Switzerland’s four reactors produced an average of 20.7 TWh a year over 2016–2025, including years with long outages. Losing that output adds to the supply already needed for electrification.
We recommend keeping Gösgen and Leibstadt while safety and owner economics support them, and developing two new reactors now. Beznau’s planned retirement comes before the new plants in our delivery scenarios, so its replacement needs a separate, earlier supply package.
Ten-year operating record · Replacement and delivery datesElectricity from existing reactors
The record shows substantial production and prolonged outages. Keep useful output while arranging cover for weeks or months without a large unit. The ten-year average includes those interruptions.
Historical dataDefinition and limits of the ten-year average
Annual generation in TWh · bars start at zero; full scale 30 TWh
- 201617.3
- 201716.5
- 201821.5
- 201922.2
- 202023.0
- 202118.5
- 202223.1
- 202323.3
- 202423.0
- 202518.4
Beznau 1, Beznau 2, Gösgen and Leibstadt throughout. Mühleberg is excluded. All outage years remain included.
BFE generation series · Chart data and arithmetic
BFE’s Gösgen figure for 2025 differs from the operator’s electricity-only total. Definitions and source difference.
Annual plant results and prolonged outages
Capacity factor across three periods
A capacity factor compares actual output with output at full nominal capacity throughout the period. It reflects both operating time and output while running.
On a small screen, scroll the table to see all columns.
| Reactor | 2016–2025 Ten years | 2021–2025 Five years | 2025 |
|---|---|---|---|
| Beznau 1 | 68.0% | 86.0% | 91.0% |
| Beznau 2 | 89.2% | 87.8% | 77.7% |
| Gösgen | 85.3% | 79.2% | 38.8% |
| Leibstadt | 75.4% | 80.8% | 88.5% |
Generation and capacity: BFE table 15 and footnotes · Period calculations
How much electricity per year?
| Reactor | 2016–2025 Annual mean | 2021–2025 Annual mean | 2025 |
|---|---|---|---|
| Beznau 1 | 2'177 | 2'751 | 2'909 |
| Beznau 2 | 2'855 | 2'807 | 2'483 |
| Gösgen | 7'552 | 7'012 | 3'429 |
| Leibstadt | 8'094 | 8'697 | 9'558 |
BFE table 15 · Inputs and arithmetic means . The same four reactors are compared throughout. Mühleberg is excluded, including before its closure.
Gösgen: operating history and the 2025–2026 shutdown
Gösgen stopped for planned maintenance on 24 May 2025. A potential design weakness in the feedwater system delayed restart.
The plant replaced check valves and submitted safety evidence. ENSI approved restart on 23 March 2026, after about ten months offline. ENSI: findings and restart approval
The operator confirms that electricity delivery resumed that day. Gösgen: return to the grid
Gösgen’s capacity factor was 90.4% across 2016–2024, before the extended outage. Across the full 2016–2025 period, including the outage, it was 85.3%. Historical calculations
Calculated capacity factor. Each bar uses the same 0–100% scale.
