Atomeromu.mvm.hu
Prime Minister Péter Magyar said Wednesday that Hungary will immediately begin laying 150,000 cubic meters (5.3 million cubic feet) of rock into the Danube to build a submerged riverbed dam near the Paks Nuclear Power Plant — and that two 80-meter (262-foot) barges may be deliberately sunk across the river if the structure is not finished fast enough. The announcement, made after an on-site cabinet meeting at the plant, comes with a blunt cost comparison baked in: the intervention is estimated at HUF 6 billion (approximately $19 million), while a complete shutdown of the plant would cost Hungary’s state budget at least HUF 50 billion per month (approximately $158 million per month). The numbers make the engineering gamble look cheap. But they do not explain why it became necessary in the first place.
The short answer is that the water was always there. The pumps just were not deep enough to reach it.
Pumps That Couldn’t Reach the Water
The Paks plant’s four VVER-440 Soviet-designed reactors cool themselves by drawing water from the Danube into a once-through cooling system — water in, heat absorbed, warmer water returned to the river downstream. For four decades, with the Danube running at normal depths, the intake system worked reliably.
This summer, the Danube fell below the suction inlets. Not slowly or ambiguously — the river surface physically dropped below the point where the plant’s pumps could draw water in. Hungary’s plant operator, Paksi Atomerőmű Zrt., confirmed the constraint plainly: a full shutdown had become unavoidable because the water level is lower than the suction pipes used to extract water from the river. Work to lower those suction pipes is now underway, the company said, but the process will take years.
Tamas Cseh, a member of parliament for the town of Paks, confirmed the engineering context in terms that leave little room for ambiguity: the Danube’s water, he said, would still be sufficient to cool the reactors had the pump inlets sucking up the water been set deeper below the surface.
This is what the 2026 Danube crisis actually is, stripped to its structural core: a plant running Soviet-era intake engineering in a 21st-century drought.
A Warning Raised in 2018 and Ignored Until 2026
The engineering failure was predicted. Attila Aszódi, the former government commissioner responsible for Hungary’s Paks expansion project, told reporters he had warned the government as early as 2018 that future Danube water levels and temperatures could create serious problems for reliable plant operation — particularly once the planned new reactors came online.
The new reactors are still years away. The warning is now eight years old. And Hungary is currently moving 150,000 cubic meters (5.3 million cubic feet) of rock into one of Europe’s great rivers because the intake pipes still have not been moved.
Magyar, for his part, framed the situation without pretense, saying in remarks at the plant: “We can no longer consider this extraordinary situation as a one-time natural extreme.”
How the Weir Works — and Why the Barges
The structure Hungary is building is called a bed sill, a submerged version of a weir — a low barrier in the riverbed that slows the current and backs up water upstream, raising the surface elevation on the plant’s side of the structure. Unlike a conventional dam, a bed sill is designed to remain fully submerged at normal water levels, causing minimal disruption during wet seasons while providing a critical water-level buffer during droughts.
Construction is scheduled to start Thursday between the Paks plant and the settlement of Dunaszentbenedek. Phase 1 will use 35,000 cubic meters (1.24 million cubic feet) of rock; Phase 2 adds 110,000 cubic meters (3.9 million cubic feet) more. The project is classified as a priority national investment, with 100 soldiers deployed for around-the-clock construction under the coordination of General Pál Kádár, head of the Defense Administration Office. Technical planning falls to the Budapest University of Technology and Economics.
The goal is to ensure the Danube’s water level at the plant’s cooling channel does not fall below negative 90 centimeters (negative 35.4 inches) — high enough to keep the pump suction inlets submerged. Magyar said the completed weir could raise the local water level by up to one meter (3.3 feet).
