Fusion industry headlines tend to focus on similar themes: plasma temperatures hotter than the sun, record-breaking confinement times, and increasing gains towards breakeven. These are extraordinary achievements, and they have rightly dominated the public discourse. But a working fusion power plant requires more than excellent plasma confinement — and several massively underfunded challenges remain in the path to commercialization.
Between the plasma — typically a deuterium-tritium (D-T) fuel mixture — and the power grid lie three largely overlooked problems whose solutions will ultimately determine whether D-T fusion is viable at scale: tritium breeding, tritium extraction, and remote maintenance. This post examines each in turn, surveys the technological advancements to date, and summarizes the gains that must still be accomplished to achieve commercialized fusion.
Why Deuterium-Tritium Fusion?
As of 2025, the Fusion Industry Association counted 53 private fusion companies, the vast majority of which are pursuing reactions between deuterium (D) and tritium (T). Add to that the major public programs to date — ITER, the USA’s NIF, China’s EAST, the UK’s JET— and Deuterium-Tritium is clearly the consensus fuel choice for first-generation commercial plants.
The reason is primarily physics-driven. As seen in the figure below, the D-T reaction is significantly more reactive than other fuel types, which allows for a lower operating temperature at around 150 million °C. The lower plasma temperature and superior reactivity of D-T enable less plasma heating and a smaller reactor vessel for an equivalent power output.
Thermal fusion reactivities vs. Ion Temperature, Wurzel & Hsu, 2022
The trade-off is neutrons. Approximately 80% of the energy in a D-T reaction is carried away by high-energy neutrons, which must be caught and converted to heat. In most reactor concepts, this is accomplished with some form of Lithium-based blanket that lines the reactor walls. This blanket must simultaneously absorb and multiply neutrons, breed tritium to sustain the fuel cycle, extract heat for power generation, and shield the rest of the reactor from radiation damage. The tritium fuel then needs to be reprocessed at unprecedented rates for the fuel cycle to be economically viable. Lastly, the radioactivity caused by high-energy neutron bombardment necessitates radiation-hardened, high-strength, and high-precision robots to conduct regular maintenance in the fusion core. For any other endeavor, these challenges would be a focal point of development effort. For fusion power plants, given the extreme challenge of achieving net energy gain, they are relegated to secondary priority.
Tritium Breeding — Making More Fuel Than You Burn
No experiment has yet demonstrated a TBR greater than one. In fact, the best direct measurement of TBR to date was published in 2025 by Delaporte-Mathurin et al., which achieved a TBR of 3.57x10-4 in a small volume of FLiBe (Li2BeF4) molten salt. MIT’s Plasma Science & Fusion Center is currently working on scaling this experiment to a larger testbed called LIBRA, with the aim of eventually achieving a TBR > 1. The UK Atomic Energy Authority is developing a dedicated test facility under its Lithium Breeding Tritium Innovation (LIBRTI) program. The facility will enable engineering-scale lithium blanket modules to be assembled, instrumented, and exposed to fusion-relevant neutron conditions in order to study tritium breeding and heat extraction. In parallel, a 1.4 MW supercomputer called “Sunrise” will use AI-accelerated tritium breeding simulations to anchor LIBTRI’s experimental results. But there is a large gap in blanket performance still left to cover– particularly given that high TBR is just one of many (often conflicting) design requirements.
Tritium breeding blankets can take a solid or liquid form. Both forms use lithium, which reacts with high-energy neutrons from the fusion reaction to create tritium and helium. Solid breeders generally use ceramic (Li₄SiO₄ or Li₂TiO₃) pebbles in a packed bed formation. Helium gas sweeps through the pebbles to remove heat and tritium. Solid breeder blankets require periodic batch replacements due to neutron damage and lithium depletion. Liquid blankets, on the other hand, are not susceptible to neutron-induced mechanical damage and can be continuously processed and replenished without affecting plant availability. In liquid-blanket reactors, a lithium-containing liquid metal (such as PbLi) or molten salt (such as FLiBe) circulates through closed channels or, in some cases, through open jets lining the vessel walls. Liquid blankets could enable simpler reactor designs in the long-term, but they come with significant R&D challenges. Their multifunctionality as a breeder, shield, and coolant drives stringent material requirements:
Strong neutron multiplication and absorption characteristics to achieve the required TBR
High radiation damage tolerance to enable longer service life
Melting and boiling points that are compatible with the plant’s required operating temperatures and the surrounding structure
High specific heat and thermal conductivity for efficient heat transfer
Low tritium solubility to enable clean fuel extraction from the blanket
Low electrical conductivity to reduce interference from nearby high-strength magnetic fields
Low corrosivity
Abundant and cheap feedstock(s)
Low toxicity and flammability
No blanket material has yet been found to satisfy these requirements simultaneously– more R&D work is needed to develop a viable liquid blanket at scale. This material R&D draws from a database consisting of fission data, small-scale breeder experiments, and neutronics simulations– none of which fully replicate the integrated blanket environment of a fusion power plant. Commonwealth Fusion Systems has been one of the more proactive and public actors in the tritium breeding domain, with its ARC blanket concept shown below. Their smaller demonstrator, SPARC, is slated to begin operations in 2027 and does not include a breeder blanket.