- 201692.1%
- 201791.4%
- 201892.4%
- 201987.7%
- 202093.0%
- 202188.5%
- 202289.2%
- 202390.2%
- 202489.4%
- 202538.8%
2025 includes the extended shutdown. BFE table 15 · Data and calculation
Annual data and calculation method
| Year | Beznau 1 | Beznau 2 | Gösgen | Leibstadt |
|---|---|---|---|---|
| 2016 | 0 GWh 0.0% | 2'994 GWh 93.4% | 8'174 GWh 92.1% | 6'108 GWh 57.0% |
| 2017 | 0 GWh 0.0% | 2'768 GWh 86.6% | 8'091 GWh 91.4% | 5'649 GWh 52.9% |
| 2018 | 2'452 GWh 76.7% | 3'022 GWh 94.5% | 8'178 GWh 92.4% | 7'813 GWh 73.1% |
| 2019 | 2'815 GWh 88.0% | 2'803 GWh 87.7% | 7'757 GWh 87.7% | 8'820 GWh 82.5% |
| 2020 | 2'747 GWh 85.7% | 2'930 GWh 91.4% | 8'254 GWh 93.0% | 9'059 GWh 84.5% |
| 2021 | 3'026 GWh 94.6% | 2'824 GWh 88.3% | 7'828 GWh 88.5% | 4'852 GWh 45.4% |
| 2022 | 2'521 GWh 78.8% | 2'944 GWh 92.1% | 7'895 GWh 89.2% | 9'753 GWh 91.3% |
| 2023 | 2'892 GWh 90.4% | 2'780 GWh 86.9% | 7'980 GWh 90.2% | 9'682 GWh 89.6% |
| 2024 | 2'406 GWh 75.0% | 3'006 GWh 93.8% | 7'929 GWh 89.4% | 9'642 GWh 89.0% |
| 2025 | 2'909 GWh 91.0% | 2'483 GWh 77.7% | 3'429 GWh 38.8% | 9'558 GWh 88.5% |
Annual capacity factor = generation in GWh × 1,000 ÷ (net capacity in MW × calendar hours) × 100.
For several years, divide total generation by the sum of each year’s capacity × hours. Use the underlying energy, capacity and hours before rounding. Leap years have 8,784 hours; other years have 8,760.
Beznau uses 365 MW per reactor and Gösgen uses 1,010 MW. Leibstadt uses 1,220 MW through 2022, then 1,233 MW. These are BFE’s dated nominal net capacities. Table 15, footnotes 1, 4 and 6
BFE’s published utilization rates include heat for Beznau and Gösgen. Our capacity factors are calculated from its generation rows. Hourly and winter reliability require output and outage data at those time scales.
Gösgen’s 2025 total differs between BFE and the operator. Source definitions and difference.
Source definitions and the Gösgen 2025 difference
BFE lists 3,429 GWh for Gösgen in 2025. The operator reports 3,397 GWh of electricity and 31 GWh of process steam in electrical-equivalent units. BFE table 15 · Operator report, page 9
The tables use BFE’s series throughout. The difference from the operator’s figures remains unresolved. Using the operator’s electricity-only figure gives a calculated capacity factor of 38.4%. Calculation and boundaries
Download the data, methods and source notes (JSON) · Download annual rows (CSV)
Long outages in the operating record
Extended outages at Leibstadt or Gösgen overlapped at least 5 of the ten winters from 2016/17 to 2025/26. Some began as planned upgrades and lasted longer than scheduled.
| Plant / winter | Shutdown and return |
|---|---|
| Leibstadt 2016/17 | August 2016–February 2017 Fuel inspections and an interrupted restart delayed resumed power operation until 20 February. Operating record Operating record |
| Leibstadt 2017/18 | September–December 2017 Major maintenance was followed by fuel-replacement and approval delays. Grid return: 18 December. Operating record |
| Leibstadt 2018/19 | September–November 2018 A planned four-week shutdown extended to about seven weeks. Grid return: 2 November. Operating record |
| Leibstadt 2021/22 | May–December 2021 Major condenser and circulation-system upgrades took longer than scheduled. Grid return: 3 December. Operating record |
| Gösgen 2025/26 | May 2025–March 2026 Feedwater-system work prolonged the shutdown. Grid return: 23 March, before winter ended. Operating record Operating record |
These events support arranging replacement energy as well as immediate backup power. Their different causes and durations do not establish the probability of the model’s chosen 90-day outage. Returning to the grid can precede full output.
Continued operation has conditions
ENSI’s national report describes the system for safety assessments and oversight. Our support for existing plants depends on meeting the applicable safety requirements and on a viable operating plan.
ENSI: safety assessment and oversightWe recommend pursuing operation to 80 years at Gösgen and Leibstadt, subject to safety and the owner’s investment decision. The extension study provides an economic basis for pursuing that option. Its maintenance, outage, waste and funding assumptions remain part of the decision.
60-to-80-year extension studyThe costs chapter compares Swiss life-extension scenarios, including reduced operation and early closure.
