Evidence cutoff: 6 October 2026. An analytical paper distinguishing research milestones, legal authority, proposed projects and operating electricity supply.
Nuclear power needs a delivery programme, not a shared label
Italy’s nuclear return will be judged by its ability to turn legal permission into repeatable electricity supply, without borrowing credibility from fusion milestones or modular branding. The Senate’s vote on 23 September 2026 reopened a legislative path; it did not approve a reactor. China’s reported handover of the BEST research campus on 1 October advanced an experiment; it did not establish a generating business. Confusing those decisions would distort industrial procurement, financing and energy-security planning before either programme delivers its intended product. Rome’s immediate task is to establish the institutions and project conditions behind its electricity ambitions. The consequential choice is how to convert Law 169 into an executable programme while keeping research expenditure, commercial construction and operating generation on separate accounts.
Fusion gain cannot underwrite an electricity contract
CCTV’s account of the Burning Plasma Experimental Superconducting Tokamak in Hefei describes main-machine assembly as more than halfway complete, with completion targeted for the end of 2027 and an electricity demonstration around 2030. Global Times repeated that state-media account; repetition supplies no independent technical validation. The Institute of Plasma Physics of the Chinese Academy of Sciences, ASIPP, specifies first plasma in late 2027 and scientific gain Q ≥ 1 before the end of 2030. These are development milestones whose commercial significance depends on what the apparatus subsequently demonstrates.
For BEST, Q measures fusion power against heating power injected into the plasma. It excludes the full electricity consumed by heating equipment, cryogenics, pumps and other systems, and does not establish what a turbine could export after those loads. ASIPP’s later burning-plasma objective near Q ≈ 5 and EUROfusion’s broader ambition above five therefore cannot be translated into a power-purchase commitment. Tritium breeding, component durability and remote maintenance remain part of the engineering distance between an experiment and dependable electricity. A successful scientific programme can remain a net electricity consumer.
The distinction survives comparisons with other named programmes. ITER seeks a burning-plasma demonstration and has no electricity-generation mission; its revised deuterium–tritium campaign is scheduled for 2039. Commonwealth Fusion Systems’ SPARC is a gain experiment, while ARC is a proposed power plant. The UK’s STEP programme targets an electricity prototype for 2040. BEST’s earlier demonstration aspiration measures a different undertaking, not an already available substitute for Italian fission procurement.
Italy’s electricity ambition implies a fleet and a financing structure
The World Nuclear Association reports 2,702 TWh of nuclear electricity in 2025, approximately 9% of global electricity, supplied by fission. At the report’s 6 October 2026 cutoff, PRIS lists 417 reactors “in operation”, with 379.611 GW net capacity, and separately records suspended units. Administrative status is not instantaneous electricity delivery. The existing system’s relevance lies in measured production, rather than a catalogue of future designs.
Italy’s PNIEC scenarios place nuclear electricity at approximately 11–22% in 2050, corresponding to about 8–16 GW. These are scenarios, not construction commitments. Against the nuclear case’s approximately 583.4 TWh annual electricity requirement, the report’s simplified calculation produces 63.1–126.1 TWh a year at an assumed 90% capacity factor. That utilisation rate is an analytical assumption, not an operating result for Nuward, and the calculation does not reproduce every cogeneration provision in the model.
If that capacity consisted entirely of units rated at 400 MWe, the arithmetic would require 20–40 units, subject to verification of their net rating. Across an illustrative fifteen-year commissioning window, the report derives average additions of 0.53–1.07 GW annually. A first plant around 2035 would therefore open the delivery problem rather than complete it. The PNIEC share requires recurring site approvals, manufacturing throughput, fuel deliveries and operating personnel; it cannot be secured by demonstrating that one reactor can be built.
The French Cour des comptes also shows why the financing boundary changes the judgement. Its January 2025 audit reproduces EDF’s Flamanville estimate assessed in March 2024: €18.4 billion in constant 2023 euros for overnight construction, against €22.6 billion including financing. These are dated estimates, not a certified final outturn, and neither is an Italian modular-reactor price. Their relevance is the distinction between constructing equipment and carrying a project through time. Edison–EDF–Nuward has no demonstrated site-specific Italian overnight cost; modularity alone does not establish a lower electricity price.
Factory repetition must earn its advantage in delivered plants
Edison’s declaration of intent with EDF, Nuward and Italian industrial participants, announced at Antibes on 25 June 2026, establishes commercial interest in a European reactor pathway. Construction discussion from 2030 and deployment around 2035 remain aspirations, rather than a final investment decision or licensed programme. The transmission mechanism runs through a completed design, a site, contracts, financing and qualified production. An industrial partnership can organise those inputs; its announcement does not prove that they have been secured.
Nuward’s current proposal is a single pressurised-water fission reactor with electrical output up to 400 MWe. It exceeds the IAEA-style small modular reactor band of approximately 300 MWe per module and is plainly outside the microreactor category. The vendor’s SMR label cannot erase that distinction, while the earlier two-module Nuward concept cannot be treated as the same design. Procurement must examine the current machine’s construction and operating requirements, rather than infer them from the word “small”.
Akademik Lomonosov and HTR-PM demonstrate that modern small or modular fission systems can generate electricity. Their operation provides evidence about those installations, not a mature market for the Edison–EDF–Nuward proposal. A repeat-build advantage becomes commercially meaningful when successive plants demonstrate delivery and availability, not when unrelated regulatory stages are presented as equivalent progress.
Law 169 starts a regulatory clock, not a construction programme
Law 169 of 29 September 2026 enters into force on 15 October. Its ordinary twelve-month decree window therefore reaches 15 October 2027, with a specified possible ninety-day extension for parliamentary scrutiny. The delegation covers generation, fuel activities, waste, legacy decommissioning and fusion research, development and use. It identifies neither an authorised site nor an approved reactor. The immediate institutional output is a workable allocation of rules and responsibilities, not permission to pour nuclear concrete.
Gilberto Pichetto Fratin, Italy’s Minister of the Environment and Energy Security, has indicated first nuclear electricity in 2033–34; the company pathway discussed in the report points towards around 2035. Neither date resolves the required sequence of project approvals and construction. ISIN, Italy’s independent nuclear-safety and radiation-protection authority, already oversees legacy facilities, research, waste and transport. The gap is recent licensing practice for new commercial power reactors.
Law 169’s provisions for siting, consultation and financial guarantees connect the new programme to obligations that survive any technology choice. SOGIN’s decommissioning and radioactive-materials responsibilities remain attached to the closed Latina, Garigliano, Trino and Caorso fleet. A future fleet would add spent fuel, interim storage and eventual disposal requirements. The 1987 and 2011 referendums also remain part of the institutional setting in which siting must proceed. Waste provision and public consent are components of project delivery, not matters that a modular label settles.
