Nuclear fusion is extraordinarily expensive today, but not necessarily for the reasons most people assume. The fuel itself is cheap. The real costs pile up in the reactor hardware, the exotic materials needed to contain a miniature star, and the engineering required to keep everything running. Whether fusion can become affordable enough to matter on the grid depends on hitting capital-cost targets that current technology has not yet reached, with recent modeling suggesting early commercial plants will produce electricity at prices well above what wind, solar, and fission already deliver.
Why Capital Cost Dominates the Conversation
When energy economists talk about whether fusion is expensive, they are almost always talking about the upfront cost of building the plant. Fuel costs for a deuterium-tritium reactor are trivial compared to the steel, concrete, superconducting magnets, shielding, and specialized components that go into the machine. A 2023 grid-modeling study estimated that for fusion to capture about 100 gigawatts of capacity on the U.S. grid, roughly the size of the current fission fleet, the capital cost of a plant would need to fall somewhere between about $2,700 and $7,500 per kilowatt of net electric output.1Joule. The value of fusion energy to a decarbonized United States electric grid That is a wide range, and where a given design lands within it depends on the cost of competing technologies, especially fission. The study found that fusion’s competition with other clean-energy sources becomes significant only after it has already displaced fission from the grid.
To put those numbers in context, a modern onshore wind farm costs roughly $1,300 to $1,700 per kilowatt, and a large fission plant can run anywhere from $5,000 to over $10,000 per kilowatt depending on the country and the regulatory environment. Fusion sitting at the low end of its target range would be competitive with advanced fission. At the high end, it would struggle to justify itself unless it offered something fission and renewables cannot, such as firm, carbon-free baseload power without long-lived radioactive waste.
The Magnet Cost Trap
One of the most counterintuitive findings in fusion economics is that stronger magnets do not straightforwardly make reactors cheaper. In theory, a more powerful magnetic field lets you build a smaller device to achieve the same plasma performance, and a smaller device should cost less. In practice, this logic runs into several walls. A systems-engineering analysis published in Nuclear Fusion concluded that increasing the magnetic field does not actually reduce device size once you account for the massive structures needed to handle electromagnetic forces, the thick shielding required to protect the magnets from neutron damage, and the need for divertor technology that does not yet exist at the necessary performance level. On top of that, the cost of high-field magnets scales roughly with the square of the field strength at current price levels.2Nuclear Fusion. Relationship between magnetic field and tokamak size—a system engineering perspective and implications to fusion development
A separate study in Fusion Engineering and Design reached a related conclusion: compact, high-field designs do lower the construction cost, but they do not strongly reduce the cost of electricity because higher neutron wall loading (more neutrons hitting each square meter of the reactor interior) accelerates damage to internal components, driving up replacement costs over the plant’s lifetime.3Fusion Engineering and Design. Impact of plasma, magnet and wall performances on tokamak and helical reactor economics In short, you can shrink the reactor, but the beating it takes from its own reactions gets worse, and the savings on construction get eaten by maintenance. The high cost of superconducting materials and the cryogenic systems needed to keep them cold add another layer of expense that does not disappear with clever design.4SpringerLink (Arabian Journal for Science and Engineering). Superconductivity for Nuclear Fusion: Past, Present, and Future
Maintenance and the Neutron Problem
Fusion reactors do not burn through expensive fuel the way a gas turbine does, but they do burn through their own internal components. The interior surfaces of a fusion reactor, particularly the “blanket” that surrounds the plasma and breeds tritium, absorb a relentless bombardment of high-energy neutrons. This bombardment causes atom-level damage that gradually weakens the material, and the degradation is essentially proportional to the total amount of energy the plant has produced.5arXiv. Valuing maintenance strategies for fusion plants as part of a future electricity grid The more electricity you generate, the faster your blanket wears out, and replacing it is not a weekend job. It involves shutting down the reactor, remotely handling radioactive components, and installing new modules, all of which takes time and money.
This creates a tension at the heart of fusion economics. You want the plant running as much as possible to spread the enormous capital cost over more kilowatt-hours. But running it harder means replacing components more often. Finding the right maintenance schedule is an optimization problem that grid modelers are only beginning to tackle seriously, and the answer depends on details like how long a blanket actually lasts under real operating conditions, something no one has measured in a commercial reactor because no commercial reactor exists yet.