Replace Beznau on its own timetable
Axpo announced operation of Beznau 2 until 2032 and Beznau 1 until 2033. ENSI records these intentions alongside the requirement for continued safe operation.
ENSI national report: introduction and Beznau operating intentionsTogether, the plants generated an average of 5.56 TWh a year during 2021–2025, including outages. This gives the scale of the energy to replace. Winter output and reliable supply during difficult hours also matter.
Five-year generation average and original recordsOur recommendation: arrange replacement supply before these retirements, independently of the first new reactor. Advance viable generation, efficiency and voluntary demand flexibility. Contract remaining winter supplies and reserves with their costs and availability stated. Imported electricity can help during the transition; reliance on it must remain visible in the sovereignty assessment. See the replacement timeline and the reserve services to procure.
Plan for the whole delivery period
Use 2047 as the working date for a first new station, with supply arrangements that can survive a delay to 2053 or later. This selects the planning case from our illustrative schedules. It is a programme assumption, conditional on legal change, approvals, suppliers and finance. The accelerated case, with commercial operation in 2042, remains an opportunity rather than electricity credited to the earlier bridge.
Scenario dates and dependencies Chosen scheduling referenceStart reversible development work now: establish the owner team, assess candidate sites, compare designs and test customer interest in long-term electricity contracts. Make construction depend on the required permissions and a funded delivery plan.
- Favourable caseCommercial operation 2042
Timely decisions, a prepared site and secured suppliers. First concrete: 2034. First grid: 2041.
- Planning illustrationCommercial operation 2047
Longer assessment and construction, with some work in parallel. First concrete: 2037. First grid: 2046.
- Delayed caseCommercial operation 2053
Legal or licensing delays, followed by difficult construction. First concrete: 2041. First grid: 2052.
The delayed case passes 2050. Greater delay or cancellation is possible. Further units need their own sites, suppliers, workforce and finance.
All timeline assumptions and dependencies
| Stage | Favourable case | Planning illustration | Delayed case |
|---|---|---|---|
| Legal availability assumed | 2028 | 2028 | 2031 |
| General licence process | 2028–2032 | 2029–2035 | 2032–2039 |
| Construction licence and design work | 2030–2034 | 2033–2037 | 2037–2041 |
| Construction start to first grid | 7 years | 9 years | 11 years |
| Commissioning after first grid | 1 year | 1 year | 1 year |
Design and application preparation overlap in these cases. Permission must precede the work that depends on it. Legal change, site approval, a suitable design, supplier slots and financing are conditions in every case.
First grid connection means commissioning electricity can enter the network. These scenarios count the unit’s planned annual contribution from commercial operation.
The referendum has qualified. DETEC expects a vote in 2027. Voters must approve the change before new nuclear projects can proceed under it. Plant licensing follows separately.Referendum qualification, 19 October · DETEC: vote expected in 2027
Grid approvals and trained people also determine delivery
The grid-law amendment has a referendum deadline of 21 January 2027. Faster procedures depend on commencement and implementation. No guaranteed project time saving follows from that deadline. Grid-law deadline and commencement
Fund recruitment, qualification and continuing training alongside the reactor project. ENSI’s September B10 draft covers operating staff, maintenance personnel, contractors and decommissioning. Consultation runs until 15 January 2027. It provides a planning checklist; final requirements depend on adoption. Public evidence still lacks project-specific headcounts, salaries and simulator costs.
Our proposed programme develops two standardised units, with working operation dates of 2047 and 2050. The second requires overlapping construction. The programme chapter shows the budget, owner financing and the supply still needed alongside them.
Two-unit planning case and delivery conditionsAll three cases arrive after Beznau’s announced retirements. A replacement programme therefore needs both near-term supply and long-term construction. If a project misses a key approval or financing milestone, update its expected contribution and secure the resulting gap immediately.
Start with designs that have operating references
Compare AP1000, APR1400 and EPR designs in a competitive Swiss feasibility study. Each has operating reference plants in the records below. Assess site, cooling, grid, supplier capacity and financing before selecting a project.