Foreign programmes provide evidence, not transferable licences
Ontario Power Generation’s Darlington programme illustrates how much preparation can precede a visible reactor project. The new-nuclear programme began in 2006; the construction application submitted in 2022 led to a permit in April 2025. The BWRX-300 is under construction, but an operating licence is still required and the end-2030 grid connection remains the operator’s goal. Italy can examine that sequence without assuming that a foreign assessment resolves Italian geology, water use, emergency arrangements or consent.
The UK Office for Nuclear Regulation’s assessment of Rolls-Royce SMR began in 2022 and remains in Step 3 at the dossier’s cutoff; generic assessment is not a site operating licence. Meanwhile, ASIPP and EUROfusion collaborate on BEST’s research plan, and Japan–EU JT-60SA supports plasma research without a grid-electricity product. These relationships transmit knowledge and regulatory experience. They do not collapse public research cooperation, vendor qualification and domestic construction authority into one international endorsement.
China’s industrial expansion likewise cannot be imported as an Italian timetable. PRIS records 37 Chinese units under construction, while the World Nuclear Association attributes nine of the eleven construction starts in 2025 to China. Yet that year’s three first grid connections were distributed across China, India and Russia. The distinction between projects started and electricity delivered is precisely what Rome must preserve: a construction stock shows organisational activity, while energy security depends on completed capacity and its subsequent availability.
The next two years will expose the cost of leaving the programme undefined
Over the next twelve to twenty-four months, Law 169 makes regulatory execution the test of Italy’s decision. By the ordinary October 2027 decree horizon, the government must have converted the delegation into rules through which proponents can establish a site, submit a safety case and allocate lifecycle obligations. By October 2028, progress means movement through those stages, rather than repetition of the ministerial electricity date. Neither milestone warrants an expectation of Italian nuclear generation within that interval.
For Edison, EDF, Nuward and their industrial partners, an undefined Italian approval route would leave production commitments and financing dependent on unresolved project conditions. ISIN would face the corresponding task of preparing assessment capacity without a settled new-project framework. SOGIN’s existing decommissioning obligations would continue independently of progress on generation. The cost of inaction would therefore fall first on proponents carrying development uncertainty and on institutions required to turn political permission into assessable projects.
BEST faces a different test over the same period: its targeted end-2027 completion must lead to commissioning and measured experimental performance, without treating the around-2030 electricity aspiration as an established commercial product. Italy’s legislative option and China’s research infrastructure can both acquire value while remaining different undertakings. If Law 169 is left without an executable project pathway, industrial electricity users gain no domestic nuclear contribution from the announced dates. Their exposure is the continued absence of the proposed supply, while the costs and obligations of preparing it still have to be assigned.
Introduction
On 1 October 2026, CCTV reported the handover for use of the research campus of the Burning Plasma Experimental Superconducting Tokamak, BEST, in Hefei, Anhui. Its account described main-machine assembly as more than halfway complete and entering a nested integration phase involving the vacuum vessel, thermal shield, toroidal-field magnets and internal components. Completion by the end of 2027 and an electricity demonstration around 2030 are reported programme targets. Global Times repeated the CCTV account; this is a second publication of substantially the same state-media claim, rather than independent technical validation. Campus handover does not mean that the tokamak has operated, achieved fusion gain or delivered electricity. [1] [2]
Italy’s Senate approved the Pichetto Fratin nuclear enabling bill on 23 September 2026, by 81 votes to 51, with seven abstentions. The resulting Law 169 of 29 September was published on 30 September and enters into force on 15 October. Thus, at this paper’s cutoff, Italy has a promulgated legal framework for delegation, awaiting commencement; it has neither an authorised site nor a licensed new power reactor. The ordinary twelve-month decree window runs from commencement. This milestone concerns the rules under which a later project might proceed. [3] [4] [7]
The operating world is in a different category. WNA reports 2,702 TWh of nuclear electricity in 2025, about 9% of global electricity, generated by fission. Its end-2025 operable capacity is approximately 400 GW. A fusion experiment, an Italian reactor proposal and that existing generating fleet therefore cannot be placed on the same shelf as interchangeable electricity options. Legitimate comparison requires a declared reaction, product, institutional stage and time horizon before evaluating cost, waste or grid usefulness. [10]
The analysis has three substantive pillars: the engineering object called BEST; the operating fission fleet and its small-reactor subset; and the Italian institutional proposition. Targets are identified as targets, company declarations as commercial intent, and Chinese state-media construction or performance claims as such. No national ranking or judgement for or against the Italian law follows from these classifications.
A taxonomy of nuclear energy products
Fission splits heavy nuclei, ordinarily in uranium-based fuel, and sustains a controlled neutron chain reaction. Present power reactors extract heat and normally convert part of it into electricity through a turbine cycle. Fusion joins light nuclei. The near-term magnetic-confinement programmes discussed here principally envisage deuterium–tritium, D–T, reactions; a helium nucleus and a high-energy neutron carry the released energy. A hot plasma, a heat-producing nuclear system and a net-electricity plant are different engineering products. ITER’s explicit exclusion of electricity generation makes this distinction concrete. [18]
A fusion experiment establishes plasma behaviour or tests components. A pilot integrates enough fuel, heat-removal, maintenance and electricity equipment to demonstrate a plant pathway. A commercial plant must additionally achieve dependable operation, a licensed safety case and an economic delivery model. No country operates a commercial fusion electricity plant. BEST is an experiment with a reported electricity-demonstration ambition; that ambition does not change its present category. [12] [18]
For fission, this paper uses the requested IAEA-style power bands: microreactors are roughly below 20 MWe; SMRs have up to about 300 MWe per module; designs above 300 MWe lie outside that small-reactor band. Microreactors are a subset of small reactors, and the boundaries are approximate nomenclature rather than universal statutory thresholds. A plant containing several 77 MWe modules can exceed 300 MWe while each module remains small. Conversely, one 400 MWe reactor is not a microreactor and exceeds the strict SMR band even if its supplier calls it an SMR. “Above 300 MWe” also does not necessarily mean a gigawatt-scale machine. [11] [29] [32]
| Definition box | Meaning and units |
| Scientific gain, Q | Fusion power / external heating power actually injected into the plasma; dimensionless. Installed heating equipment is not necessarily the injected power in a particular shot. |
| Engineering gain and net electricity | Declare the system boundary and averaging interval. A useful electrical gain is gross generated electricity / total electrical input. Net export is gross electricity minus all plant consumption; measured in MWe or MWh. |
| Capacity factor | Actual net MWh over an interval / (net nameplate MWe × interval hours); dimensionless or %. Availability and pulse duration are different measures. |
| Overnight cost | Construction capital cost as if built instantaneously, excluding financing during construction; currency/kWe or a total. State constant-currency year, inclusions and capacity basis. |
| Annual energy | TWh/year = net GW × capacity factor × 8.76 for a 365-day year. MWth measures heat; MWe measures electrical power. LCOE is currency/MWh, not a construction price. |
The power balance explains why Q > 1 is insufficient. A schematic relation is net electric power = thermal-cycle efficiency × captured heat − electrical power for plasma heating − all other electrical loads. Heating’s wall-plug efficiency determines how much electricity is needed to inject a given plasma-heating power. Magnets and cryogenics, pumps, current drive and the fuel system consume additional power. A commercial D–T system must also replace its consumed tritium through a breeding and recovery cycle. Neither a value of Q nor a fusion-thermal target specifies those balances. Definitions above concern different boundaries, and are not numerically interchangeable. [18] [19]
Institutional boundaries matter equally. Enabling legislation creates powers to establish rules; a licence authorises a defined activity at a defined project; first nuclear concrete marks a construction stage; operation produces measured energy. Public research infrastructure has scientific outputs and public sponsors. A utility’s project has a proposed buyer, tariffs or power-purchase arrangements, financing obligations and an operating licence. Scientific usefulness and contracted annual MWh are distinct measures of success.
