Too cheap to meter? A stochastic analysis of projected future fusion costs

Abstract: In recent years, technological developments and activities by private actors have led a reemerged discussion of the potential of nuclear fusion to meet growing global energy demands. So far, however, fusion technologies remain at comparatively low development levels and their deployment in commercial power plants is probably still decades away. Regardless, over the last decades, many cost studies have been conducted that estimate the future cost of potential fusion power plants. But to date, there is no systematic and harmonized assessment of these projections. Therefore, this study conducts a stochastic analysis of future fusion power plant costs for three distint technology lines, magnetic confinement, inertial confinement, and magneto-inertial confinement fusion, including cost assessments of different technology maturity levels. These levels are further assessed to determine projected learning rates for future fusion costs. For mature technologies, mean LCOE are determined at 114.6, 110.3, and 143.9 USD per MWh for MCF, ICF, and MIF devices, respectively. This implies learning rates of more than 30%. We find that these projected values are rather optimistic when compared to other literature or comparable technologies like fission. We therefore urge policymakers to caution when potential fusion developers refer to the potential economic competitiveness of fusion power plants.

• The paper presents a rigorous, probabilistic analysis of fusion power costs using Monte Carlo sampling on a harmonized dataset of 590 records.
• It standardizes diverse techno-economic inputs to compute LCOE and learning rates across MCF, ICF, and MIF, highlighting high sensitivity to discount rate assumptions.
• The study concludes that fusion energy remains economically uncertain, cautioning policymakers against optimistic projections and emphasizing capital cost challenges.

Stochastic Analysis of Projected Future Fusion Costs

Overview and Motivation

This study conducts a comprehensive stochastic evaluation of projected levelized costs of electricity (LCOE) for future fusion power plants (FPPs) across three principal technology lines: magnetic confinement fusion (MCF), inertial confinement fusion (ICF), and magneto-inertial fusion (MIF). The primary goal is to provide a harmonized and systematic treatment of the wide-ranging, heterogeneous techno-economic cost data for FPPs, addressing the limitations of prior deterministic studies which fail to capture the substantial epistemic uncertainty inherent in pre-commercial fusion projections. By integrating a Monte Carlo-based uncertainty quantification on a curated literature dataset, the results offer probabilistic distributions for both LCOE and technology-specific implied learning rates, directly informing the discussion around fusion’s prospective economic competitiveness and sectoral policy strategy.

Data Harmonization and Stochastic Modelling Methodology

The analysis utilizes a harmonized database of 590 records from 56 sources, standardizing all cost figures to constant 2018 USD and normalizing technology readiness levels across confinement concepts and reactor archetypes. For cost parameters—specifically overnight construction cost (OCC) and O&M cost—lognormal or triangular distributional forms are fit to published estimates depending on data coverage and skewness. Other LCOE model inputs, such as fuel cost, plant lifetime, capacity factor, discount rate, and construction time, are uniformly derived from the median of the available literature to avoid overfitting to outliers.

The LCOE and learning rates are computed via Monte Carlo sampling (N=10,000), with output distributions characterizing not only central tendencies (mean, median) but also the range and asymmetry of possible outcomes, indexed by maturity stages (FOAK, NOAK, 10th-OAK, mature NOAK). Sensitivity analysis examines financial scenarios by varying the weighted average cost of capital (WACC) parameter.

Numerical Results and Interpretation

LCOE Projections

For mature FPP configurations (10th-OAK and beyond), mean LCOE values are as follows:

• MCF (10th-OAK): $114.6$/MWh
• ICF (10th-OAK): $110.3$/MWh
• MIF (10th-OAK): $143.9$/MWh

The probabilistic results reveal pronounced uncertainties, particularly for lower-trl (FOAK) installations, with median FOAK MCF values at $206.7$/MWh. As maturation progresses, the distribution narrows and shifts leftwards; however, there remains residual positive skew for all classes except for MIF.

Sensitivity to Financial Assumptions

LCOE outcomes are highly elastic to discount rate assumptions. Reducing WACC from 7% to 3% drives mean/median LCOE values downward by approximately 19–38%, holding other parameters constant. The dominant role of capital cost in LCOE is thus reaffirmed, indicating that economic viability is contingent on both technological advances and access to low-cost financing similar to established large-scale generation assets.

Learning Rates

The stochastic derivation of learning rates—using risk-adjusted FOAK estimates (when required for ICF and MIF)—demonstrates the following means per doubling of cumulative installed capacity:

• ICF: 30.0%
• MCF: 19.5%
• MIF: 20.7%

These values substantially exceed historically observed learning rates in analog fission technologies (which cluster between 2–15% and can even be negative due to project complexity or regulatory drag). The finding aligns with recent meta-analyses that flagged systematic optimism in fusion learning rate assumptions [Tang et al., Nature Energy 2026].

Implications for Fusion Energy Economics

The projected LCOE figures, even in the fully mature scenarios, reside in the upper quantiles of current cost distributions for fission reactors and do not approach the cost domain of contemporary renewables plus storage. The analysis directly contradicts industry narratives of fusion being “too cheap to meter” and supports the assertion that cost claims by some fusion proponents are predicated on overly optimistic or unobservable learning and cost parameter assumptions. These results also highlight the fact that capital costs—both in terms of OCC and financing—are the principal cost drivers and sources of uncertainty, underscoring the importance of robust supply chain and project management innovation beyond mere physics or engineering breakthroughs.

Operational cost projections remain highly speculative, particularly the negligible fuel cost assumption, which presumes the technical feasibility of sustained D-T cycle tritium breeding—a process yet unproven at industrial scales. Reported plant lifetimes and capacity factor estimates parallel those for large-scale fission but are not empirically validated and likely understate the practical challenges associated with component irradiation and replacement logistics.

Policy and Research Outlook

The findings have several actionable implications:

• Caution for Policymakers: Current cost and learning-rate claims should not be used to rationalize large-scale investment or delay continued scaling of established low-carbon technologies. There is no empirical basis to treat fusion as an imminent, economically competitive baseload option in the near-decarbonization horizon (pre-2050).
• Technology Development: Focused investment in R&D should prioritize not just plasma physics or confinement, but manufacturability, modular deployment, and supply chain establishment to counterbalance the inherently high capital intensity and mitigate cost escalation risk.
• Integrated Assessment Models (IAMs): Scenario studies that incorporate fusion as a low-cost dispatchable resource should treat such assumptions as sensitivity cases, not as central projections, given >30% learning rate and sub-$100/MWh LCOE requirements are not theoretically or empirically justified at this stage.

Conclusion

This work provides a rigorous, probabilistic assessment of FPP costs based on the full published range of techno-economic literature, conclusively demonstrating that expectations for “competitive” fusion power are predicated on optimistic, often non-empirical assumptions regarding capital cost reduction and replicable learning. Given the unresolved technical and supply chain uncertainties, FPP LCOEs are currently projected to track or exceed those of fission and systematically diverge from state-of-the-art renewables. The study urges a recalibrated discussion grounded in stochastic, not deterministic, cost mapping, and reiterates the risk of policy distraction via over-reliance on future, as-yet-unproven fusion advances. Robust, transparent uncertainty quantification and empirical realism should form the basis for future fusion-related energy system planning (2606.26536).