Reference designs, construction dates and source recordsYears · common scale from zero to 20
These examples include faster builds and major delays. Country, design, contractors and project organisation differ. The range gives useful stress tests for a Swiss plan.
Milestones, selection and source differences
Six completed units span four designs and four countries. This selection illustrates variation; it gives no probability of a Swiss completion date.
All intervals use IAEA PRIS dates and calendar days divided by 365.25. First grid connection precedes commercial operation while commissioning continues.
| Unit | Construction start | First grid | Commercial operation | Source |
|---|---|---|---|---|
| Fuqing 5 | 2015-05-07 | 2020-11-27 | 2021-01-30 | IAEA PRIS |
| Barakah 1 | 2012-07-19 | 2020-08-19 | 2021-04-01 | IAEA PRIS |
| Taishan 1 | 2009-11-18 | 2018-06-29 | 2018-12-13 | IAEA PRIS |
| Sanmen 1 | 2009-04-19 | 2018-06-30 | 2018-09-21 | IAEA PRIS |
| Vogtle 3 | 2013-03-02 | 2023-03-31 | 2023-07-31 | IAEA PRIS |
| Olkiluoto 3 | 2005-08-12 | 2022-03-12 | 2023-05-01 | IAEA PRIS |
For Olkiluoto 3, the operator’s regular-production date differs from the PRIS commercial date. This chart consistently uses PRIS’s milestone.
The faster examples show what some projects achieved. The delayed examples show the consequences a Swiss owner must prepare for. Neither outcome transfers automatically to Switzerland.
Keep small modular reactors as an option. Give them a construction commitment when reference performance, suppliers and executable offers justify it. The first project should earn its place through demonstrated capability and acceptable commercial terms.
Build on Switzerland’s earlier proposals
Start the site comparison with Beznau, Mühleberg and Niederamt. Their earlier proposals contain geological investigations, flood studies, cooling concepts and generic plant layouts. This work gives a new owner a concrete starting point for assessing locations.
ENSI: scope of the earlier assessments| Project | Proposed power | Historical findings to check against later work |
|---|---|---|
| Beznau | 1'450 MW | Flood protection, including effects on neighbouring nuclear facilities. Page 2, hydrology paragraph |
| Mühleberg — Niederruntigen | 1'450 MW | Slope stability, landslides and rockfall. Page 1, final paragraph |
| Niederamt — near Gösgen | 1'100 MW or 1'600 MW | Regional geology and reliable grid connection, with any necessary network expansion. Printed page 28, recommendations 4.1.6 and 4.1.7 |
Each listed power class allowed ±20%. Niederamt’s two sizes were alternatives for one plant. All three proposals included cooling towers.
ENSI: proposed power, cooling and site conditions, pages 1–2 Download the proposal data (CSV)
How far did the earlier projects get?
- 2008Applications submitted
Three projects enter the general-licence process. Source
- November 2010Site assessments
ENSI finds geological conditions suitable in principle and requires further investigations. Source
- March 2011Procedures suspended
The government suspends the procedures after Fukushima. Source
- October 2016Applications withdrawn
Owners cite changed market conditions and the political direction. Source
These were general-licence applications: an early assessment of the site and project outline. The withdrawn applications provide no current permission to build.
Fund a comparative site study with a decision at the end: select a feasible project or stop that option.
What must a fresh comparison establish?
- Safety and land: check present site conditions, earthquake and flood hazards, available land and the space a proposed reactor needs.
- Cooling: test water availability, heat discharge, electricity used by cooling equipment and costs during hot, dry conditions.
- Connection: establish how reliably the plant can feed the grid, what network expansion it needs and when that work can finish.
- Delivery: identify a willing owner, a suitable reactor and suppliers, current approvals, finance and responsibility for overruns.
These are our recommended checks. ENSI’s earlier findings and the Nuclear Safety Commission’s Niederamt recommendations provide specific starting questions.
ENSI: site findings KNS: owner, geology and grid recommendations, page 28
The value of reuse depends on how much earlier work remains applicable. A credible estimate of years or money saved is still missing. The delivery scenarios above therefore assume no time saving from these dossiers.
