Image source: Hungary to shut sole nuclear plant for first time on Sunday 

One of Europe’s largest and most famous rivers, the Danube, is experiencing record breaking low water levels. The previous months’ historic heat wave, combined with months of low average rainfall, have contributed to dangerously low water levels. The resulting shutdown of the Paks nuclear power plant has left Hungary with a multitude of energy dilemmas, all while costing the country millions. Aside from the Paks nuclear plant, ship traffic, tourism, and even other nuclear power plants are all being hit hard. While the crisis has been tempered thus far, it can certainly happen again, and how hard will the financial toll hit when it does? 

Why This Matters 

• The Danube River water levels are the lowest they’ve ever been.  

• On July 26th, the Hungarian day-ahead energy prices were at around €92/MWh, a normal price for a hot summer day. However, as the river fell towards the critical levels, prices climbed above €150, peaking at €186.90/MWh on August 4—roughly double the peak prices seen in the preceding weeks. 

• Forecasting models are falling behind climate change risk. 

The Science: The Nuclear Plant is Water-Run 

A nuclear power plant and reactor generate electricity the same way a coal or gas plant does: by boiling water, producing steam, which powers a turbine. Most nuclear power plants in Europe cool the heat generated through recirculating cooler water already stored in cooling towers, but the Paks plant does it a different way: it takes water straight from the Danube River, cools the plant, and deposits it downstream. This has historically worked well, but it has depended on one key assumption: the Danube is too large of a river to ever run short of water. 

This assumption failed, not because there wasn’t enough water, but because the river’s surface dropped so drastically and rapidly, due to drought, months of below-average rainfall, and the recent heat waves plaguing Europe. The physical structure of the Paks plant couldn’t reach into the water, thus creating a positive feedback loop built on a grid stressed by heat, and the reactor not able to generate energy to compensate for it, resulting in energy crises across multiple locations.  

Details: The Danube River Water Level Lowest Ever 

Figure 1. Daily Danube gauge level at Dunafoldvar, near the Paks power plant. Summer 2025 (gray) versus summer 2026 (red). Data pulled from LSEG Workspace/Refinitiv. 

In 2025, the Danube rose and fell over a long period of time, oscillating around its’ zero mark. It even spiked up over 2 meters with August rain. The summer of 2026, however, told a drastically different story: it never came up at all. Already over a full meter below its average level, long periods of lack of rainfall and a brutal heatwave in mid-July resulted in a historic low.  

At its lowest point, a massive difference in water levels near the Paks water plant resulted in the Danube being 459 cm lower than the previous summer, not only shattering the record set in 2018, but also crossing a vital threshold: the river was not able to provide a direct water source to the Paks plant. The prevailing action resulted in the plants shutdown, with only one singular turbine available for power generation. 

Details: Energy Prices Soar 

Figure 2: Hungarian day-ahead electricity prices throughout the duration of the Paks crisis. Prices pulled from LSEG Workspace/Refinitiv. 

• The day-ahead energy prices tell the story. On July 26th, the Hungarian day-ahead energy prices were at around €92/MWh, a normal price for a hot summer day. However, as the river fell towards the critical levels, prices climbed above €150, peaking at €186.90/MWh on August 4—roughly double the peak prices seen in the preceding weeks. The forward market moved even faster and with the Hungarian week-ahead contract, assessed at €217.90/MWh on 30 July, it was trading at €250–263 within a day as the shutdown became “inevitable”. 

The consequences were immediate, with Hungary taking extreme action to secure the energy reserves available and asking large industrial consumers to reschedule production and electricity imports, furthering increasing costs. During a critical time window, major users of the high voltage grid were asked to curb electricity use, as well as functions in governmental buildings. Additionally, requests to the public were made to move electricity-intensive activities during the peak period, including the use of air-conditioning units, which further human health risks. 

With climate forecasts showing that these events will only become more frequent with time, and Europe continuing to be the fastest-warming continent in the world, these heat waves and their growing intensity and frequency mean that crisis like this one are sure to grow as well. 

Details: Forecasting Models are Falling Behind Climate Change Risk 

Figure 3: LSEG Power Research, Hungarian pricing forecast versus actual, over the period of the time of the crisis. Actual prices (blue) compared to the forecast (red and pink). Pulled from LSEG Workspace/Refinitiv. 

The visual here, taken as a snapshot from Refinitiv’s forecasting tool, highlights the difficult nature of predicting how climate change affects them. From July 30^th onwards, the actual interlay prices (blue) ran above the actual predicted forecast (pink). The model was pricing for a normally hot summer, but not for the main source of energy to lose its ability to use water. The gap peaked during the crisis (August 3-4), when prices nearly doubled the forecast. Interestingly, the model then rebounded and was higher than the actual prices (Fig 2) ended up being in the period after, meaning the risk never materialized. In essence, not only does the model not properly map out potential disasters like the one early in August, but it also overestimates the effects on the way out, highlighting the key idea that physical climate risk simply moves too fast for current tools to predict.  

Take Action 

Map water risk and dependencies:  

• The Paks crisis revealed that water is quickly being exposed to a wide variety of climate-related risks, some that are not simply just the fault of a lack of attention, but that of previous structural and fundamental design (the intake of water at Paks). 

Thresholds are key to climate risk:  

• The Paks facility failed because the water levels reached a previously unheard-of level; the actual structural design is what called the failure. While previous tools track information such as gauge level, dryness, temperature etc., they do not account for when assets, like Paks, simply fail because of thresholds (water level) 

Utilize frameworks that aim to price nature-related dependencies: 

•  Frameworks like TNFD give instructions to organizations to learn how to disclose nature-related dependencies, rather than simply relying on tools that are not equipped to handle the dynamic nature of climate risk.