mile Island unit 1 restart deal with Microsoft, and it has discussed SMR partnerships for future capacity. Many investors who want nuclear exposure without the development-stage risk buy CEG for its existing operating assets.
SMR Stocks at a Glance
| Company | Ticker | Reactor Type | Public / Private | Key Hyperscaler Link |
|---|---|---|---|---|
| NuScale Power | SMR | Light water SMR | Public (NYSE) | NRC design approval secured; pursuing data center clients |
| Oklo | OKLO | Fast fission microreactor | Public (NYSE) | Sam Altman (OpenAI) as chairman; data center focus |
| GE Vernova | GEV | BWRX-300 (BWR SMR) | Public (NYSE) | Diversified; SMR is one segment of broader energy business |
| Constellation Energy | CEG | Existing large nuclear fleet | Public (Nasdaq) | Microsoft Three Mile Island restart deal |
| Kairos Power | Private | Fluoride salt-cooled reactor | Private | Google signed power purchase agreement |
| X-energy | Private | Xe-100 pebble-bed HTGR | Private | Amazon / AWS equity investment + power deal |
The Regulatory Timeline That Determines Everything
The single most important variable for any SMR investment is the regulatory timeline, and it is one that most retail investors underestimate.
In the United States, the Nuclear Regulatory Commission (NRC) oversees reactor licensing. The NRC’s combined license (COL) process, which covers both construction and operation, can take several years even for experienced applicants with well-documented designs. The NRC has been working to reform its advanced reactor licensing framework under the ADVANCE Act, signed into law in 2024, which aims to reduce fees and processing times for new reactor designs. The reforms are meaningful but not transformative: a novel reactor design still requires extensive safety review, environmental impact assessment, and public comment periods.
Beyond licensing, there are supply chain constraints that add real delay. High-assay low-enriched uranium, or HALEU, the fuel that many advanced reactor designs require, is not yet commercially available at scale in the United States. The Department of Energy has been working to establish a domestic HALEU supply chain, but the timeline for commercial availability remains uncertain. Any SMR developer whose design requires HALEU is taking on fuel availability risk in addition to licensing risk.
The realistic first-power timeline for most U.S. SMR projects currently in development is the late 2020s for small demonstration units and the early-to-mid 2030s for commercial-scale plants. That is a long time to hold a pre-revenue stock, and it means the investment horizon is measured in years, not quarters.
The NRC’s official reactor licensing portal at nrc.gov is the authoritative source for current license application status, public dockets, and regulatory milestones for every active SMR application.
How to Think About Nuclear Energy Stocks More Broadly
Pure-play SMR developers like NuScale and Oklo sit at the highest-risk end of the nuclear investment spectrum. They have no revenue, significant cash burn, and time horizons that depend on regulatory, technological, and commercial variables that are genuinely hard to predict. For investors who want nuclear exposure with lower binary risk, the spectrum is wider than SMR developers alone.
Uranium mining companies benefit from any increased nuclear buildout because more reactors mean more fuel demand, regardless of which reactor design wins. Cameco (ticker: CCJ) is the largest publicly traded uranium producer and trades on reactor fleet expansion expectations globally, not just in the U.S. Uranium enrichment companies, fuel fabricators, and reactor services firms are all part of the nuclear supply chain that would benefit from a genuine SMR buildout.
S4Tips coverage of nuclear energy stocks covers the full spectrum: uranium miners, enrichers, existing fleet operators, and developers, so you can calibrate your exposure to the parts of the chain where you have highest conviction.
The Honest Speculative Risk Picture
The SMR investment case is built on a chain of assumptions, every one of which needs to hold for a pure-play developer to become a valuable business. Regulatory approval must come within a workable timeframe. Construction costs must stay within projections that make power affordable enough for data center buyers. The hyperscaler deals must not be cancelled or renegotiated when timelines slip. Competing technologies, whether grid-scale storage, enhanced geothermal, or next-generation gas with carbon capture, must not solve the baseload problem more cheaply first.