BEST as an engineering object
The ASIPP-led technical research description gives BEST a major radius of 3.6 m, minor radius of 1.1 m, magnetic field of 6.15 T and 50 MW of installed auxiliary heating. It specifies D–T plasma work, scientific breakeven Q ≥ 1, long-pulse technology development and exploration of alpha-particle and burning-plasma physics near Q ≈ 5. Its early milestone is first plasma in late 2027 and Q ≥ 1 before the end of 2030. These are statements of mission and schedule in a prepublished abstract for an IAEA meeting scheduled for 20–23 October 2026; they are not results already presented or achieved. [12]
EUROfusion’s public announcement of the ASIPP–EUROfusion research plan describes a broader Q > 5 ambition. The careful formulation is therefore an initial Q ≥ 1 milestone, a technical burning-plasma objective near Q ≈ 5, and a public programme ambition above five. Reporting only “five-fold gain” obscures both the milestone sequence and the plasma boundary. In a D–T reaction, approximately one fifth of fusion energy is carried by alpha particles: near Q = 5, alpha heating can be comparable to external plasma heating. This is physically significant but does not establish electrical self-sufficiency. [12] [14] [18]
Chinese industry reporting identifies ASIPP as the design lead and Fusion New Energy (Anhui) Co. Ltd as the construction company; the latter is also rendered more briefly in English reports as Fusion Energy (Anhui). Its account of delivered vacuum-vessel and thermal-shield components is industrial reporting, not a technical performance paper. The October campus and four-layer assembly milestone remains a CCTV/Global Times claim. A percentage of assembly completed does not quantify the remaining integration, commissioning or nuclear-operational difficulty. [1] [2] [15]
EAST supplies experience in superconducting tokamak operation, long pulses, control and plasma-facing components. CAS carried Xinhua’s account of a 1,066-second high-confinement plasma result on 20 January 2025. That is a reported plasma-duration achievement: it establishes neither a D–T burning plasma nor net electrical output. BEST is intended to move towards burning-plasma and integrated engineering questions between this experimental base and later Chinese engineering demonstration programmes commonly described as CFETR or CFEDR. Those later names and evolving designs should not be treated as an operating plant or a fixed procurement package. [17] [12]
Xinhua reports a 20–200 MW fusion-power target range. It is a state-media statement of intended fusion heat, not an electrical nameplate. A separate ASIPP scenario abstract models an inductive reference discharge with about 41 MW of fusion power over a 50-second flat-top and studies longer-pulse scenarios. Those calculations are more specific than the broad media envelope, but remain modelled design performance. No stated, validated thermal-cycle efficiency or plant-wide consumption allows this paper to translate either figure into net MWe. [16] [13]
After mechanical completion, BEST must establish reproducible plasma performance, D–T operations and a defensible gain measurement, while managing disruptions, impurities and intense divertor heat loads. Progress towards a power plant also requires neutron-resistant materials, tritium inventory control and breeding, remote replacement of activated components, and a heat-to-electricity system with measured parasitic loads. BEST can investigate some of these questions without proving all of them. A limited electricity demonstration might generate gross power while the facility remains a net consumer; the around-2030 claim does not publicly settle that boundary. [13] [1]
| Fusion programme | Mission and fuel | Electricity scope and dated target |
| BEST — public experiment | D–T tokamak; plasma gain and integrated technology. | State-media electricity demonstration around 2030; no validated net MWe. [1] [12] |
| ITER — international experiment | D–T; 500 MW fusion / 50 MW plasma heating, Q = 10. | No electricity mission. Revised schedule: full magnetic energy 2036, D–T operation 2039. [18] |
| CFS SPARC / ARC — private sequence | SPARC is a D–T gain experiment; ARC is a proposed tokamak power plant. | ARC design targets 400 MWe net, early-2030s company ambition. Simulation and a customer agreement do not establish operation. [19] |
| UK STEP — public prototype | D–T spherical-tokamak prototype with an integrated plant mission. | Electricity prototype targeted for 2040; construction ambition from 2030. [20] |
| Japan–EU JT-60SA — experiment | Hydrogen/deuterium plasma research for ITER and DEMO; no planned tritium campaign in its present mission. | No grid-electricity product. First plasma in 2023. [21] |
| Type One Infinity Two — proposed US stellarator | Power-plant development following experimental engineering work. | 400 MWe proposal, revised from 350 MWe in early 2026; TVA letter of intent, conditional on final approvals and agreements. [22] |
BEST’s public 2030 electricity aspiration is earlier than ITER’s revised D–T campaign, but the deliverables differ. ITER seeks a larger controlled burning-plasma demonstration without a turbine. Private tokamak and stellarator developers pursue plant products on ambitious schedules; none has established commercial fusion generation. Collaboration, mission and the set of unresolved demonstrations explain these differences better than a country ranking. The prudent system-planning interpretation is a research option with potentially significant later benefits, whose first electricity date alone cannot establish bankable annual output.
