Record-low river levels are simultaneously reducing hydroelectric generation, restricting nuclear cooling systems and widening the price gap between solar-heavy daytime hours and increasingly expensive evening supply.
Water availability has become a direct electricity-market variable in Europe during the summer of 2026. The problem is particularly acute across Central and Southeast Europe, where the Danube supports hydropower production, nuclear cooling systems, industrial water supply and cross-border transport at the same time. The European Commission’s Joint Research Centre reported that the Danube, Rhine, Loire and Po had fallen to record-low August levels, while 50% of the EU and the UK was experiencing some degree of drought and 9% was already at the most severe alert level. Forecasts continued to indicate warmer and drier conditions through August and into September.
This is not a conventional generation outage affecting one power plant or one fuel. It is a correlated hydrological shock. Low water reduces output from hydropower plants, or HPPs, while simultaneously limiting the cooling-water systems of nuclear power plants, or NPPs. Heatwaves then increase air-conditioning demand precisely when hydroelectric and nuclear availability is falling. The same weather pattern can also reduce wind generation during stagnant high-pressure periods, leaving gas, coal, imports, batteries and demand response to balance the evening system.
Hydropower is affected directly because electricity output is determined primarily by water flow, hydraulic head and turbine efficiency. Run-of-river plants experience the impact almost immediately: lower river discharge means fewer cubic metres of water passing through the turbines and therefore fewer megawatt-hours. Reservoir plants can temporarily maintain production by drawing down stored water, but this consumes the system’s most valuable seasonal flexibility. Operators therefore tend to preserve reservoir water for evening peaks, emergency balancing or the coming winter rather than maximise summer baseload production.
Nuclear plants face a different problem. The nuclear fuel and reactor may remain fully available, but the conventional side of the plant needs large volumes of cooling water to condense steam after it leaves the turbines. Low river levels can place water-intake structures and pump inlets outside their normal operating range, while high river temperatures can restrict how much heated water may be returned to the environment. Output reductions and controlled shutdowns are therefore preventive operating measures rather than indications of a nuclear safety failure.
The clearest example is Romania’s Cernavodă nuclear power plant. Unit 1 had already been shut as the Danube declined, and Nuclearelectrica began the controlled shutdown of Unit 2 on the morning of 13 August 2026 because of the continuing fall in the river level. Each unit has approximately 700 MW of installed capacity, leaving around 1.4 GW of Romanian baseload generation unavailable. The company confirmed that both units remained in a safe shutdown state and that the event had no impact on personnel, the population or the environment.
Cernavodă normally supplies around one-fifth of Romania’s electricity requirements. Its absence is therefore materially different from the loss of a small thermal unit. It increases Romanian import requirements, removes a major source of low-variable-cost baseload electricity and raises balancing exposure during evening demand peaks. Nuclearelectrica also obtained force-majeure certificates covering electricity-delivery contracts because the severe Danube drought prevented the company from meeting all contracted supply obligations from its own generation. The water shortage has consequently moved beyond an operational problem and into electricity-contract, counterparty and earnings risk.
Hungary has faced a similar problem at the 2,000 MW Paks nuclear power plant, which normally provides close to half of the country’s domestic electricity generation. Low Danube levels forced Units 1, 3 and 4 to shut, while Unit 2 operated at half output for eleven days before beginning a gradual increase on 10 August. The latest official government update located, dated 13 August, said Paks was operating at only 500 MW, equivalent to approximately 25% of installed capacity, while the Danube was again forecast to decline.
The Paks case demonstrates that the constraint is not necessarily the total amount of water flowing down the river. Plant operator MVM explained that sufficient water could theoretically have been available for cooling, but the water level had fallen below the effective suction elevation of the existing pumps. Running units require roughly 100 cubic metres of water per second, while four shut units require no more than approximately 2.5 cubic metres per second for residual cooling. The plant was operating at a river level 28 centimetres below the previous 2018 record and more than one metre below the century-minimum level assumed when the station was designed more than four decades ago.