The NuScale Idaho project failure is a cautionary case. It had regulatory approval. It had a customer group. And when construction cost estimates rose materially, the economics collapsed, the customer exited, and the stock sold off severely. That was not a regulatory failure or a technology failure. It was an economic failure, which may be the hardest kind to model in advance.
Oklo has no completed reactor and no operating license. Its current valuation reflects significant optimism about what the company might become rather than what it has built. That is a speculative premium, not a fundamental one. Investors who buy it at current prices are making a bet on future execution, future licensing outcomes, and future power purchase agreement conversions. That bet might pay off significantly. It might not pay off at all. Both outcomes are plausible.
The hyperscaler backing is real and creates genuine signal: these companies do extensive due diligence before committing capital, and their participation de-risks the demand side of the equation meaningfully. But it does not de-risk the supply side. Building a reactor that has never been built at commercial scale, on budget, on schedule, at a cost per megawatt-hour that pencils for a data center, remains an unsolved engineering and financial challenge. SMR stocks are a long-duration bet on an industry that does not yet exist at commercial scale. Position size accordingly, and treat anything you allocate here as genuinely at risk until a first commercial unit delivers power to a paying customer.
Frequently Asked Questions
What does SMR stand for in SMR stocks?
SMR stands for small modular reactor, a nuclear reactor design that generates up to roughly 300 megawatts of electricity per unit, compared to 1,000 megawatts or more for a conventional large-scale nuclear plant. The “modular” part refers to the ability to factory-manufacture components and deploy multiple units at a single site to scale capacity incrementally.
Are any SMR stocks currently generating revenue?
No commercial SMR is operating in the United States as of mid-2026. NuScale and Oklo, the two primary public pure-play SMR stocks in the U.S., are pre-revenue development-stage companies. GE Vernova and Constellation Energy generate revenue from their existing energy businesses, but not from SMR operations specifically.
Which hyperscalers have made concrete SMR commitments?
Amazon Web Services invested in X-energy and agreed to purchase power from a proposed Energy Northwest SMR project. Google signed a power purchase agreement with Kairos Power. Microsoft signed a deal for power from the restarted Three Mile Island unit, and Oracle announced plans for a data center campus built around three SMR units. All of these are agreements for future power delivery, not operational capacity today.
What is the main regulatory body overseeing SMR approvals in the U.S.?
The U.S. Nuclear Regulatory Commission, or NRC, is the federal agency responsible for licensing nuclear reactors, including SMRs. It oversees the design certification process, the combined license process covering construction and operation, and ongoing safety inspection once plants are operating. The NRC’s ADVANCE Act reforms, signed in 2024, are intended to streamline licensing timelines for advanced reactor designs.
What is HALEU and why does it matter for SMR investors?
HALEU stands for high-assay low-enriched uranium, a fuel enriched to between 5 and 20 percent uranium-235, compared to the less-than-5-percent enrichment used in conventional light-water reactor fuel. Many advanced SMR designs, including some fast reactors and high-temperature gas reactors, require HALEU. There is currently no commercial-scale domestic production of HALEU in the United States, which means developers whose designs require it are taking on fuel supply chain risk in addition to licensing and construction risk.
How is an SMR stock different from a uranium mining stock?
An SMR developer is betting on building and operating a new type of reactor. A uranium miner is betting on the price of the fuel those reactors would consume. Both benefit from nuclear expansion, but they carry very different risk profiles. Uranium miners have operating businesses, producing assets, and revenues tied to commodity prices. SMR developers are largely pre-revenue and their timelines extend across multiple years of regulatory and construction work before any cash flows.
Daniel Reyes is a markets writer for S4Tips covering the AI infrastructure and semiconductor supply chain. He focuses on the companies that build and power the AI compute stack. His articles are for information only and are not financial advice.