What exists worldwide that produces energy
The live PRIS world feed retrieved on 6 October 2026 lists 417 reactors “in operation”, 379.611 GW net, and separately 21 reactors in suspended operation, 19.387 GW. Adding those categories gives 438 reactors and 398.998 GW; it does not mean all are currently generating. PRIS also lists 78 units under construction, totalling 81.349 GW net. WNA’s approximately 400 GW end-2025 operable figure is a differently dated, broader snapshot. The datasets must retain their status definitions. [8] [9] [10]
Generation also requires a source label. The live PRIS feed reports 2,635.25 TWh for 2025, whereas WNA reports 2,702 TWh. This retrieval does not resolve the difference in coverage or revisions; it would be unjustified to average them or explain the whole gap by suspended capacity. WNA’s figure and approximately 9% world share are used for its 2025 global account, with the PRIS value disclosed. Capacity, annual energy and the instantaneous generating fleet are separate observations. [8] [10]
| Region (paper’s grouping) | In-operation reactors | Net GW | Under construction | Construction GW |
| Americas | 118 | 114.733 | 3 | 1.665 |
| Europe, excluding Russia | 129 | 120.839 | 6 | 6.955 |
| Russia | 34 | 27.969 | 5 | 4.655 |
| East Asia | 100 | 96.961 | 43 | 47.965 |
| South Asia and Middle East | 34 | 17.255 | 17 | 15.709 |
| Africa | 2 | 1.854 | 4 | 4.400 |
| World total | 417 | 379.611 | 78 | 81.349 |
Source: author’s aggregation of PRIS country feed, retrieved 6 October 2026. Europe here includes Armenia and Belarus, and excludes Russia and Türkiye; Türkiye is grouped with the Middle East. East Asia comprises China, Japan and South Korea. Construction includes countries with no operating units, notably Bangladesh, Türkiye and Egypt. “In operation” is an administrative status, not a capacity-factor assumption or a claim that every listed Ukrainian or Japanese unit is presently exporting electricity. [9]
PWRs and BWRs, both light-water technologies, dominate capacity and electricity. PHWRs, including CANDU and Indian heavy-water designs, remain important regional alternatives. Russia retains RBMK-type graphite-moderated units; the UK retains an ageing AGR subset. Fast-reactor experience includes Russia’s operating BN-600 and BN-800, so “fast” is not synonymous with an unbuilt concept, although many proposed advanced systems remain prototypes or development projects. None of these reactors is fusion. [11] [27]
Large light-water machines dominate kilowatt-hours because numerous mature units combine substantial net capacity with established fuel manufacture, turbine systems, operating procedures and repeat maintenance. Large annual output is a multiplication of scale and utilisation, not an inherent consequence of the word nuclear. WNA reports an average reactor age of about 33 years in August 2026 and an 83.7% fleet capacity factor for 2025. Its 2025 construction-start account records eleven starts, nine in China and two in Russia. China is the principal centre of current expansion, with Korea’s domestic and export programme and Russia’s domestic/export projects also material; the OECD project pipeline is thinner. However, 2025 itself had only three first grid connections, one each in China, India and Russia, rather than a Chinese majority that year. [10]
France illustrates an operating system rather than a proposed technology: RTE reports nuclear generation of 373.0 TWh and 68.1% of mainland French generation in 2025. Italy’s domestic nuclear generation is zero. This comparison concerns generation shares, not all energy consumption or the origin of Italy’s imported electricity. Reactor number and installed GW alone cannot reproduce France’s annual energy contribution. [23] [42]
Recent delivery experience spans a wide interval. The table uses recorded construction start, ordinarily first nuclear concrete, and explicitly separates first grid connection from commercial operation. It excludes prior licensing and development time. The interval therefore cannot be read as the time from a political decision to delivered electricity. [44] [45] [46] [47] [48]
| Project — uranium fission | Recorded construction start | First grid connection | Commercial-operation endpoint | Approximate interval |
| Olkiluoto 3, Finland | 12 Aug 2005 | 12 Mar 2022 | 1 May 2023, PRIS record | 16.6 years to grid; 17.7 to commercial. [45] |
| Flamanville 3, France | 3 Dec 2007 | 21 Dec 2024 | Not populated in current PRIS record; progressive commissioning reported by RTE | 17.0 years to grid. [46] [23] |
| Vogtle 3 and 4, US | Mar / Nov 2013 | Mar 2023 / Mar 2024 | 31 Jul 2023 / 29 Apr 2024 | About 10.4 years each to commercial. [47] [26] |
| Barakah 1 and 4, UAE | Jul 2012 / Jul 2015 | Aug 2020 / Mar 2024 | Apr 2021 / Sep 2024 | About 8.7 / 9.1 years to commercial. [48] |
| Zhangzhou 2, China | 4 Sep 2020 | 22 Nov 2025 | 1 Jan 2026 | 5.2 years to grid; 5.3 to commercial. [44] |
Cost comparison is narrower than schedule comparison because currencies, vintages, financing and project boundaries differ. The January 2025 French audit reproduces EDF’s Flamanville estimate, assessed in March 2024: €18.4 billion in 2023 euros for overnight construction including stated additional items, compared with €22.6 billion including financing; the auditor’s broader full-cost total is €23.7 billion in 2023 euros. The overnight estimate is about €11,300 per kWe when divided by the current PRIS net rating of 1,630 MWe. It is an audited presentation of a forecast, not a certified final 2026 outturn. [24] [46]
For a consistent international range, IEA/NEA’s 2020 study reports prospective new-build overnight estimates in constant 2018 US dollars. Table 3.4a ranges from US$2,157/kWe for Korea to US$6,920/kWe for the Slovak submission; China is US$2,500/kWe, France US$4,013/kWe and the US US$4,250/kWe. These are country submissions for assumed future, generally repeat builds, not audited invoices for Barakah, Zhangzhou, Flamanville or Vogtle. The US$6,041/kWe US figure sometimes extracted from that table is investment cost at a 10% discount rate, not overnight cost. [25]
The retrieved evidence does not provide a defensibly harmonised constant-price overnight outturn for every named project. Olkiluoto contractor totals are not the owner’s fixed-price purchase obligation; Vogtle participant net accounts are not a full project overnight estimate; Barakah financial-close packages include financing and other items; China’s generic NEA submission is not a Zhangzhou project audit. Those quantities are therefore excluded from a common numerical outturn range. Repricing a multiyear nominal aggregate with one end-year inflation multiplier would introduce false precision. This is an evidence limitation, not evidence that the excluded projects share the same cost. [24] [25]
For BEST there is no verified commercial overnight cost per net kWe because there is no validated electrical plant rating. Likewise, an Italian project without a selected site, completed design and financing has no demonstrated Italian overnight price. Smaller unit size may reduce absolute project exposure, while factory repetition might improve delivery; neither establishes lower cost per kWh before construction and operating evidence.
