How Much Does a Small Modular Reactor Cost? 2026 Guide

• Short answer: what an SMR really costs
• What an SMR costs to build
• Cost per MWh vs large nuclear and solar
• Why SMR costs overrun
• Case study: the Vogtle cost overrun
• Who actually pays for SMRs
• Are SMRs cost-effective?

Short answer: what an SMR really costs

A small modular reactor likely costs several billion dollars to build, but no honest number exists yet because almost none are running. In the United States, there are no operating commercial SMRs as of mid-2026. So every price you read is an estimate, not a receipt.

The clearest real-world data point is a canceled project. The NuScale Carbon Free Power Project, planned with the Utah Associated Municipal Power Systems (UAMPS), had a construction estimate near $9.3 billion for about 462 megawatts before it was scrapped in 2023. That works out to roughly $20,000 per kilowatt of capacity.

Newer designs promise to be cheaper once factories build them at scale. But that lower price is a target, not a proven cost. Be skeptical of any confident SMR price. The honest answer is: expensive, unproven, and rising. For how this connects to power demand, see our guide to nuclear power and data centers.

What an SMR costs to build

The best public build cost comes from NuScale's failed Idaho project, which climbed to about $9.3 billion before cancellation. The plan was six 77-megawatt reactors near Idaho Falls, totaling 462 megawatts. That single case shows how fast estimates can move.

NuScale first told customers in 2021 it could sell the power for about $58 per megawatt-hour. By early 2023, that target price had jumped to roughly $89 per megawatt-hour, a 53% rise. The construction estimate rose from about $5.3 billion to $9.3 billion over the same stretch, a 75% jump. UAMPS and NuScale ended the project in November 2023.

For a broader benchmark, the International Energy Agency (IEA) estimated in 2025 that SMRs cost around $10,000 per kilowatt to build in Europe. By comparison, the IEA put traditional large nuclear near $6,600 per kilowatt. So the "small" reactor can cost more per unit of power, not less.

Key build cost numbers:

• NuScale project total: about $9.3 billion for 462 MW (NuScale and UAMPS, 2023).
• NuScale overnight cost: roughly $20,000 per kilowatt at cancellation.
• IEA 2025 SMR estimate: around $10,000 per kilowatt in Europe.
• IEA 2025 large nuclear: around $6,600 per kilowatt.

These are estimates from different places and years, so they do not line up perfectly. They all point the same way: SMRs are costly, and small size does not guarantee a small price.

Cost per MWh vs large nuclear and solar

On a per-megawatt-hour basis, SMR power looks expensive next to solar and wind. The clearest yardstick is the levelized cost of energy, which spreads a plant's lifetime cost over the power it makes. The 2025 report from Lazard gives useful ranges, and nuclear sits near the top.

Here is how the main options compare on an unsubsidized basis, per Lazard's 2025 analysis:

• Utility-scale solar: about $38 to $212 per megawatt-hour.
• Nuclear: about $141 to $220 per megawatt-hour.
• NuScale SMR target price: about $89 per megawatt-hour, but only with heavy subsidies (NuScale and UAMPS, 2023).

Lazard found that solar and wind remained the cheapest new power in the United States on an unsubsidized basis. Nuclear, including SMRs, sat far above them. Even the NuScale $89 figure was propped up by federal support, so it is not a fair apples-to-apples number.

Solar has one real weakness: it does not run at night without storage. That is the argument SMR backers use for round-the-clock nuclear power. But adding batteries to solar is often still cheaper than building a reactor, and battery prices kept falling in 2025 per Lazard. The cost gap is wide, and it favors renewables today.

Why SMR costs overrun

SMR costs overrun mainly because these are first-of-a-kind machines with no factory line behind them yet. When you build the very first version of anything complex, you pay for every surprise. NuScale's jump from $5.3 billion to $9.3 billion is a textbook example of that risk.

Several forces push the price up:

• First-of-a-kind penalty: the first units cost far more than later copies. The U.S. Department of Energy and the IEA both expect real savings only after many reactors are built.
• Supply chain gaps: few factories make reactor parts, so buyers face high prices and long waits.
• Special fuel (HALEU): many new designs need high-assay low-enriched uranium. Right now, mainly Russia and China make it at scale, and the United States banned Russian uranium imports in 2024. Building a domestic supply adds cost and delay.
• Financing: reactors take years to build. Interest piles up the whole time, which can quietly double the sticker price.