Reuse assessment and remaining uncertaintyMake owner financing a condition
| Annual output as a share of full power | All output earns revenue | Only 75% earns revenue |
|---|---|---|
| 70% | 192 | 256 |
| 85% | 163 | 217 |
| 93% | 151 | 201 |
Lower availability and unsold output both raise the price the owner needs. The 75% case keeps costs fixed and gives the remaining output zero revenue. It is a sales-risk sensitivity, not a prediction of nuclear curtailment or electricity prices.
The disposal allowance is an explicit assumption, pending a site-specific fund assessment. The original owner case uses 85% annual output; the separate financing demonstration uses 90% to isolate financing effects. Neither replaces observed reactor histories or the hourly supply cases with explicit maintenance and outages.
Recent European projects justify a wide cost range
EDF’s 2025 accounts put Hinkley Point C’s two-reactor project estimate at £35 billion in 2015 money. EDF’s Flamanville figure of €13.2 billion in 2015 money covers construction. The comparison below shows the broader investment totals in the French audit. These project estimates support testing finance and overruns; a Swiss project needs its own scoped offer.
EDF accounts, completion estimates and price years
Billion euros at 2015 prices · each bar is a different total for the same project
- Construction estimate13.2
- Plant and owner costs before financing15.6
- EDF total including financing19.3
- Broader auditor estimate20.4
Financing and owner costs change the investment an owner must recover. These totals overlap; adding the bars would count the same costs several times.
Estimate dates and scope
The January 2025 audit reproduces EDF’s November 2023 estimates, then gives a broader auditor total. These are estimates, not final accounts. First-cycle operating costs and Swiss project conditions need separate treatment. All bars use 2015 euros; values in 2023 euros differ.
Our preferred financing route
Start with an owner consortium and long-term customer contracts. Require owners and suppliers to carry defined construction risks. If viable offers still require public price support, compare a capped, two-way contract: support below an agreed price, repayment above it. Publish the maximum public exposure, end date and overrun allocation before approval.
This is our proposed negotiating position. Evidence does not yet establish an affordable guarantee cap or prove that public support is unavoidable. We would not promise unlimited overrun cover or automatic customer charges during construction.
Require customers, owners and suppliers to agree who carries delays, outages and overruns. Long-term electricity contracts must leave enough income for maintenance, fuel, waste and closure. Compare offers with alternatives delivering the same dependable supply.
Conditions for committing to construction
- A suitable design and permitted site, with cooling, grid and safety requirements costed.
- Reserved supplier capacity and a schedule covering commissioning and delay allowances.
- A complete price with currency, price year, finance and contingency stated.
- Named parties responsible for overruns, delays and reduced output.
- Funded fuel, maintenance, insurance, waste and dismantling obligations.
Public guarantees or price support would transfer risk to taxpayers or customers. Any such proposal needs a separate, explicit decision. The cost chapter shows how financing and construction duration change the price a plant must recover.
Costs, waste and supply dependence
Fuel independence needs more than a reactor
Nuclear fuel can be stocked ahead of use, but Switzerland depends on foreign uranium, conversion, enrichment and fuel fabrication. Axpo reports Canadian and Kazakh uranium contracts and a Western processing chain. It also reports strategic stocks of Russian origin. Public evidence does not establish an audited duration for those stocks.
Our requirement: diversify the complete fuel chain and contract reloads and stocks before counting dependable generation. Axpo fuel-chain explanation, June 2026
Fund waste services for the full reactor lifetime
The repository application includes reserves for longer operation of existing reactors, including an 80-year case. It explicitly excludes new reactors. Each new project therefore needs its own funded plan for interim storage, fuel packaging and final disposal. Application inventory · Nagra: existing and new reactors
Make these services and their costs part of the investment decision. Owners must fund the obligations, and the regulator must approve the relevant capacity and safety case. Our added disposal-cost allowance remains an assumption until a project-specific assessment is available. Licence and funding conditions · Disposal-cost assumption
What the existing reserves cover, and why timing matters
The proposed maximum is 2'500 m³ of unpackaged high-level waste. It starts from a rounded 1'550 m³ baseline and includes 550 m³ for 80 rather than 60 years of existing-reactor operation, plus other margins. Packaging and the space needed to dissipate heat are separate. The application quantity establishes no approved capacity for new units.
