pith. sign in

arxiv: 2606.26536 · v1 · pith:NKLNLXSBnew · submitted 2026-06-25 · 💰 econ.GN · q-fin.EC

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

Stefania B\"ohnlein , Fanny B\"ose , Christian von Hirschhausen , Claudia Kemfert , Alexander Wimmers This is my paper

Pith reviewed 2026-06-26 02:28 UTC · model grok-4.3

classification 💰 econ.GN q-fin.EC
keywords fusion power costsLCOE projectionslearning ratesstochastic analysismagnetic confinement fusioninertial confinement fusionmagneto-inertial fusionenergy economics
0
0 comments X

The pith

Projected mature fusion plants show mean LCOE of 110 to 144 USD per MWh, requiring learning rates above 30 percent that appear optimistic against fission benchmarks.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper performs a stochastic analysis on cost estimates drawn from existing literature for three fusion technology lines at varying maturity stages. It calculates expected levelized costs of electricity for mature devices and derives the implied learning rates needed to reach those figures. The resulting values sit well above costs already achieved by fission plants and other comparable technologies, leading the authors to advise policymakers against relying on claims of near-term economic competitiveness from fusion developers.

Core claim

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%. These projected values are rather optimistic when compared to other literature or comparable technologies like fission.

What carries the argument

Stochastic model applied to harmonized literature cost studies across technology maturity levels for magnetic confinement, inertial confinement, and magneto-inertial confinement fusion.

If this is right

  • Mature fusion devices would need sustained cost reductions exceeding 30 percent per doubling of capacity to reach the projected LCOE values.
  • The three technology lines converge on similar mean costs once mature, with ICF slightly lowest and MIF highest.
  • Current projections exceed realized costs for fission and other low-carbon sources, reducing the likelihood of near-term competitiveness.
  • Policymakers should treat developer claims of economic viability with caution given the gap between modeled and observed benchmarks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the optimistic projections hold, fusion could still enter power markets only after decades of deployment experience comparable to early nuclear fission.
  • The analysis highlights the risk that policy support based on these figures may overstate near-term contributions to decarbonization targets.
  • Future work could test whether private-sector cost studies omitted from the sample alter the derived learning rates.

Load-bearing premise

The collected cost studies from the literature are representative of realistic future fusion plants and the stochastic model correctly captures the uncertainty and learning dynamics across maturity levels.

What would settle it

Direct comparison of actual LCOE from the first commercial-scale fusion plants against the paper's calculated means for mature devices, or measured learning rates during initial deployment that fall below 30 percent.

Figures

Figures reproduced from arXiv: 2606.26536 by Alexander Wimmers, Christian von Hirschhausen, Claudia Kemfert, Fanny B\"ose, Stefania B\"ohnlein.

Figure 1
Figure 1. Figure 1: Assumed overnight construction costs for FPPs by confinement type and technological [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
read the original 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.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper conducts a stochastic meta-analysis of published cost projections for three fusion technology lines—magnetic confinement fusion (MCF), inertial confinement fusion (ICF), and magneto-inertial confinement fusion (MIF)—across varying technology maturity levels. It reports harmonized mean LCOE values for mature plants of 114.6, 110.3, and 143.9 USD/MWh respectively, derives implied learning rates exceeding 30%, and concludes that these projections are optimistic relative to fission and other literature, urging policymakers to exercise caution when developers cite economic competitiveness.

Significance. If the data collection, harmonization criteria, and stochastic propagation of uncertainty are robust and transparent, the work supplies a systematic, quantitative synthesis of the fusion cost literature that has been lacking. The explicit comparison of implied learning rates to fission and the policy-oriented conclusion could usefully inform energy-economics debates, provided the underlying assumptions hold.

