# Owner economics and buildable supply evidence Updated 10 September 2026. Physical adequacy is now paired with historical trading and capital-recovery screens. No technology is assumed to provide storage or flexibility without compensation. The screens remain pre-tax and omit identified project-specific costs; they are not investment valuations. ## Battery income from actual Swiss day-ahead prices The new archive contains complete Swiss calendar 2024 and 2025 day-ahead hourly prices: 8,784 and 8,760 observations. Energy-Charts distributes the underlying Bundesnetzagentur/SMARD prices with its stated attribution/license. Prices are EUR/MWh; CHF results use a fixed illustrative 0.95 CHF/EUR, not a historical exchange-rate series. [ch-prices-2024](); [ch-prices-2025](). A 1 MW/4 MWh plant is optimized with perfect foresight, 85% round-trip efficiency, 365 annual output-equivalent cycles, CHF 1/MWh each-way transaction cost, cyclic stocks and exclusive charging/discharging. The full-year mixed-integer benchmark permits every trading hour. | Historical year | Net energy-trading margin CHF/MW-year | Discharged MWh/MW-year | Cycles | |---|---:|---:|---:| |2024|58,215|1,460|365| |2025|75,649|1,460|365| | Four-hour initial capital assumption | Annual capital recovery + O&M CHF/MW | |---|---:| |CHF 250/kWh|149,795| |CHF 375/kWh|224,692| |CHF 500/kWh|299,589| Even the perfect-foresight day-ahead benchmark falls below all three assumed annual cost levels. This does not prove every Swiss battery business is unprofitable. Intraday, reserves and voluntary tolling can add value, but require evidence, available capacity and market depth. Historical price-taking returns cannot be multiplied by national GW without considering their price effects. The 18 sensitivity runs cover 2/4/8 hours and 100/50/25% intraday price-spread amplitude around each daily mean. These runs restrict charging to at/below the daily median and discharging to the other hours; they are a feasible heuristic, not the unrestricted upper bound. For 4 hours in 2025, net margin falls from about CHF 75,615 to 29,763 and 13,456 per MW-year as spreads compress. Compression is an exogenous sensitivity, not a modeled market equilibrium. Constant-real-margin 15-year NPVs and the additional revenue needed to break even are recorded for each capital band. Swissgrid reports 2025 general ancillary-services/balance-energy procurement costs of CHF 346.3 m, including CHF 104.0 m for control-power provision/unintentional deviation and CHF 177.3 m for ancillary-services energy. These are whole-system historical expenditures shared across services and suppliers, not a revenue entitlement for batteries. They put a scale boundary around claims that unlimited GW can all earn today’s reserve revenues. [swissgrid-financial-2025](). ## Swiss delivery evidence and remaining procurement uncertainty Axpo and energieUri inaugurated Gurtnellen systems on 9 September 2026. The Axpo design is 50 MW/100 MWh; energieUri’s operating announcement says 8.4 MW, superseding the 8.6 MW construction announcement. Fluence was named supplier for both. Their full installed costs, guaranteed AC energy and contract exclusions remain undisclosed in these sources. [axpo-gurtnellen-operation](); [axpo-gurtnellen-build](). Primeo reported four of five Kappel strings operating since spring 2026, with the fifth targeted for year-end. Full buildout is 30 MW/80 MWh. We do not count the entire 80 MWh as already operating or infer equal string ratings. Reserve services are the main reported purpose, with intraday testing; no audited plant profit is reported. [primeo-kappel-operation](). These projects corroborate that a Swiss supply and installation chain exists. They do not prove procurement of tens of GWh at our assumed cost. The earlier US fleet, international project budgets and supplier shipment evidence remain in 029; they have not been replaced with more optimistic learning forecasts. ## Hydro investment and water opportunity cost A seasonal shift must earn the winter price minus the net summer value sacrificed, after operating differences. Charging electricity is not free for pumped hydro; existing water likewise has an alternative use. For illustration, the 2024 tender paid€66.12/MWh to reserve 250 GWh for winter 2024/25, or€16.53 m. That was a reservation payment; activation energy was paid separately. It was neither construction cost nor the price of the stored electricity. [swissgrid-reserve-awards](). The table allocates each reported project budget to its stated winter-shift service, assuming 7% real financing, 60 years and 0.5% annual O&M, with no grant. Other benefits are not valued. Published CHF budgets are held constant in their reported money basis; this is not an independently inflation-adjusted tender comparison. | Project/service case | Budget CHFm | Winter shift GWh | Required winter-minus-summer net value CHF/MWh | |---|---:|---:|---:| |Mattmark|83|60|105.5| |Gornerli_2040|510|435|89.4| |Moiry_40 GWh|120|40|228.7| |Moiry_50 GWh|120|50|183.0| The 168 hydro cases vary financing 4/7/10%, life 40/60/80 years, O&M 0.5/1%, and cost 1/1.5 times. Moiry also has a six-year sensitivity: hypothetical 2033 completion to 2039 concession expiry with no terminal compensation. Long physical asset life does not guarantee the owner can retain its cash flows. Concession extension or residual-value compensation must be explicit. Construction finance, lost production during works, taxes, actual maintenance, hydrological underfill and negotiated concession compensation remain outside these numbers. Other project services may add value. Consequently, the table neither proves that hydro should be subsidized nor that it is uneconomic; it identifies the revenue that must be evidenced under the chosen boundary. ## New nuclear: equal treatment of delivery and finance EDF’s February 2026 estimate for 3.3 GW Hinkley PointC is£35 bn in 2015 prices, about£48 bn current money, excluding interim interest. Unit 1 targets 2030 with a 2031 delay case; Unit 2 follows 12 months later. The currency distinction matters:£48 bn is not€48 bn. These are operator estimates, not Swiss offers. [edf-hinkley-2026](). The six-unit EPR2 programme estimate is€72.8 bn in 2020 money, with first commissioning targeted 2038 and 12–18 months between later units. EDF’s April 2026 registration document confirms the DINN audit occurred inQ 1; recommendations were to be implemented by year-end. The support framework includes a subsidized State loan covering 60% of construction, a CfD and risk sharing. Thus it does not demonstrate unsupported merchant financing. [edf-urd-2025](). The July 2026 issuer release adds manufacturing progress and an ongoing EU State-aid investigation. It also reports three reactors temporarily shut and nine reduced during the June/July heatwaves. That supports keeping cooling/climate constraints explicit; it is not a Swiss outage probability. [edf-halfyear-2026](). CEZ’s primary timeline records the two-unit KHNP contract inJune 2025. Swiss parliamentary action inJune 2026 is a legislative milestone, not a site license, financing decision or commissioning schedule. [cez-dukovany](); [parliament-nuclear-2026](). The 54 nuclear sensitivity cases use CHF 6,000/10,000/15,000 per kW overnight, 6/10/15 construction years, 4/7/10% real financing and 75/90% capacity factor. A uniform expenditure schedule capitalizes interest to operation. At 7%, 10 years and 90%CF, the resulting partial cost screens are about CHF 108/159/224 per MWh. The assumed CHF 120/kW-year fixed and CHF 15/MWh fuel/variable/waste allowances are unbundled. Decommissioning funding, fund shortfalls, taxes and site-specific items remain additional: these are not complete Swiss project LCOEs or private insurance estimates. Equations and full results are in [owner-economics.json](); reproduce with `python -m feasibility_model.economics`. Capital recovery is r/[1−(1+r)^−n]; constant-margin NPV is −capital +(annual margin−annual O&M)/capital-recovery-factor. No future technology learning or public subsidy is credited.