The existing programme’s reference schedule places high-level waste in the repository in 2060–2074. A new reactor starting in 2050 and operating for an assumed 60 years produces its final spent fuel around 2110, before cooling and disposal. Its storage, packaging and disposal services must remain available beyond the existing programme’s reference intake period.
Safety report, Appendix A · Reference inventory schedule, section 2.5 · Lifetime illustration and its assumptions
Existing-fleet closure costs
The submitted 2026 cost study estimates CHF 23.602 billion for decommissioning and disposal, including federal waste and costs already incurred. A separate CHF 1.629 billion covers post-operation activities after electricity production ends. These are lifecycle estimates at January 2026 prices, not the remaining bill or a fund shortfall.
Independent review precedes final contribution decisions. These existing-fleet estimates do not price our proposed new reactors or 80-year extensions. Those need separate cost and inventory assessments.
KS26: lifecycle totals and assumptions · Submission and review status
Unlimited liability is not unlimited insurance
Swiss operators have unlimited liability for nuclear damage. Required cover starts at EUR 1.2 billion, plus 10% for interest and legal costs. Operator assets beyond that cover remain exposed. Federal cover fills specified private-insurance gaps; international supplementary compensation has a separate limit.
The ordinance permits certain exclusions or limits for exceptional natural events, war, terrorism and late claims. It does not disclose each plant’s private policy. Any new project must show its actual insurance, federal premiums and remaining exposure. An extreme loss could still require a parliamentary decision on public intervention.
BFE: liability and compensation · Liability ordinance, Article 7
What Swiss system models tell us
The ETH–PSI study finds that construction cost, required returns and policy decide whether its models choose new nuclear. Its market-driven cases disagree on that choice. Some modelled net-zero systems use no new nuclear. The study does not test how either mix supplies electricity during restricted imports and simultaneous outages.
Study results and limitsOur conclusion: develop a real Swiss offer and test its delivery, costs and winter service. The study helps identify favourable terms; it cannot establish that a supplier will offer them.
ETH–PSI: assumptions, model differences and excluded supply tests
Construction cost, finance and policy affect whether new nuclear earns a place in the electricity mix. The June 2026 ETH–PSI study compares four computer models, including nuclear alongside hydro, renewables, storage and electricity trade. It also finds modelled net-zero systems without new nuclear, using renewables, flexibility, storage and trade.
ETH–PSI: methods and conclusions, pages 6–16 and 27–28In 2024 money, before construction financing.
Lower investment risk, such as through public guarantees.
Non-hydro renewables and new nuclear share the target. Both receive support to meet it.
A model result for this low-cost case. Actual projects need supplier prices, finance and approval.
The financing rate is the average annual return required by lenders and investors. Researchers call it the weighted average cost of capital, or WACC. Higher construction costs reduce the nuclear capacity selected in this case.
Report: pages 21–22, Figure 6 Appendix: financing and policy definitions, pages 7–11
The previous case requires 45 TWh a year from renewables excluding hydro and from new nuclear, together. This comparison removes that requirement.
The models can choose how much electricity Switzerland generates or imports, within their other assumptions. The net-zero emissions rule remains.
Policy change and retained climate rule, pages 10–11
Each row is a computer model. These tools calculate a low-cost energy system under chosen assumptions. Their names identify different tools, whose scope is explained beside each result.
The columns set the annual return that lenders and investors require on money invested in new nuclear. The study uses 8% above inflation to represent private investment. Its 5% case assumes reduced investment risk, such as through government support.
Both columns test the same assumed construction prices: CHF 5'000, 8'000 and 12'000 per kilowatt of generating capacity.