Small and micro, as built rather than proposed
Two frequently cited modern examples actually supply electricity. Russia’s Akademik Lomonosov floating plant has two KLT-40S fission reactors, each about 35 MWe gross and 32 MWe net: 70 MWe gross and 64 MWe net for the plant. Its commercial operation began in 2020. China’s HTR-PM comprises two roughly 250 MWth helium-cooled reactor modules feeding a common turbine, and entered commercial operation in December 2023. Its commonly quoted approximately 210 MWe plant rating is electrical, not the 500 MWth combined reactor heat. [11] [27] [44]
The current PRIS HTR-PM record distinguishes 211 MWe gross, 200 MWe design net and 150 MWe listed net. This paper preserves that discrepancy rather than silently presenting 200 MWe as current net output or assuming a reason for the change. Similarly, PRIS still lists Linglong-1, the Chinese ACP100, under construction, with 125 MWe gross and 100 MWe net design. It is therefore not counted here as a third operating modern SMR. Older small power reactors demonstrate small-scale fission, but do not establish a modern factory-repeat SMR market. No modern commercial microreactor is verified as supplying a civilian electricity grid at the cutoff. [44] [28]
| Design / project | Reaction and module class | Status at cutoff | What the status does not establish |
| Nuward, current design | Uranium-fission PWR, 400 MWe; above strict SMR band | In pre-licensing / design development | No Italian licence or construction authority. Earlier 2 × 170 MWe design is a different version. [29] |
| Rolls-Royce | Uranium-fission PWR, 470 MWe; above strict SMR band | In pre-licensing: UK GDA Step 3 | Generic assessment is not a site operating licence. [30] [11] |
| GE Hitachi BWRX-300, Darlington | Uranium-fission BWR, 300 MWe | Under construction; construction licence granted April 2025 | Operating licence still required; end-2030 grid date is OPG’s goal. [31] |
| NuScale US460 | Uranium-fission integral PWR, 77 MWe/module; up to six modules | Licensed/approved at design level: NRC standard design approval, May 2025 | Design approval is not a licensed construction or operating project. [32] |
| Westinghouse AP300 | Uranium-fission PWR, 300 MWe | In pre-licensing / development | AP1000 heritage does not itself license AP300. [33] [11] |
| X-energy Xe-100 / Long Mott | Uranium-fission HTGR, 80 MWe/module | Formal construction-permit review; not yet shown licensed in retrieved evidence | Environmental assessment approval is not a construction permit. [34] |
| Kairos Hermes 1 / Hermes 2 | Uranium-fission salt-cooled demonstration reactors; TRISO fuel | Hermes 1 under nuclear construction; Hermes 2 construction permits and 2026 groundbreaking | Hermes 1 is non-power. Hermes 2’s revised 50 MWe proposal must be distinguished from original permitted configuration. [35] [36] |
| Pele / DOE microreactor concepts | Uranium fission; Pele proposes 1.5 MWe; MARVEL about 0.01 MWe | Prototype manufacture/test infrastructure; other concepts range from pre-licensing to paper | Military or laboratory authorisation does not establish a civilian commercial fleet. [37] [38] [39] |
The requested five-stage vocabulary is useful only with its object specified. A design approval, an environmental finding, a construction permit and an operating licence are not one legal event. Long Mott is already in formal permit review, so calling it merely “paper” would understate progress and calling it “licensed” would overstate it. The table retains that intermediate stage explicitly. Italy’s discussed candidates remain proposals or pre-licensing designs in the Italian jurisdiction; progress abroad supplies evidence to review, not an Italian authorisation.
Kairos warrants particular version control. Its April 2026 company announcement describes Hermes 2 as an eventual up-to-50 MWe electricity supplier. The original permits described in INL’s regulatory update cover a two-reactor demonstration configuration, each 35 MWth. A revised electrical target must not be substituted for the original permitted thermal specification without the relevant amendment. Groundbreaking also does not itself prove completion of nuclear construction. The sources establish substantial development progress, not operation. [35] [36]
US microreactor programmes show the same boundary. DOE’s MARVEL project envisages approximately 85 kWth and 10 kWe, with criticality and electricity milestones still ahead. DOME is a test bed capable of accommodating reactors up to 20 MWth; its thermal test limit is not a 20 MWe grid rating. BWXT reports Pele core stacking completed in June 2026 and further delivery and integrated testing still to come, for a proposed 1.5 MWe military demonstration. Factory production, fuel qualification, transport, staffing and civilian licensing remain material issues for wider deployment. [37] [38] [39]
The Italian proposition, reconstructed
Law 169 delegates a comprehensive reorganisation rather than approving a reactor. Articles 1 and 2 encompass nuclear generation, fuel manufacture and reprocessing, legacy decommissioning, radioactive waste and spent fuel, and fusion research, development and use. Article 3 expressly includes modular and advanced technologies, refers to IAEA definitions, and provides principles for authorisation, siting, financial guarantees and consultation. Ordinary decrees are due within twelve months of 15 October 2026, subject to a specified possible ninety-day extension for parliamentary scrutiny. The statute itself identifies no approved site or reactor. [4] [5] [6]
The June Quirinale/Antibes political and industrial sequence should be read through the actual agreements. Edison’s 25 June Antibes statement records a declaration of intent with EDF, Nuward and Italian industrial participants to develop a European reactor pathway, aiming at deployment around 2035. Construction talk from 2030 and first Italian electricity around 2035 are commercial aspirations, not a final investment decision or licensed schedule. Pichetto Fratin’s separately reported 2033–34 date is a ministerial target. It does not override a designer’s timetable or the sequence of future licences. [40] [41]
The current Nuward product page describes one 1,150 MWth reactor with electrical output up to 400 MWe and a heat-supply option. That is modular pressurised-water fission of a few hundred megawatts. It is not a microreactor; under this paper’s strict band it is also above SMR size, despite the vendor’s label. The earlier two-module, 340 MWe Nuward concept must not be carried forward as if unchanged. No retrieved named microreactor agreement supports treating Italy’s Edison–EDF–Nuward proposition as a microreactor project. [29] [40]
The 1987 and 2011 referendums frame public legitimacy and institutional continuity, but are not interchangeable votes on today’s designs. Italy’s former Latina, Garigliano, Trino and Caorso stations are closed; Garigliano had ceased generation before 1987. SOGIN’s legacy task is decommissioning and management of associated radioactive materials and waste, not operation of a new generating fleet. A future reactor programme must resolve additional spent-fuel inventories, interim storage and eventual disposal alongside that inherited obligation. [42]
ISIN already exists as Italy’s independent nuclear-safety and radiation-protection authority. It oversees legacy nuclear facilities, research, waste, transport and associated safety matters. The relevant gap is therefore not an absence of a regulator or all domestic nuclear skills: it is the lack of a continuous recent practice licensing, constructing and operating new domestic commercial power reactors after the phase-out. Restoring that practice requires retained technical competence, recruitment and training, assessment capacity and a clear allocation of responsibilities under the decrees. These are institutional inferences from the current mandate and decades-long generation gap, not a judgement that ISIN is incapable. [43] [42]
| Conditional stage | Earliest programme aspiration / boundary | International evidence and implication |
| Commencement and decrees | 15 Oct 2026 commencement; ordinary decree horizon 15 Oct 2027, possible statutory extension | This is rule-making. Neither a site nor a reactor automatically becomes licensed. [3] [4] |