Pulsed Operation and the Duty-Cycle Question
Most current tokamak designs, including the EU-DEMO concept that serves as a reference for many cost studies, operate in pulses rather than continuously. The EU-DEMO baseline, for instance, is designed for a burn duration of about two hours per pulse with a net electric output of 500 megawatts.6Elsevier / Fusion Engineering and Design. Economic study on the key aspects required for a commercial fusion power plant design based on EUROfusion-DEMO to become commercially viable Pulsed operation means the reactor periodically stops producing power while the plasma is re-established for the next pulse. That downtime lowers the plant’s capacity factor, the fraction of time it is actually generating electricity, which drives up the effective cost per kilowatt-hour.
Steady-state operation, where the plasma runs continuously, is an engineering goal that would eliminate this penalty, but it requires plasma-current-drive technologies that consume significant power themselves. Whether the gain from higher uptime outweighs the power consumed to sustain the plasma continuously is an open question and varies by design. Some private companies are betting on alternative confinement concepts that might sidestep pulsed operation entirely, but none have demonstrated net electricity production.
What Would Fusion Electricity Actually Cost?
The standard metric for comparing power sources is the levelized cost of electricity, which rolls up capital costs, fuel, maintenance, financing, and decommissioning into a single price per megawatt-hour over the plant’s lifetime. A 2023 analysis in Energy Policy concluded that for fusion to be competitive on grids beyond 2040, it will likely need to deliver electricity at or below about $80 to $100 per megawatt-hour in 2020 dollars. The same study found this will be hard to achieve for early fusion designs of any size, with modeling showing energy costs greater than $150 per megawatt-hour even after accounting for production learning.7Energy Policy. Can fusion energy be cost-competitive and commercially viable? An analysis of magnetically confined reactors
That $150-plus figure for first-generation plants is not a death sentence for the technology, but it does mean early fusion reactors will not compete on price alone. They would need to offer grid services that cheaper sources cannot, like reliable baseload power that does not depend on weather or energy storage. As more plants are built and designs mature, costs should fall along a learning curve, but how steeply and how quickly is uncertain. A recent extension of the standard fusion-plant costing framework now includes probabilistic modeling that layers materials-price uncertainty, technology-readiness uncertainty, and learning-curve uncertainty into cost projections, acknowledging that single-point cost estimates for a technology that does not yet exist commercially are inherently unreliable.8arXiv. Extension of the fusion power plant costing standard
Decommissioning and Waste
Fusion is often sold as producing no long-lived radioactive waste, and compared to fission that framing is roughly correct. Fusion does not generate spent fuel rods or plutonium. But it does produce radioactive waste. Neutrons streaming out of the plasma activate the reactor’s structural materials, turning ordinary steel and other alloys into low-level radioactive waste. Tritium, the radioactive hydrogen isotope used as fuel, also contaminates internal surfaces and must be managed carefully. A recent radiological study confirmed that future fusion reactors will produce radioactive waste primarily in the form of neutron activation products and tritium, and that robust decommissioning strategies will be necessary.9The European Physical Journal Plus. Radiological characterization for future fusion reactor decommissioning: balancing accuracy, time, and radiation protection
The good news is that most of this waste decays to safe levels within decades rather than millennia, which means decommissioning a fusion plant should be far cheaper and faster than decommissioning a fission reactor. No deep geological repository is expected to be necessary. But “cheaper than fission decommissioning” is a low bar; fission decommissioning is notoriously expensive, often costing billions of dollars per reactor. The fusion industry has not yet produced detailed, independently verified decommissioning cost estimates, because there is nothing to decommission yet. This remains an area where costs could surprise in either direction.
Fuel Is Cheap, but the Supply Chain Is Not Simple
Deuterium, one half of the standard fusion fuel mix, can be extracted from ordinary seawater at negligible cost. Tritium, the other half, is a different story. It does not occur naturally in useful quantities and must be bred inside the reactor itself by bombarding lithium with the neutrons produced by the fusion reactions. This means every fusion plant needs a lithium blanket that can reliably produce and recover enough tritium to sustain the reaction, with some margin to start up the next reactor.
Lithium is not scarce in absolute terms, but the fusion fuel cycle requires lithium enriched in a specific isotope, lithium-6, which makes up only about 7.5 percent of natural lithium. Enrichment adds cost and raises questions about supply-chain readiness and, in some analyses, nuclear non-proliferation considerations.10Joule. Getting the balance right—Considerations of lithium enrichment for the fusion fuel cycle with a focus on nuclear non-proliferation None of this makes fusion fuel expensive in the way coal or natural gas is expensive, but it does mean the “fuel is essentially free” talking point glosses over real engineering and supply-chain costs that show up elsewhere in the budget.
Could Alternative Fuels Change the Economics?