For an illustrative 1 GW station, those prices mean CHF 5 billion, 8 billion or 12 billion. These are construction assumptions in 2024 money, before financing costs. Scale calculation
| Model and what it covers | Higher financing cost8% above inflation | Lower financing cost5% above inflation |
|---|---|---|
| STEMSwiss TIMES Energy System ModelElectricity, heating, transport and other energy uses. It follows the transition over time. | Includes new nuclearAt all three construction prices. | Includes new nuclearAt all three construction prices. |
| Nexus-eElectricity supply, networks and trade with neighbouring countries for a chosen future year. | No new nuclearAt any of the three construction prices. | Includes new nuclearOnly at CHF 5'000/kW. |
| SESSwiss EnergyScopeElectricity, heating and transport together, for a chosen future year. | No new nuclearAt any of the three construction prices. | No new nuclearAt any of the three construction prices. |
| SES-ETHSwiss EnergyScope — ETH variantA separately developed version of EnergyScope, also covering electricity, heating and transport for a chosen future year. | No new nuclearAt any of the three construction prices. | No new nuclearAt any of the three construction prices. |
STEM includes new nuclear at every tested price. Nexus-e includes it only when both construction and financing are at the lower tested levels. The two EnergyScope versions choose other sources in these cases.
The tools differ in demand, renewable potential, trade and operation as well as scope. Their results describe those assumptions; an actual plant needs an owner, suppliers and finance.
Model descriptions, pages 1–3; financing, pages 8 and 13; results, Figures 21 and 25 System costs and owner decisions, page 15 Results for every price and financing case (CSV)
Reliability and delivery need their own tests
The study excludes security-of-supply stress analysis and leaves the 5 TWh winter net-import goal out of all models. Its nuclear and no-new-nuclear cases therefore leave our sovereignty test open: dependable supply during import shortages, difficult weather and coincident outages.
Security-of-supply scope, page 2 Winter import goal and reasons for exclusion, pages 7 and 11New nuclear starts in 2050 by assumption, and the study excludes extensions of existing reactors. For our plan, continued safe operation and the actual delivery of new stations remain separate decisions, with the timetables in the delivery section above.
Start year and plant lifetimes, page 10How to interpret the capacities and model differences
The models choose capacity continuously from a minimum of 300 MW, using generic light-water reactor assumptions. A real programme needs discrete reactor sizes and suitable sites.
Demand treatment, renewable resource limits, trade representation and operating assumptions differ between models. Each result reflects those choices as well as its model structure.
The selected scenarios allow carbon removal abroad to offset some emissions. Their net-zero result depends on that accounting choice.
The appendix also leaves implementation questions about SES construction-finance treatment, some solar financing rates and STEM dismantling costs. It notes that nuclear operating-cost assumptions based on high use may understate costs at lower use.
These questions need the exact model inputs and code. The published results are useful conditional evidence; we have not independently reproduced the models.
Tables 2–3 and methods, pages 3–12 Selected findings, assumptions and unresolved implementation questions
Continue to the cost of dependable electricity.