Commonwealth Fusion’s ARC design uses a liquid immersion blanket (LIB), Meschini et al., 2023
Tritium Extraction and Recycling — Closing the Fuel Loop
Breeding tritium in the blanket is only the beginning. The tritium must then be rapidly extracted, purified, separated from deuterium, and returned to the plasma — all within a closed loop that loses as little tritium as possible. This makes the tritium plant — comprising fuel clean-up units, isotope separation systems, and storage — as consequential as the blanket itself.
As mentioned previously, tritium extraction has already been demonstrated within CANDU fission reactors. However, a CANDU reactor produces tritium incidentally at low concentrations, whereas a D-T fusion plant must extract it continuously and return it to the plasma at a rate and purity that no existing tritium plant has ever demonstrated.
Key Tritium Plant Challenges
In steady state, a self-sufficient D-T fusion plant’s rate of tritium production must be greater than or equal to its rate of consumption. If on-site tritium production lags, the plant must outsource large amounts of tritium to maintain smooth operation, which is expensive and can drive higher regulatory burdens.*
A key metric that drives tritium production requirements is the tritium burn efficiency (TBE), or the fraction of injected tritium actually fused. Doubling the TBE will halve the amount of tritium that must travel through the fuel cycle. ITER is expected to burn less than 1% of injected tritium (Abdou et al., 2021). The highest burn efficiencies from inertial confinement experiments are slightly larger, but still in the single digits. Thus, it is imperative that the large amounts of unburnt tritium be quickly recycled to eliminate bottlenecks. While the blanket-bred tritium goes through the full fuel cycle shown below, the purest portion of the unburnt plasma goes through a bypass called Direct Internal Recycling (DIR). This innovation dramatically reduces the required TBR and tritium startup inventory. However, DIR does require real-time isotopic composition monitoring at the pump outlet, since the exact ratio of deuterium to tritium is unknown in the plasma exhaust.
A simplified schematic of tritium flows through each stage of the fuel cycle
The remainder of the fuel must travel through the inner fuel cycle within hours, including the fuel clean-up unit (FCU) and isotope separation system (ISS). The FCU typically employs membranes that are selectively permeable to Hydrogen gases, which are then separated in the ISS through a multi-stage cryogenic distillation process. Chemical separation of bred tritium from the blanket fluid adds a further time constraint upstream of the inner fuel cycle. A 2023 analysis by Meschini et al. establishes five interlocking performance thresholds that must all be met simultaneously for tritium self-sufficiency to be achievable:
Tritium burn efficiency (TBE) >0.5%
A tritium processing time of <4 hours through the inner fuel cycle
Plant availability > 70%
A DIR fraction > 30%
A required TBR < 1.2
Of these, TBE has emerged as the lever with the most cascading effects on the rest of the fuel cycle, and is consequently the subject of new design proposals. Recent theoretical work by Parisi et al.— affiliated with the company Marathon Fusion — proposes that TBE can be raised by an order of magnitude through two complementary mechanisms that increase the fusion productivity of tritium in the core.
These targets are not independent: a low burn efficiency means more unburnt tritium must be recovered and reprocessed per pulse, which could increase the required DIR fraction; a slow tritium plant increases the inventory tied up in the system at any given time, which could force a higher TBR to compensate; and low plant availability compounds these effects by reducing the time available for breeding. Miss any single threshold by a meaningful margin and the required TBR climbs above what blankets can physically provide. ITER has the most mature tritium plant design to date, but does not claim self-sufficiency. The project will rely on CANDU-sourced tritium to close the fuel cycle, which is a non-starter for future commercial power plants.
*Regulation is another weighty consideration when assessing the economic potential of D-T fuel. However, regulatory factors lie outside the scope of this posting.