The deuterium-tritium reaction is the easiest to achieve, but it produces fast neutrons that cause the structural-damage and activation problems described above. Some researchers are pursuing reactions that produce few or no neutrons, the most discussed being proton-boron-11 (p-B11). The appeal is straightforward: if you eliminate the neutrons, you dramatically simplify the reactor. No breeding blanket, far less activation waste, potentially cheaper materials, and a simpler path to decommissioning. The fuel itself, hydrogen and boron, is abundant and non-radioactive.11PubMed. Improving the feasibility of economical proton-boron-11 fusion via alpha channeling with a hybrid fast and thermal proton scheme
The catch is physics. The p-B11 reaction requires far higher temperatures than deuterium-tritium and has a much smaller reaction rate even at those temperatures. Achieving ignition, the point where the fusion reaction sustains itself without external heating, is thought to be barely possible, with fusion power exceeding radiation losses by only around 3 percent under standard conditions. Researchers have proposed techniques like “alpha channeling,” which redirects energy from the reaction products back into the fuel protons, and modeling suggests this could reduce the required energy confinement time by a factor of roughly three to seven depending on the approach.11PubMed. Improving the feasibility of economical proton-boron-11 fusion via alpha channeling with a hybrid fast and thermal proton scheme The startup HB11 Energy is pursuing a laser-driven approach to this reaction, arguing that if net energy gain can be achieved, the engineering simplicity of an aneutronic reactor would yield a dramatically simpler and safer plant.12Journal of Fusion Energy. HB11—Understanding Hydrogen-Boron Fusion as a New Clean Energy Source That “if” is doing heavy lifting. The physics challenges are severe enough that most mainstream fusion programs remain focused on deuterium-tritium for the foreseeable future.
Selling Heat Instead of Electricity
One way to change the cost equation is to change what you are selling. A fusion reactor produces enormous amounts of heat, and converting that heat to electricity through a turbine cycle throws away a significant fraction of the thermal energy. Industrial processes like steelmaking, cement production, and chemical manufacturing need high-temperature heat directly, and buying it from a fusion plant could be cheaper than generating electricity and then converting it back to heat.
A recent analysis explored this idea and reached a striking conclusion: in a highly electrified future grid, a fusion plant acting as a heat source effectively works like a giant heat pump with a coefficient of performance greater than one at temperatures much higher than any competing technology can reach. At sufficiently high temperatures, heat from a low-cost fusion plant would be cheaper than using grid electricity regardless of what happens to electricity prices.13arXiv. Fusion for high-value heat production This matters because industrial heat is responsible for a large share of global carbon emissions, and there are few clean alternatives at the temperatures some industries require. If fusion’s electricity is too expensive to compete with solar and wind, its heat might still find a market willing to pay a premium.
Licensing, Regulation, and Hidden Costs
The regulatory framework for fusion is still being written in most countries. Historically, fusion research facilities have been regulated under nuclear rules designed for fission reactors, which impose requirements around spent-fuel management and proliferation risks that do not map cleanly onto a technology that produces no chain reaction and no fissile material. Several countries, including the United Kingdom and the United States, have begun developing fusion-specific regulatory pathways that treat fusion more like an industrial radiation source than a nuclear reactor. How quickly and predictably these frameworks mature will have a direct impact on development costs. Delays in licensing add years of interest payments on borrowed capital, and for a technology where capital costs dominate, financing costs can easily add 20 to 40 percent to the final price of electricity.
The private fusion industry, which now includes dozens of startups backed by billions in venture capital, is keenly aware of this. Companies like Commonwealth Fusion Systems and TAE Technologies have lobbied for streamlined regulatory processes, arguing that applying fission-era rules to fusion would impose unnecessary costs and delays. Whether regulators agree depends partly on unresolved technical questions, like how much tritium a plant will handle and what happens during off-normal events, that will only be answered by building and operating pilot plants.
Where the Uncertainty Really Lives
The honest answer to whether fusion is expensive is that nobody yet knows with confidence, because nobody has built a commercial fusion power plant. Every cost estimate in the literature is a projection based on engineering models, analogies to fission, and assumptions about technologies that have not been demonstrated at scale. The range of projected costs is wide enough that fusion could plausibly end up anywhere from “comparable to offshore wind” to “too expensive to matter.”
The factors most likely to determine where fusion lands on that spectrum are not the glamorous physics breakthroughs that make headlines. They are the mundane engineering questions: Can blanket modules last long enough to keep maintenance costs manageable? Can superconducting tape be manufactured cheaply enough at scale? Can the tritium fuel cycle be closed reliably without leaks? Can plants be built on a predictable schedule without the cost overruns that have plagued large fission projects? The answers will come from building things and seeing what breaks, which is exactly what the next generation of fusion pilot plants is designed to do.