Hungary has responded with an emergency river-engineering programme. The government approved a bed sill requiring approximately 145,000 cubic metres of rock, with two 80-metre barges prepared as an additional temporary measure. The intervention is expected to cost around HUF6.1 billion, while the government estimates that a complete Paks shutdown would impose a burden of at least HUF50 billion per month. Even allowing for uncertainty around the latter estimate, the comparison illustrates the economics of climate adaptation: relatively modest investment in intake security can protect a generating asset whose monthly replacement-power cost is many times larger.
Farther downstream, the same Danube shortage is suppressing Serbian hydropower production. Elektroprivreda Srbije reported that Đerdap 1 was producing only around 20% of its usual output, with historically low inflow of approximately 1,400 cubic metres per second, while Đerdap 2 was operating at roughly 30%. Đerdap 1 has 1,140 MW of installed capacity and an average historical Danube inflow of around 5,370 cubic metres per second, meaning current flow has been only about one-quarter of the long-term average.
EPS expected Serbian electricity demand to approach 100 GWh per day during the heatwave and said approximately 10% of requirements were being procured from the market to replace missing hydroelectric generation. Thermal plants were carrying more of the domestic load, while EPS attempted to preserve reservoir stocks and coal inventories ahead of winter. Serbia is therefore exposed twice: directly through reduced output from Đerdap and indirectly through the loss of nuclear generation in neighbouring Hungary and Romania, which limits the amount of inexpensive regional electricity available for import.
This regional correlation weakens the conventional assumption that imports will always compensate for domestic weather-related outages. Interconnectors can move electricity from one bidding zone to another, but they cannot create additional generation when several interconnected systems face the same hydrological event. Romania, Hungary and Serbia may all seek imports during the same evening hours. Austria, Slovakia, Croatia and Bulgaria then become part of the same price-formation chain, while transmission congestion determines where the highest scarcity premium appears.
The European system as a whole remains adequately supplied. ENTSO-E’s pre-summer assessment found no broad systemic adequacy threat for most of Europe, supported by expanding renewable generation, stronger cross-border coordination and battery capacity that had doubled to approximately 29 GW. Nevertheless, expected European hydro storage at the start of the season was already 18% below June 2025, while summer demand was forecast to be 2.5% higher year on year. The subsequent deterioration in river conditions illustrates the difference between having sufficient aggregate capacity and having that capacity available in the correct region and hour.
France provides an important counterpoint. Water constraints do not make nuclear generation structurally unreliable. French electricity production reached 284.3 TWh in the first half of 2026, an increase of 4.6%, with nuclear output rising by 7.9 TWh and average nuclear availability improving to 73.9%. RTE nevertheless acknowledged that heatwaves and prolonged drought can temporarily restrict individual river-cooled reactors because of water-flow and discharge-temperature rules. At fleet level, RTE considers these losses secondary in volume, but they can still influence prices when they coincide with heat-driven demand and low wind production.
The market impact is most visible in the shape of hourly prices rather than in monthly averages alone. The International Energy Agency expects EU electricity consumption to increase by 2% in 2026, supported by electrification and higher cooling demand. At the same time, average EU spot prices in the second quarter were already more than 30% above the previous year, partly because of higher gas-generation costs. Water-related nuclear and hydro restrictions have added another scarcity factor to a market that was not starting from a low-cost fuel environment.
On the Hungarian HUPX market, the baseload price for delivery on 15 August was €137.94/MWh, while the conventional peak-load block averaged only €101.59/MWh. Intraday trading for 16 August showed prices around €40–60/MWh during solar-rich midday periods before rising towards approximately €200/MWh during the evening. Serbia displayed an even clearer profile on 15 August, with the SEEPEX day-ahead price falling to €0.01/MWh at noon and reaching €182.90/MWh late in the evening.
Romania showed the same inversion. The OPCOM baseload block for 16 August was RON717.86/MWh, but the daytime peak block was only RON624.92/MWh, while the late-night block reached more than RON1,046/MWh. In July, the Romanian day-ahead market had already averaged approximately €120/MWh. These figures describe a market with surplus solar power during the middle of the day and increasingly scarce firm supply after sunset. The traditional assumption that “peak” hours are necessarily the most expensive is becoming obsolete in solar-heavy markets.