Remote Maintenance — Working in a Radioactive Environment
The same high-energy neutrons that necessitate frequent replacements of irradiated components also prevent humans from replacing them. The D-T fusion reaction releases high-energy neutrons that penetrate reactor structures and induce nuclear reactions that convert stable atoms into radioactive isotopes, causing high-exposure reactor materials to become radioactive over time. Although fusion produces low level radioactive waste as compared with an equivalent fission plant, humans still cannot enter a D-T reactor vessel for years after operation. Every task, from tightening a bolt to replacing a 10-tonne blanket module, must be performed by radiation-hardened robots working in confined spaces with limited lines of sight. And these tasks must be completed as quickly as possible, given the economic consequences of operational downtime. A technoeconomic analysis on tokamaks by Lindley et al. cites capacity factor (determined in large part by maintenance downtime) as the single most influential variable on a plant's levelized cost of electricity. Reactors must therefore be designed with remote handling (RH) considerations in mind.
Key Remote Maintenance Challenges
A D-T plant generally requires several distinct RH systems operating in concert. These systems might include: articulating robotic arms for replacing blanket modules and divertor cassettes; autonomous transporters that shield and convey radioactive components between the reactor and the hot cell; and finally, remote cranes and welding equipment inside the hot cell itself.
There is an inherent tension between optimizing for low RH costs and high plant availability. Key tradeoffs must be considered to balance CAPEX, performance, accessibility, and plant availability. For example, a larger number of ports can improve plant availability by accelerating and simplifying maintenance — but each port is a penetration through the vacuum vessel that reduces tritium-breeding coverage and weakens structural integrity. Conversely, an architecture with few ports may not be able to accommodate as many parallel operations, thus hurting plant availability while also simplifying the reactor vessel design. Other plant concepts like the UK’s Spherical Tokamak STEP have a vertical lid that requires low-resistance joints in the magnet coils but allows for large components to be lifted by an overhead crane directly to the hot cell.
JET, the UK’s Joint European Torus, operated from 1983 - 2023 and was a pathfinder for fusion remote handling. JET’s RH system, pictured below, has demonstrated in-vessel divertor tile replacement and was used as a foundation for ITER and DEMO remote handling designs. ITER is developing the most ambitious RH system ever built, though these systems will not truly be tested until 2039 when deuterium-tritium operations are scheduled to begin. Meanwhile, the robotics powerhouse of China has reached a significant milestone with the completion of their remote-handling test platform. The system includes a blanket-maintenance robot that has demonstrated the ability to move 60-tonne loads with ±3.1 mm positioning accuracy.
Renderings of JET’s two-arm manipulator and boom transporter, Todd 2018
While the recent advancements in RH technology are encouraging, there is significant development work ahead. RH systems must demonstrate an unprecedented level of reliability and repeatability within high-radiation environments. Early RH systems will likely be bespoke to their co-designed reactor geometry, limiting economies of scale; in the longer term, standardization of RH components and reactor interfaces could yield significant cost reductions. Each fusion plant design must balance the upfront cost of RH systems with the lost revenue from maintenance downtime. Not prioritizing maintenance considerations when designing an inherently high-maintenance D-T power plant is a financial liability.
Conclusions
At the current pace of fusion progress, net power operation may be achieved before a viable blanket design exists to capture that power, or a viable tritium plant exists to close the fuel cycle, or a viable remote maintenance scheme exists to minimize downtime. Each of these fields must advance substantially for D-T fusion to become economically competitive in future power markets with low-cost alternatives like natural gas and solar + storage.
Many well-funded fusion companies have not yet published or disclosed detailed plans for the breeding, extraction, and remote handling of tritium. They appear to be deferring these questions to publicly-funded programs like LIBRTI and ITER. And yet delaying tackling the engineering challenges posed by tritium can have sizable impacts to a plant’s levelized cost of electricity (LCOE). For example: a tritium breeding or extraction shortfall forces lower plant availability or relies on externally-sourced tritium, thus increasing LCOE. Similarly, a suboptimal remote maintenance configuration could result in high CAPEX, low reactor performance, or prolonged downtimes, all of which hurt the LCOE.
Plasma gain (Q) alone cannot determine a fusion plant’s cost of electricity. A plant with outstanding plasma performance but poor tritium/RH infrastructure may not be able to produce electricity at an economically viable cost .
At 1cFE, quantifying these gaps is part of the work, and the early signal from our models is that tritium and maintenance assumptions can shift LCOE estimates by as much as the plasma physics does.
