Fusion energy has attracted more than $700 million in annual U.S. federal funding and billions more in private capital, all premised on the idea that it will one day deliver cheap, limitless clean power. A pair of papers published in Nature Energy in spring 2026 challenges that premise directly, arguing that the two dominant fusion designs will almost certainly fail to match the cost trajectory of solar and wind power.
The reason is not physics. It is economics.
The learning-rate gap at the heart of the debate
Every energy technology gets cheaper as more of it gets built. Economists measure this through “experience rates,” the percentage by which cost falls each time cumulative production doubles. Solar panels have achieved experience rates near 24 percent over the past two decades. Onshore wind sits around 15 percent. These steep curves are why renewable electricity is now the cheapest new power source in most of the world.
Many fusion cost models assume their technology will follow a similar path, plugging in experience rates of roughly 8 to 20 percent, according to an open-access analysis in Nature Energy. However, these figures are modeled estimates drawn from analogies with other large energy technologies, not empirical fusion data, because no commercial fusion plant has ever been built. The same analysis concludes that magnetic confinement reactors (tokamaks) and laser inertial fusion systems, because of their enormous unit size, extreme design complexity, and reliance on custom engineering, would realistically achieve experience rates of only about 2 to 8 percent. That gap between what fusion advocates assume and what historical patterns predict is the core finding.
The argument draws on established research in technology diffusion. Work published in Joule has shown that technologies requiring high design complexity and heavy customization tend to have lower global experience rates. Separate peer-reviewed research has found that large unit size correlates with slower learning and slower market diffusion. Modular renewables sit at the opposite end of that spectrum: small, standardized, and mass-produced, with each factory run and rooftop installation feeding cost data back into the next generation. Fusion reactors, which are closer in character to nuclear fission plants or large hydroelectric dams, do not benefit from that feedback loop at scale.
A companion policy brief in Nature Energy puts the implication bluntly: policymakers should not treat fusion as a core pillar of decarbonization strategies unless designs with fundamentally different characteristics emerge. That is a pointed recommendation from a top-tier journal, aimed at governments that have been increasing fusion budgets while climate deadlines tighten.
The moving target fusion must hit
The competitive bar is not standing still. The U.S. National Renewable Energy Laboratory’s Annual Technology Baseline for 2024 projects continued cost declines for solar, wind, and battery storage through 2050. Under NREL’s moderate scenario, utility-scale solar capital costs fall below $800 per kilowatt by mid-century, and the levelized cost of electricity from combined solar-plus-storage drops well under $40 per megawatt-hour in favorable regions.
Any fusion plant entering service in the 2040s or 2050s would need to beat or match those numbers to justify its capital investment. When NREL’s projections are paired with the modest 2-to-8-percent experience rates the Nature Energy analysis expects for fusion, the arithmetic is unforgiving. A technology that learns slowly is chasing a target that moves fast.
What the research does not settle
No operational fusion power plant exists today. Every cost estimate, optimistic or pessimistic, relies on modeling rather than invoices from a real facility. The Nature Energy analysis builds its case on technology analogies and historical patterns, a defensible method but not a substitute for empirical data from actual fusion construction and operation.
The research also focuses on “dominant” designs, specifically tokamaks like ITER, the $22-billion-plus international project in southern France that has faced repeated delays and cost overruns, and laser inertial systems like the National Ignition Facility, which achieved a net energy gain shot in December 2022. Alternative concepts fall outside the scope. Some private developers are pursuing smaller, modular reactor designs that could, in theory, behave more like the granular technologies that learn quickly. Companies such as Commonwealth Fusion Systems, Helion Energy, and TAE Technologies have collectively raised billions in venture capital on the premise that their compact approaches differ meaningfully from traditional tokamaks.
Whether those designs can deliver net energy gain at commercial scale remains unproven. Without independent cost audits of private fusion efforts, the tension between the Nature Energy findings and industry projections cannot be fully resolved. Investors and policymakers face a familiar problem: how much weight to give peer-reviewed modeling versus company forecasts that have not yet been tested by construction crews and power grids.
Another open question concerns system-level value rather than plant-level cost. Fusion advocates argue that a steady, dispatchable power source could complement variable renewables and reduce the need for storage or transmission upgrades. The Nature Energy work focuses on cost learning at the plant level and does not model those broader grid interactions. A technology with a slower learning curve could still earn a role in a decarbonized grid if it provides unique reliability benefits, but quantifying that value requires integrated modeling that does not yet exist for commercial fusion scenarios.
Where this leaves climate strategy in spring 2026
The strongest evidence in this debate comes from the Nature Energy papers and the supporting literature in Joule. The experience-rate comparison, 2 to 8 percent for fusion versus 8 to 20 percent assumed in promotional projections, is the most concrete number in the discussion. Because these figures are modeled from analogies with other large-scale energy systems rather than measured from actual fusion deployments, they should be treated as the best available estimate, not a settled fact.
Scientific feasibility and economic viability are distinct questions, and answering the first does not resolve the second. The National Ignition Facility proved that fusion ignition is physically possible. That milestone did not tell us what a fusion kilowatt-hour would cost. The Nature Energy research fills part of that gap by stress-testing the cost assumptions that underpin fusion investment cases, and the results are sobering for anyone expecting fusion to be cheap on arrival.
For governments allocating limited climate budgets, the practical calculus is straightforward. Solar, wind, and battery storage are on steep, well-documented learning curves and can cut emissions now. Fusion remains decades from commercial deployment under even optimistic timelines. The Nature Energy findings do not argue for abandoning fusion science. They argue against treating it as a near-term climate solution or assuming it will inevitably become cost-competitive. The most defensible strategy, based on what the evidence supports as of May 2026, is to fund fusion as a long-term research bet while accelerating deployment of the technologies already proving themselves on the grid.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