References
- Winter supply, lake access and nuclear fleet comparisons
WattInfo · Calibrated hourly scenarios, 26 September 2026 - WattInfo electricity programme: 66 hourly cases, delivery dates and owner finance
WattInfo · Conditional programme, 19 September 2026 - WattInfo proposed supply path: decisions and supporting calculations
WattInfo · Proposed priorities and fixed-charging supply screen, 18 September 2026 - Economics of extending Gösgen and Leibstadt to 80 years
BFE commissioned research · Report dated 12 March 2026; publication date not established - Visual explanations: seasonal stress, demand and storage
WattInfo · Visual chapter data and explicit teaching assumptions - Swiss Electricity Statistics 2025
Swiss Federal Office of Energy (BFE) · 2025 - Convention on Nuclear Safety: tenth Swiss report
Swiss Federal Nuclear Safety Inspectorate (ENSI) · Tenth report - Nuclear generation and capacity factors, 2016–2025
WattInfo / Niko Storni · Historical comparison v1 - Gösgen restart permission after the extended shutdown
Swiss Federal Nuclear Safety Inspectorate (ENSI) · Restart decision of 23 March 2026 - Return to grid after extended outage, 23 March 2026
Kernkraftwerk Gösgen · 2026-03-23 - Gösgen annual report 2025
Kernkraftwerk Gösgen-Däniken AG · 2025 annual report - Nuclear construction records and conditional Swiss delivery cases
WattInfo · Delivery evidence, 17 September 2026 - IAEA PRIS country units CN
International Atomic Energy Agency · Snapshot retrieved 17 September 2026 - IAEA PRIS country units AE
International Atomic Energy Agency · Snapshot retrieved 17 September 2026 - IAEA PRIS country units US
International Atomic Energy Agency · Snapshot retrieved 17 September 2026 - IAEA PRIS country units FI
International Atomic Energy Agency · Snapshot retrieved 17 September 2026 - DETEC current nuclear policy and referendum status
DETEC / UVEK · Snapshot retrieved 17 September 2026 - ENSI comparison of the three proposed nuclear plants
ENSI; archived by BFE · 2010-11-15 - Niederamt: Nuclear Safety Commission opinion
KNS; archived by BFE · KNS document dated 15 December 2010; BFE publication date 10 January 2011. - Swiss nuclear development and the withdrawn applications
BFE · Snapshot retrieved 17 September 2026 - Owners withdraw the nuclear general-licence applications
Alpiq, Axpo and BKW · 2016-10-12 - Historical nuclear proposals: selected data and reuse assessment
WattInfo · Historical proposals, reviewed 18 September 2026 - ETH–PSI: Nuclear Power and the Future Swiss Energy System
ETH Zurich and PSI; 19 authors including Manera · 2026-06-29 - ETH–PSI: nuclear white paper modelling documentation
ETH Zurich and PSI · 2026-06-29 - ETH–PSI study: conditions, model inputs and reading notes
WattInfo · Study review, 18 September 2026 - ETH–PSI comparison: model scopes, financing and construction-price examples
WattInfo · Reader explanation checked against the June 2026 documents, 18 September 2026 - EDF 2025 consolidated financial statements
EDF · Selected primary-source passages - Auswirkungen neuer KKW auf das Tiefenlagerprojekt
ENSI · Selected primary-source passages - Where does the uranium for Beznau and Leibstadt come from?
Axpo · Selected primary-source passages - Nuclear Energy Liability Ordinance
Swiss Confederation · 2024-01-01 - Nuclear energy liability
BFE · Undated page; checked 2 October 2026 - Swiss energy consumption and production: nuclear policy status
DETEC · 2026-06-29 - Supply scale, fuel and nuclear owner-cost sensitivities
WattInfo · 2026-10-02 - NTB24-01: General licence application safety report
Nagra / SFOE archive · 2025-05 - Sind neue Kernkraftwerke im Tiefenlager eingerechnet?
Nagra · 2024-09-04 - NTB22-05 Model radioactive-waste inventory for repository application
Nagra · 2023-04 - Measured heating demand and repository planning boundaries
WattInfo · 2026-10-04 - Federal Chancellery nuclear and electricity-law referendum observations
Federal Chancellery / LINDAS · Official snapshot modified 13 October 2026; checked 14 October - ENSI B10 personnel draft, September2026
ENSI · 2026-10-09 - ENSI B10 consultation announcement, rendered browser excerpt
ENSI · 2026-10-09 - Kernenergiegesetz: Referendum zustande gekommen
Federal Chancellery · 2026-10-19 - UVEK current votes: nuclear-law referendum in 2027
UVEK · 2026-10-23 - Electricity Act network acceleration amendment, adopted 2 October 2026
Federal Assembly / Fedlex · 2026-10-23 - Kostenstudie 2026: Mantelbericht
swissnuclear · 2026-09-30 - Kostenstudie 2026 eingereicht
STENFO · 2026-09-30 - EPR industry: renewed momentum and persistent risks — full audit report
Cour des comptes · 2025-01-14
Find definitions, calculations and model inputs in the source register.