| Design, site and applicant review | After applicable rules; work can overlap, but no Italian approval date established | Darlington benefited from an earlier environmental/site process; its 2022 construction application led to an April 2025 permit. [31] |
| Financing and first nuclear concrete | Industry discussion from about 2030, conditional on licences, site, contracts and investment | A vendor’s factory or assembly duration excludes much preconstruction work. Groundbreaking and nuclear concrete must be reported separately. [29] [40] |
| Fuel loading, grid connection and commercial operation | Minister: 2033–34; company aspiration: about 2035 | Recent fission projects took roughly five to eighteen years from nuclear construction to grid or commercial endpoint; no automatic Italian duration follows. [41] [44] [45] [47] [48] |
Two licensing analogues caution against compressing the entire programme into the nominal build period. Rolls-Royce’s UK generic assessment began in 2022 and remains in Step 3 at the cutoff, without being a site licence. Darlington’s new-nuclear programme began in 2006, much earlier than its current construction permit. Neither proves how long Italy must take, but both show why mature institutions, a complete application and preparatory work matter. A design review completed elsewhere cannot decide Italian geology, water use, emergency arrangements or community consent. [30] [31]
Fusion is explicitly within the Italian statute’s scope, so it is legitimate to discuss research and future regulatory preparation. It does not follow that an Italian commercial fusion plant is authorised, funded or available to buy. BEST remains Chinese research infrastructure. The distinction between a research line and a power-purchase project must survive any reference to international collaboration. The paper reaches no conclusion for or against the law; it identifies the conditions separating legislation from measured Italian MWh. [4] [5]
Comparison only after classification
The table compares four different objects, rather than four competing turnkey products. Future success is an evaluative criterion, not a forecast. Ratings are electrical unless explicitly stated; the absence of comparable cost evidence is preserved. Sources: BEST [1] [12] [13] [16]; Italy [3] [4] [29] [40] [43]; Flamanville [23] [24] [46]; Akademik Lomonosov [11] [27].
| Object | Reaction | Electrical output | Readiness / stage | Earliest grid electricity | Fuel and waste | Overnight-cost evidence | Regulatory position | Success in 2030 / 2040 |
| BEST | D–T fusion | None demonstrated; no validated net rating | Experiment in advanced assembly, state-media report | Around 2030 demonstration target; net export boundary unspecified | D/T; tritium control; neutron-activated structures. Closed breeding cycle not proved | No commercial cost/net-kWe evidence | Chinese research and applicable radiation/nuclear-safety oversight; specific public licence not verified | 2030: measured gain and defined electrical balance. 2040: evidence supporting a later maintainable pilot. |
| Italian Nuward-class proposal | Uranium fission; PWR | Up to 400 MWe vendor design; none in Italy | Proposed; design pre-licensing. One reactor exceeds strict 300 MWe SMR band | 2033–34 ministerial / circa 2035 company targets; conditional | Enriched uranium; spent fuel, activated equipment and disposal obligations | No site-specific Italian constant-price overnight estimate | ISIN presently; allocation/reorganisation under future decrees, plus project authorisations | 2030: licensed, financed project if conditions met. 2040: audited operation and any repeat units, not announcements. |
| Flamanville 3 | Uranium fission; EPR PWR | 1,630 MWe net PRIS rating | Operating / progressive commissioning | First grid connection 21 Dec 2024 | Enriched uranium; spent fuel and decommissioning waste | €18.4bn, 2023 euros, overnight estimate assessed Mar 2024; not final outturn | France’s ASNR; operator EDF | 2030: dependable measured generation. 2040: maintained safety, availability and waste funding. |
| Akademik Lomonosov | Uranium fission; two KLT-40S PWRs | 64 MWe net; 70 MWe gross, plant total | Operating small floating plant | Grid 2019; commercial 2020 | Enriched uranium; spent fuel and activated equipment; shore/logistics obligations | No verified comparable constant-price overnight figure in this paper | Russian nuclear-safety authority Rostekhnadzor; operator Rosenergoatom | 2030: electricity/heat availability and maintenance. 2040: demonstrated lifecycle, refuelling and waste management. |
A reported fusion pulse cannot inherit a fission plant’s capacity factor. Nor can the price of a research campus be divided by its peak fusion heat to obtain a commercial electricity construction cost. For the Italian proposal, a manufacturer’s output specification concerns a possible future machine; it does not constitute installed Italian generating capacity. For the two operating fission examples, success includes audited generation and lifecycle performance, not merely the existence of a design.
System implications
Existing fission offers dispatchable low-carbon electrical energy, generally most economically used at high annual utilisation. “Baseload” describes a frequent operating and economic role, not an inability to vary output: RTE documents increasing French nuclear modulation as the generation mix changes. Grid usefulness depends on planned outages, unplanned common-mode problems, transmission and cooling constraints, reserve provision and integration with other supply and demand resources. A large station’s nameplate rating does not guarantee an equally large firm contribution during every system stress. [23] [25]
Smaller fission units may fit smaller load increments, share site services or supply electricity and heat. High-temperature designs could extend industrial-heat applications, but an HTR-PM turbine demonstration does not by itself establish every proposed industrial process. Heat customers need a temperature, duty cycle and contractual service; electricity customers need reliable net MWh. Factory learning and phased investment remain hypotheses to be tested through a sequence of delivered units, not attributes proved by a rendering. [11] [34]
Fusion’s present system contribution is research information rather than material electricity supply. A successful 2030 pilot or demonstration would still need evidence on availability, fuel self-sufficiency, component life, maintenance duration and capital cost before becoming a repeatable grid option. Treating fusion as a potential post-2040 contributor is therefore a conditional planning horizon, not a claim that no earlier demonstration can occur or that commercial success is guaranteed after 2040. BEST, ITER and private pilots resolve different portions of that uncertainty. [12] [18] [19] [20]
Italy’s PNIEC scenarios are quantitative but not deployment commitments. The parliamentary dossier reproduces a conservative 2050 case of about 8 GW, corresponding to roughly 11% nuclear electricity, and a higher-potential case of 16 GW, about 22%. The model’s nuclear case has electricity requirements around 583.4 TWh/year. The stated capacities also include a cogeneration component, so a simplified all-electric calculation below is an interpretation for scale, not a reconstruction of every model dispatch and heat assumption. [42]
| Transparent calculation | 8 GW case | 16 GW case |
| Annual electricity at assumed 90% capacity factor | 63.1 TWh | 126.1 TWh |
| Share of 583.4 TWh/year scenario requirement | 10.8% | 21.6% |
| Annual electricity at assumed 85% capacity factor | 59.6 TWh | 119.1 TWh |
| Number of 400 MWe units, if all capacity uses that size | 20 | 40 |
| Number of 300 MWe modules, rounded up to meet capacity | 27 | 54 |
| Average additions over a 15-year commissioning window | 0.53 GW/year | 1.07 GW/year |
Calculations use a 365-day year and declared net-capacity assumptions; the vendor’s 400 MWe output must be verified as the appropriate net basis before project modelling. At 90%, one 400 MWe unit would produce 3.15 TWh/year, whereas a 1 GW unit would produce 7.88 TWh/year. Twenty or forty such units over an illustrative fifteen-year window imply about 1.3 or 2.7 completions annually. These are arithmetic implications, not forecasts and not measured Nuward capacity factors.