major comments (2)
  1. [Abstract and Methods (data collection and stochastic model)] The headline LCOE means and >30% learning-rate claims rest on the representativeness of the collected studies and the functional form of the stochastic maturity model. The abstract supplies no counts of studies per technology-maturity bin, no explicit inclusion/exclusion or harmonization criteria (e.g., capacity factor, discount rate, plant size), and no equations for the stochastic process or learning-rate derivation. These omissions make it impossible to verify whether the reported means are load-bearing or sensitive to selection bias, directly undermining the optimism conclusion and the fission comparison.
  2. [Results (comparison to fission and other literature)] The claim that the projected values are 'rather optimistic when compared to other literature or comparable technologies like fission' is presented without a transparent sensitivity table or alternative harmonization scenarios. If the literature sample is skewed toward optimistic projections (as the weakest-assumption note flags), the policy caution loses force; the manuscript must demonstrate that the central result survives plausible changes in study weighting or maturity binning.
minor comments (2)
  1. [Abstract] Abstract: 'distint' should read 'distinct'.
  2. [Methods] The manuscript should include a table listing the number of underlying studies, their publication years, and the specific harmonization adjustments applied to each technology-maturity combination.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive report and the recommendation for major revision. We address the two major comments point by point below, agreeing where transparency improvements are needed and outlining specific revisions.

read point-by-point responses

  1. Referee: [Abstract and Methods (data collection and stochastic model)] The headline LCOE means and >30% learning-rate claims rest on the representativeness of the collected studies and the functional form of the stochastic maturity model. The abstract supplies no counts of studies per technology-maturity bin, no explicit inclusion/exclusion or harmonization criteria (e.g., capacity factor, discount rate, plant size), and no equations for the stochastic process or learning-rate derivation. These omissions make it impossible to verify whether the reported means are load-bearing or sensitive to selection bias, directly undermining the optimism conclusion and the fission comparison.

    Authors: We agree the abstract is too concise and omits key details needed for immediate verification. In revision we will expand the abstract to state the number of studies per technology-maturity bin, summarize the inclusion/exclusion and harmonization criteria (capacity factor, discount rate, plant size, etc.), and briefly describe the stochastic maturity model and learning-rate derivation. The methods section already contains the full equations and criteria; we will add explicit cross-references from the abstract and ensure the stochastic process is presented with the functional form shown. This directly addresses verifiability. On representativeness, the studies were drawn from a systematic literature search; any selection effects are flagged in the existing weakest-assumption discussion, which we will expand with a short table of study counts. revision: yes

  2. Referee: [Results (comparison to fission and other literature)] The claim that the projected values are 'rather optimistic when compared to other literature or comparable technologies like fission' is presented without a transparent sensitivity table or alternative harmonization scenarios. If the literature sample is skewed toward optimistic projections (as the weakest-assumption note flags), the policy caution loses force; the manuscript must demonstrate that the central result survives plausible changes in study weighting or maturity binning.

    Authors: We accept that the fission comparison would be strengthened by explicit sensitivity checks. We will add a new results subsection and accompanying table that reports the harmonized LCOE means under (i) equal versus inverse-variance study weighting, (ii) alternative maturity bin boundaries, and (iii) varied harmonization assumptions for capacity factor and discount rate. The table will also show the implied learning rates under each scenario. Where the central optimism conclusion is robust we will state so; where it is sensitive we will qualify the policy recommendation accordingly. This directly responds to the concern about potential skew noted in the weakest-assumption passage. revision: yes

Circularity Check

0 steps flagged

No circularity: meta-analysis aggregates external cost studies

full rationale

The paper collects published cost studies on fusion technologies at varying maturity levels, applies stochastic modeling to compute mean LCOE values for mature MCF/ICF/MIF (114.6/110.3/143.9 USD/MWh), and derives implied learning rates (>30%) by comparing across maturity bins. These steps use external literature inputs; the outputs are descriptive aggregates and comparisons to fission/other literature rather than self-referential definitions, fitted parameters renamed as predictions, or load-bearing self-citations. No equations or claims reduce the central results to the paper's own fitted values or prior self-citations by construction. This is a standard external meta-analysis with independent content.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No free parameters, axioms, or invented entities are identifiable from the abstract alone; the work relies on external literature studies whose selection and modeling assumptions are not detailed here.

pith-pipeline@v0.9.1-grok · 5766 in / 1086 out tokens · 52706 ms · 2026-06-26T02:28:21.953324+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

8 extracted references · 7 canonical work pages

  1. [1]

    STARFIRE, a Commercial Tokamak Power Plant Design

    “STARFIRE, a Commercial Tokamak Power Plant Design.” Nuclear Engineering and Design 63 (2): 199–231. https://doi.org/https://doi.org/10.1016/0029-5493(81)90044-3. BBk. undated. “Data Basket.” Deutsche Bundesbank. https://www.bundesbank.de/en/statistics/time - series-databases/data- basket/748932!submitdataBasket?mode=its&databasketActionDownload=Go+to+dow...