The IEA reported that the spread between midday lows and evening highs reached as much as $600/MWh in several European markets during the June heatwaves. Spain recorded negative prices in 17% of wholesale hours during the first half of 2026, up from 10% in 2025. Europe can therefore experience excess electricity and acute scarcity within the same day. Average baseload prices conceal much of the commercial value now migrating towards batteries, pumped-storage plants, flexible hydropower, demand response and fast-starting thermal generation.
More solar capacity will help cover daytime air-conditioning demand, but solar alone cannot replace nuclear and hydro output between approximately 19:00 and 23:00. Without storage, additional photovoltaic generation can deepen midday price compression while leaving the evening ramp largely unchanged. Wind has a different production profile, a generally higher capacity factor and a lower correlation with solar output, giving it greater diversification value. It must nevertheless be assessed separately because major heatwaves can coincide with low-wind conditions across several European markets.
The investment response should therefore combine generation, flexibility and water resilience rather than rely on a single technology. EPS has identified a 1 GW solar project, the proposed Niš gas plant, the Bistrica pumped-storage project and the longer-term Đerdap 3 development as elements of Serbia’s security-of-supply strategy. Hungary is accelerating wind, storage and grid development, while Romania’s nuclear programme will increasingly have to integrate cooling-water resilience into the economics of refurbishing Cernavodă Unit 1 and developing additional units.
An illustrative Southeast European resilience programme for 2027–2032 would require approximately €4–7 billion. Around €300–800 million could be directed towards deeper nuclear and thermal intake structures, pumping systems, cooling optimisation, dredging and river-control works. A further €800 million–€1.5 billion could cover hydro-turbine rehabilitation, reservoir digitalisation and improvements that produce more electricity from each cubic metre of water. Approximately €1–2 billion could support 2–4 GWh of battery storage and industrial demand-response infrastructure, with the remaining €1.5–2.7 billion allocated to pumped storage, transmission reinforcement and strategically congested interconnectors. These are analytical ranges rather than announced project budgets.
For a two-hour battery located in a volatile Central or Southeast European node, an illustrative model using all-in investment costs of €250,000–€400,000 per MWh, 250–300 annual equivalent cycles, a net captured spread of €90–130/MWh and additional balancing-market income could support a nominal equity IRR of approximately 12–16%. Returns could rise towards 15–20% where evening scarcity remains persistent or capacity payments are introduced. A normalisation of spreads to €50–70/MWh would reduce the return towards the high-single-digit range. A 12–18 month grid-connection delay could cut equity IRR by approximately 2–5 percentage points, while increasing effective project cost by 5–10% through interest during construction, equipment escalation and deferred revenue.
Hydropower investment requires a different valuation. A reservoir HPP may produce fewer annual megawatt-hours during dry years but earn more per megawatt-hour by concentrating production in high-price evening periods. Its value increasingly comes from flexibility, balancing services, black-start capability and avoided scarcity purchases rather than baseload energy volume alone. Run-of-river assets lack much of this optionality and should be valued using more conservative climate-adjusted production assumptions.
Nuclear projects also need a more explicit hydrological risk premium. Plant valuations, availability guarantees and long-term power-purchase contracts should incorporate intake depth, river-bed erosion, minimum-flow forecasts, water-temperature restrictions and the cost of alternative cooling arrangements. Historical capacity factors based on twentieth-century river conditions are no longer sufficient for debt sizing or contracted-output guarantees.
The immediate outlook remains dependent on rainfall across the Alps, Central Europe and the Danube basin. Significant precipitation could allow nuclear units to restart and lift hydroelectric output, while declining temperatures would reduce cooling demand. Yet depleted soils, low river flows and partially exhausted reservoirs do not recover instantly. Even after the current operational emergency passes, utilities will enter the autumn with greater sensitivity to reservoir preservation, nuclear availability and the cost of replacement electricity.
Europe’s electricity market is consequently assigning an increasingly visible premium to water that can be converted into a reliable megawatt-hour after the sun sets. The most valuable assets are no longer simply those capable of producing the largest annual volume, but those able to remain available during the narrow evening periods when heat, low river flows and declining solar generation converge.