That scale requires more than a successful first project. It requires recurring licensing, several sites or multiple-unit sites, manufacturing throughput, skilled construction and operating staff, fuel deliveries, waste capacity and sustained financing. China’s present construction stock—37 units and about 40 GW in PRIS—is evidence of what an established large programme can organise, not evidence that Italy can transfer its rates or institutions unchanged. A first Italian unit around 2035 would leave a short interval for the scenario’s repeat-build programme. The comparison concerns production capacity and organisational requirements, without ranking countries or recommending a policy choice. [9]
The PNIEC share must also use the right denominator. Applying 11–22% to present Italian demand would understate the generating scale in a 2050 electrification scenario. Applying a fission capacity factor to BEST would invent an energy product that its research mission has not delivered. Public research outcomes, prospective power-purchase projects and existing fleet TWh therefore remain separate even when all appear in a decarbonisation discussion.
Conclusion
BEST is a state-led fusion experiment entering advanced assembly, with a state-media electricity-demonstration target around 2030 and no commercial electricity product. Its technical plan specifies an early scientific-gain milestone and later burning-plasma work; neither entails net electricity. Italy has published an enabling statute for a possible return to nuclear generation, including modular and advanced fission and an explicit fusion line, but has no authorised new site or plant. At 6 October, the statute has not yet reached its 15 October commencement. [1] [3] [12]
The world’s nuclear kilowatt-hours come from uranium-based fission and related fission fuel cycles, overwhelmingly an established large-reactor fleet with a substantial ageing component. The modern SMR operating subset is thin; the development pipeline is large and institutionally diverse. Akademik Lomonosov and HTR-PM are real generating installations, whereas design approvals, reactor proposals and microreactor prototypes represent other stages. [8] [10] [11] [27] [44]
Comparison is legitimate only after these propositions are established. Research success, legislative success, construction success and operating success answer different questions. The useful common framework is therefore reaction, net energy product, maturity, dated institutional stage, cost boundary, fuel-and-waste responsibilities and evidence of repeat delivery. It permits analysis of future options without converting a scientific target or a political timetable into electricity that already exists.
Source and method notes
All sources were checked against a 6 October 2026 cutoff. The PRIS public statistics and reactor-record feeds were retrieved directly because the website is dynamically rendered. Their date is the retrieval date; administrative status and rating fields are preserved rather than interpreted as real-time production. Source discrepancies are disclosed for global 2025 TWh, HTR-PM ratings and gain milestones. Calculations are labelled as the author’s and use only stated assumptions.
The specified Nuclear Engineering International article dated 6 October 2026 could not be independently retrieved. It has therefore not been cited as read or used to corroborate CCTV. The campus/four-layer claim is attributed to CCTV and Global Times; the designer/builder is attributed to the retrieved industry account; technical objectives are grounded in ASIPP’s prepublished abstracts and EUROfusion. The missing article remains a specific evidence gap in the requested source set.
No harmonised constant-price project-cost outturn was established for every named recent build. The paper accordingly separates the French audit’s constant-2023-euro estimate from NEA’s constant-2018-dollar prospective submissions and excludes incompatible nominal financial packages from a numerical outturn comparison. A paper containing a single confident cross-project cost ranking would exceed this evidence.
[1] CCTV report, reproduced by CLS, 1 October 2026: BEST campus handover. Accessed 6 October 2026.
[2] Global Times, 1 October 2026: BEST campus and assembly milestone. Accessed 6 October 2026.
[3] Gazzetta Ufficiale: Law 169 of 29 September 2026; publication and commencement. Accessed 6 October 2026.
[4] Law 169/2026, Article 1: delegation, procedures and decree deadline. Accessed 6 October 2026.
[5] Law 169/2026, Article 2: scope and regulatory institutions. Accessed 6 October 2026.
[6] Law 169/2026, Article 3: technology, siting, financial guarantees and consultation. Accessed 6 October 2026.
[7] Italpress, 23 September 2026: Senate vote, 81–51 with seven abstentions. Accessed 6 October 2026.
[8] IAEA PRIS: live world statistics, retrieved 6 October 2026. Accessed 6 October 2026.
[9] IAEA PRIS: live country statistics, retrieved 6 October 2026. Accessed 6 October 2026.
[10] World Nuclear Association: World Nuclear Outlook Report 2026, updated 6 September 2026. Accessed 6 October 2026.
[11] IAEA: Small Modular Reactor Technology Catalogue 2024; vendor-contributed design descriptions. Accessed 6 October 2026.
[12] Qian, Kazakov and collaborators: Overview of BEST Research Plan and BEST Missions, prepublished IAEA meeting abstract. Accessed 6 October 2026.
[13] ASIPP: Physics basis of integrated scenarios towards Q ~ 1 in BEST, prepublished IAEA meeting abstract. Accessed 6 October 2026.
[14] EUROfusion, 24 November 2025: release of the BEST Research Plan. Accessed 6 October 2026.
[15] CINIE, 3 August 2026: BEST core equipment; designer and construction company. Accessed 6 October 2026.
[16] Xinhua, 24 November 2025, reproduced by 24horas: BEST 20–200 MW fusion-power target. Accessed 6 October 2026.
[17] Chinese Academy of Sciences, 20 January 2025, Xinhua report: EAST 1,066-second plasma result. Accessed 6 October 2026.
[18] ITER Organization: ITER in a few lines; gain, mission and revised schedule. Accessed 6 October 2026.
[19] Commonwealth Fusion Systems, 4 June 2026: ARC physics basis and electrical target. Accessed 6 October 2026.