    work page doi:10.1016/0029-5493(81)90044-3 1978

  2. [2]

    Potential Contribution of Fusion Power Generation to Low -Carbon Development under the Paris Agreement and Associated Uncertainties

    Oxford: Elsevier. https://doi.org/https://doi.org/10.1016/B978-0-12-819725-7.00114-8. Gi, Keii, Fuminori Sano, Keigo Akimoto, Ryoji Hiwatari, and Kenji Tobita. 2020. “Potential Contribution of Fusion Power Generation to Low -Carbon Development under the Paris Agreement and Associated Uncertainties.” Energy Strategy Reviews 27 (2020): 100432. https://doi.o...

    work page doi:10.1016/b978-0-12-819725-7.00114-8 2020

  3. [3]

    The ARIES Advanced and Conservative Tokamak Power Plant Study

    “The ARIES Advanced and Conservative Tokamak Power Plant Study.” Fusion Science and Technology 67 (1): 1–21. https://doi.org/10.13182/FST14-794. Kikuchi, M. 1996. “An Economically Competitive Fusion Reactor.” Fusion Technology 30 (3P2B): 1631–

    work page doi:10.13182/fst14-794 1996

  4. [4]

    Klinger, T., T

    https://doi.org/10.13182/FST96-A11963184. Klinger, T., T. Andreeva, S. Bozhenkov, C. Brandt, R. Burhenn, B. Buttenschön, G. Fuchert, and et al

    work page doi:10.13182/fst96-a11963184

  5. [5]

    Overview of First Wendelstein 7 -X High -Performance Operation

    “Overview of First Wendelstein 7 -X High -Performance Operation.” Nuclear Fusion 59 (11): 1–11. https://doi.org/https://doi.org/10.1088/1741-4326/ab03a7. Kovari, Michael, Matti Coleman, I. Cristescu, and Robert Smith. 2017. “Tritium Resources Available for Fusion Reactors.” Nuclear Fusion 58 (November 2017). https://doi.org/10.1088/1741 - 4326/aa9d25. Koz...

    work page doi:10.1088/1741-4326/ab03a7 2017

  6. [6]

    THE ARIES -I TOKAMAK REACTOR STUDY††This Research Was Supported by the Fusion Cost Analysis Seite 26 United States Department of Energy, Office of Fusion Energy

    “THE ARIES -I TOKAMAK REACTOR STUDY††This Research Was Supported by the Fusion Cost Analysis Seite 26 United States Department of Energy, Office of Fusion Energy.” In Fusion Technology 1990 , edited by B. E. KEEN, M. HUGUET, and R. HEMSWORTH, 253 –57. Oxford: Elsevier. https://doi.org/https://doi.org/10.1016/B978-0-444-88508-1.50033-5. Najmabadi, F., A. R...

    work page doi:10.1016/b978-0-444-88508-1.50033-5 1990

  7. [7]

    Introduction to Sensitivity Analysis

    “Introduction to Sensitivity Analysis.” In Global Sensitivity Analysis. The Primer , 1–51. John Wiley & Sons, Ltd. https://doi.org/https://doi.org/10.1002/9780470725184.ch1. Schaffer, M. J., and W. D. Bard. 1985. “A New Spatial-Axis Stellarator.” Nuclear Fusion 25 (7): 846–51. Schwartz, Jacob A., Wilson Ricks, Egemen Kolemen, and Jesse D. Jenkins. 2023. “...

    work page doi:10.1002/9780470725184.ch1 1985

  8. [8]

    1991; Entler et al

    MCF FOAK 10 0.125 0.159 0.069–0.466 (Miller et al. 1991; Entler et al. 2018; Famà et al. 2024) MCF NOAK 11 0.0457 0.0423 0.029–0.055 (Yamazaki et al. 2009; Yamazaki, Oishi, and Mori 2011; Kozaki, Imagawa, and Sagara 2009) Fusion Cost Analysis Seite 41 MCF 10th-OAK 23 0.115 0.583 0.048–2.916 (Baker et al. 1981; J. Sheffield et al. 1986; J. G. Delene et al....

    1991