Inflation in materials and labor made all of this worse from 2021 to 2023. Promised cost savings from mass production remain unproven, because the mass production has not happened. For the safety questions behind these designs, see whether small modular reactors are safe.

Case study: the Vogtle cost overrun

The Vogtle 3 and 4 reactors in Georgia are the clearest warning about nuclear cost overruns. They are large reactors, not SMRs, but they show what happens when a first-of-its-kind U.S. build meets real-world delays. The pattern is exactly what SMR skeptics worry about.

The two Westinghouse AP1000 units were first expected to cost about $14 billion and start running around 2016 and 2017. Per the U.S. Energy Information Administration, Unit 3 finally entered service in July 2023 and Unit 4 in April 2024, years late. The final price climbed to more than $30 billion, and some tallies put the full project near $36.8 billion.

Two lessons carry straight over to SMRs:

• Estimates can more than double. Vogtle roughly doubled in cost. NuScale's numbers rose fast too, before its project was even built.
• Delays cost real money. Every extra year adds interest and labor charges, which fall on customers and taxpayers.

The NuScale story followed the same script on a smaller scale. Its target price rose from about $58 to $89 per megawatt-hour between 2021 and 2023, and the build estimate jumped from $5.3 billion to $9.3 billion. Then it was canceled in November 2023. Two different nuclear projects, one shared lesson: early prices are optimistic, and final prices are not.

SMR backers say factory-built reactors will avoid the Vogtle trap by moving work off the construction site. That is the theory. Vogtle is the evidence we actually have, and it is not reassuring for anyone promised cheap nuclear power.

Who actually pays for SMRs

The public pays for SMRs, through both taxes and electric bills, long before any reactor makes power. This is the part backers rarely lead with. Public money is the reason the NuScale price looked as low as it did.

Taxpayers fund SMRs through federal programs. The Department of Energy had spent more than $1.2 billion on SMRs by 2021, issued a $900 million solicitation in March 2025, and awarded $800 million in December 2025 to the Tennessee Valley Authority and Holtec. The Inflation Reduction Act adds a production tax credit worth roughly $25 to $30 per megawatt-hour, or a 30% investment tax credit, per the Department of Energy.

Ratepayers pay too. In many states, "construction work in progress" rules let utilities charge customers for a reactor while it is still being built, before it makes any electricity, and even if it never does. That shifts the risk from the company to households. It is a big reason data centers eye nuclear power, since the public can absorb the cost. See how this shows up on bills in our explainer on whether data centers raise electric bills.

Federal loan guarantees add another layer of public backing. For the large Vogtle reactors, the Department of Energy finalized about $12 billion in loan guarantees across several rounds. A loan guarantee means the government promises to repay lenders if the project fails. So taxpayers stand behind the debt, even when a private utility runs the plant.

Stack these supports together and the picture is clear. A single SMR can lean on DOE grants, a roughly $25 to $30 per megawatt-hour production tax credit or a 30% investment tax credit, loan guarantees, and construction-work-in-progress charges on customers. Each tool moves risk off the developer and onto the public. When you read a low SMR price, ask how much of it the public quietly pays.

Data-center operators like the idea of dedicated reactors, and some tech firms have signed deals with nuclear companies powering AI. But the subsidies and ratepayer charges mean the wider public still shares the bill, even for private computing.

Are SMRs cost-effective?

Based on the evidence so far, SMRs are not cost-effective today, and that is the honest verdict. The one flagship U.S. project collapsed on cost, no commercial SMR runs in the country yet, and every price still relies on estimates and subsidies. That is a weak foundation for a "cheap power" claim.

The case for SMRs rests on a future that has not arrived. Supporters argue that factory-built reactors will get cheaper with volume, the way solar panels did. That could happen. But solar proved it with millions of installed panels and falling prices. SMRs have not delivered a single competitive commercial unit at scale.

For now, the numbers favor solar, wind, and storage on price, per Lazard's 2025 analysis. SMRs may still matter for round-the-clock, low-carbon power, especially if the public keeps funding them. Just be clear-eyed: the low prices are promises, the high prices are real, and you help pay either way. Explore where the demand comes from on our AI grid impact hub, check the data center map, and see how communities push back at fighting back.