[20] UK government, 16 March 2026: STEP prototype objectives. Accessed 6 October 2026.
[21] QST / ITER Japan: JT-60SA mission and fuel; explanatory institutional page. Accessed 6 October 2026.
[22] Type One Energy / TVA statement, 19 September 2025, revised 4 September 2026. Accessed 6 October 2026.
[23] RTE: Annual Electricity Review 2025, Generation; mainland France mix. Accessed 6 October 2026.
[24] Cour des comptes, January 2025: La filière EPR, especially printed pp. 26 and 34–35. Accessed 6 October 2026.
[25] IEA / OECD NEA: Projected Costs of Generating Electricity 2020; Table 3.4a, USD 2018. Accessed 6 October 2026.
[26] Georgia Power, 29 April 2024: Vogtle 3 and 4 commercial operation. Accessed 6 October 2026.
[27] World Nuclear Association: Nuclear Power in Russia; Akademik Lomonosov and fast reactors. Accessed 6 October 2026.
[28] CNNC, 31 March 2025: ACP100 construction, 125 MWe gross design. Accessed 6 October 2026.
[29] Nuward: current product description, 400 MWe single-reactor design. Accessed 6 October 2026.
[30] Office for Nuclear Regulation: Rolls-Royce SMR Generic Design Assessment. Accessed 6 October 2026.
[31] Ontario Power Generation: Darlington SMR, licensing chronology and summer 2026 construction. Accessed 6 October 2026.
[32] NRC, 29 May 2025: Standard Design Approval for NuScale US460. Accessed 6 October 2026.
[33] Westinghouse, 4 August 2026: AP300 development and prospective NRC approval. Accessed 6 October 2026.
[34] X-energy, 18 May 2026: Long Mott environmental assessment and construction-permit application. Accessed 6 October 2026.
[35] Kairos Power, 17 April 2026: Hermes 2 groundbreaking and revised electricity proposal. Accessed 6 October 2026.
[36] Idaho National Laboratory / GAIN: original Hermes 2 construction permits. Accessed 6 October 2026.
[37] US Department of Energy: MARVEL Microreactor Project. Accessed 6 October 2026.
[38] US Department of Energy, April 2026: DOME test bed. Accessed 6 October 2026.
[39] BWXT: Project Pele, manufacturing milestones through June 2026. Accessed 6 October 2026.
[40] Edison, 25 June 2026: declaration with EDF, Nuward and Italian industrial partners. Accessed 6 October 2026.
[41] Sky TG24 / ANSA, 13 June 2026: Pichetto Fratin’s 2033–34 target. Accessed 6 October 2026.
[42] Camera dei deputati, 16 January 2026: dossier AP0205; PNIEC scenarios and Italian nuclear history. Accessed 6 October 2026.
[43] ISIN: statutory role and current responsibilities, institutional description. Accessed 6 October 2026.
[44] IAEA PRIS: China reactor records, including HTR-PM, Linglong-1 and Zhangzhou-2; retrieved 6 October 2026. Accessed 6 October 2026.
[45] IAEA PRIS: Finland reactor records, including Olkiluoto-3; retrieved 6 October 2026. Accessed 6 October 2026.
[46] IAEA PRIS: France reactor records, including Flamanville-3; retrieved 6 October 2026. Accessed 6 October 2026.
[47] IAEA PRIS: US reactor records, including Vogtle-3 and Vogtle-4; retrieved 6 October 2026. Accessed 6 October 2026.
[48] IAEA PRIS: UAE reactor records, including Barakah; retrieved 6 October 2026. Accessed 6 October 2026.
Appendix — six distinctions and a four-object comparison
One-page reference sheet • Evidence cutoff: 6 October 2026 • Tables refer to the cited sources in the paper. Future dates and success criteria are conditional.
| Distinction | Operational definition |
| 1. Plasma gain / engineering gain / net electricity | Q = fusion power ÷ plasma-injected heating power. Declare plant-boundary gain separately. Grid export subtracts every electrical load and needs a heat cycle; Q > 1 is insufficient. |
| 2. Experiment / pilot / commercial fusion plant | Experiment proves physics or components; pilot integrates a plant pathway; commercial plant delivers a licensed, dependable product. BEST is the first category; no commercial fusion plant operates. |
| 3. Micro / SMR / larger fission reactor | Approximate electrical bands: micro <20 MWe; SMR ≤300 MWe/module; above that is outside the strict SMR band. One 400 MWe Nuward or 470 MWe Rolls-Royce reactor exceeds it. |
| 4. Law / licence / concrete / operation | An enabling statute makes rules possible; a project licence permits a defined activity; nuclear concrete marks construction; operation produces measured energy. Italy has reached the first stage. |
| 5. Nameplate / capacity factor / annual energy | Net MWe is a power rating. Capacity factor is actual MWh ÷ rated MWe-hours. Annual TWh = net GW × capacity factor × 8.76; pulse duration is not a fleet capacity factor. |
| 6. Research infrastructure / power project | A public experiment earns scientific results. A generating project needs a buyer/revenue model, financing, site licences and lifecycle obligations. Its product is net MWh and/or contracted heat. |
| Object | Reaction | Electrical output | Readiness / stage | Earliest grid electricity | Fuel and waste | Overnight-cost evidence | Regulatory position | Success in 2030 / 2040 |
| BEST | D–T fusion | No demonstrated net MWe | Experiment; assembly | ~2030 demo target; net unspecified | Tritium; activated structures | No commercial net-kWe basis | Chinese oversight; specific licence unverified | 2030: gain/power balance. 2040: pilot evidence. |
| Italian proposal | Uranium fission; PWR | 400 MWe vendor target | Proposed; above strict SMR band | 2033–34 minister / ~2035 company; conditional | Uranium; spent fuel | No Italian site estimate | ISIN; future decrees/licences | 2030: authorised project. 2040: measured operation. |
| Flamanville 3 | Uranium fission; EPR | 1,630 MWe net | Operating; commissioning | Grid: Dec 2024 | Uranium; spent fuel | €18.4bn, 2023 euros; Mar 2024 estimate | ASNR; EDF | 2030: reliable MWh. 2040: safe lifecycle. |
| Akademik Lomonosov | Uranium fission; PWR | 64 MWe net; 70 gross | Operating small plant | Grid 2019; commercial 2020 | Uranium; spent fuel | No comparable figure verified | Rostekhnadzor; Rosenergoatom | 2030: availability. 2040: maintenance/waste. |
Cost units: constant euros of 2023 for Flamanville’s dated estimate; no comparable overnight evidence asserted for the other rows. MWth denotes heat, MWe electricity. Uranium fission and D–T fusion are kept separate throughout.