The fusion industry faces no shortage of ambition, innovation, or capital. As engineering breakeven draws closer, the focus must shift towards the less glamorous but equally existential engineering challenges that lie between the burning plasma and the electricity grid. Ignition is just the beginning.
Note: updated June 22, 2026 to include reference to Parisi et al. work on boosting TBE
References
Abdou, M., Riva, M., Ying, A., Day, C., Loarte, A., Baylor, L., Humrickhouse, P., Fuerst, T.D. and Cho, S. (2021) ‘Physics and technology considerations for the deuterium–tritium fuel cycle and conditions for tritium fuel self sufficiency’, Nuclear Fusion, 61(1), p. 013001. https://doi.org/10.1088/1741-4326/abbf35
Chinese Academy of Sciences, Institute of Plasma Physics (2025) Remote Handling Test Platform Sets New Benchmark in Fusion Technology. Hefei Institutes of Physical Science. Available at: https://english.hf.cas.cn/nr/ps/202509/t20250923_1055245.html (Accessed: 15 March 2026).
Crofts, O. and Harman, J. (2014) ‘Maintenance duration estimate for a DEMO fusion power plant, based on the EFDA WP12 pre-conceptual studies’, Fusion Engineering and Design, 89(9–10), pp. 2383–2387. arXiv:1412.4008.
Delaporte-Mathurin, R., Goles, N., Dunn, C., Edwards, E., Meschini, S., Segantin, S., Ferry, S., Peterson, E., Whyte, D., Lamere, E., Weaver, C. and Woller, K. (2024) Advancing Tritium Self-Sufficiency in Fusion Power Plants: Insights from the BABY Experiment. arXiv:2412.02721.
Ferry, S.E., Woller, K.B., Peterson, E.E., Sorensen, C. and Whyte, D.G. (2023) ‘The LIBRA Experiment: Investigating Robust Tritium Accountancy in Molten FLiBe Exposed to a D-T Fusion Neutron Spectrum’, Fusion Science and Technology, 79(1), pp. 13–35. https://doi.org/10.1080/15361055.2022.2078136
Fusion Industry Association (2025) Global Fusion Industry Report 2025. Washington, DC: Fusion Industry Association. Available at: https://www.fusionindustryassociation.org (Accessed: 15 March 2026).
Glasstone, S. and Lovberg, R.H. (1960) Controlled Thermonuclear Reactions. Princeton: Van Nostrand.
Lindley, B., Roulstone, T., Locatelli, G. and Rooney, M. (2023) ‘Can fusion energy be cost-competitive and commercially viable? An analysis of magnetically confined reactors’, Energy Policy, 177, 113511.
ITER Organization (2024) ITER Project Schedule Update. Available at: https://www.iter.org (Accessed: 15 March 2026).
Jassby, D.L. (2017) ‘Fusion reactors: Not what they’re cracked up to be’, Bulletin of the Atomic Scientists, 19 April.
Lundy-Bryan, L. (2024) Nuclear Fusion: A Primer. State of the Future Substack, 20 November. Available at: https://stateofthefuture.substack.com/p/nuclear-fusion-the-state-of-play (Accessed: 15 March 2026).
Meschini, S., Ferry, S.E., Delaporte-Mathurin, R. and Whyte, D.G. (2023) ‘Modeling and analysis of the tritium fuel cycle for ARC- and STEP-class D-T fusion power plants’, Nuclear Fusion, 63(12), p. 126005. https://doi.org/10.1088/1741-4326/acf3fc
Parisi, J.F., Meschini, S., Rutkowski, A. and Diallo, A. (2024) 'Tritium-Lean Fusion Power Plants with Asymmetric Deuterium-Tritium Transport and Pumping', arXiv:2410.05238.
Schwartz, J.A., Ricks, W., Kolemen, E. and Jenkins, J.D. (2024) ‘Valuing maintenance strategies for fusion plants as part of a future electricity grid’, arXiv:2405.01514.
Segantin, S., Testoni, R. and Zucchetti, M. (2020) ‘Neutronic comparison of liquid breeders for ARC-like reactor blankets’, Fusion Engineering and Design, 160, p. 112013. https://doi.org/10.1016/j.fusengdes.2020.112013
Woller, K.B. et al. (2024) ‘Advancing tritium self-sufficiency in fusion power plants: insights from the BABY experiment’, MIT PSFC Report PSFC/JA-24-101.
Insightful! Way to make complex concepts intelligible to the layperson.