The barge contingency is the bridge solution if the river drops faster than the weir can be built. Two 80-meter (262-foot) barges will be transported to Paks — one arriving from Mohács as early as Wednesday night, the second by Thursday — and placed on standby. If the Danube level falls sharply before the permanent structure is complete, the government could deliberately sink both barges across the river to temporarily raise the surface by up to 20 centimeters (7.9 inches) at the cooling pumps. The decision on whether to sink them is expected by Friday.
Where Things Stand Now
The Danube’s level rose approximately 19 centimeters (7.5 inches) from its crisis low over the past week — enough to allow a partial output restart Monday bringing the plant from 240 megawatts up to approximately 500 megawatts. Normal capacity is 2,000 megawatts.
The breathing room is temporary. Magyar said water-management forecasts indicate the river could fall again to around negative 136 to negative 137 centimeters (negative 53.5 to negative 53.9 inches) within four to five days, and could drop below negative 140 centimeters (negative 55.1 inches) the following week without the intervention now underway. No significant rainfall is expected in the Danube’s upstream catchment area — the Alpine headwaters that feed the river — in coming days and weeks, meaning extreme low-water conditions could persist for months.
For context: the mandatory full-shutdown threshold under Hungarian environmental law is negative 134 centimeters (negative 52.8 inches). The plant already passed below that threshold in early August, completing its first full shutdown in 44 years before the recent modest recovery.
Once-Through Cooling: A Known and Unresolved Structural Vulnerability
According to Carbon Brief’s nuclear heatwave analysis, approximately 60 of the world’s 440 operating reactors use once-through river cooling — drawing water from a river, extracting heat, and returning it downstream at a higher temperature. The architecture was designed for historical river conditions: flows and temperatures that left sufficient thermal margin within environmental discharge limits. Under those conditions, it works. Under prolonged drought and elevated base temperatures, it does not — because the river has too little volume to absorb the waste heat without the discharge exceeding legally mandated temperature ceilings that protect aquatic ecosystems.
European regulators have documented this failure mode across multiple countries in 2003, 2019, 2022, 2023, and repeatedly throughout 2026. France’s nuclear fleet — which includes river-cooled plants on the Rhône, Meuse, and Garonne — has experienced output cuts in each of those years. Romania’s Cernavodă plant, which uses the same Danube, had its Unit 1 forcibly shut down on July 28 after just 23 days back online from routine maintenance. Romania’s navy detonated 180 kilograms (397 pounds) of explosives to blast submerged rocks and redirect Danube flow toward the plant’s cooling intakes on August 3 — an intervention that made international headlines. Hungary’s response is structurally similar but larger in scale.
A 2026 study in Energy Policy from McMaster compiled the first global inventory of weather-linked nuclear curtailments and found that the climate-related pressures on once-through cooled nuclear plants are growing as intake-water temperatures rise. No European government has announced a mandatory program to retrofit river-cooled plants with alternative cooling architectures — such as dry cooling towers that use air rather than river water — though each such retrofit is expensive and reduces plant operating efficiency.
Does Hungary Have a Permanent Fix?
The bed sill weir is not a permanent solution to the underlying problem. It raises the river to the pumps. The permanent solution, as confirmed by Paksi Atomerőmű Zrt., is to lower the pumps to the river — an infrastructure modification now underway that the company says will take a few years.
In the interim, Hungary will manage both: the weir provides a water-level buffer for drought conditions while the intake redesign proceeds in parallel.
What the Crisis Is Accelerating
Magyar announced Wednesday that Hungary will tender 700 megawatts of wind power generation capacity by August 31 and is targeting 4 gigawatts of total wind capacity by 2030. Hungary currently has approximately 330 megawatts of installed wind capacity — a figure unchanged since 2011, after a decade-long effective moratorium on new wind development driven by restrictive siting rules. The tender volume has since been raised to approximately 1,000 megawatts amid strong developer interest.
Energy Minister István Kapitány has separately noted that the 4 GW wind target is roughly equivalent to the nominal generating capacity of the four reactor units at Paks — the plant currently producing less than a quarter of that.
The connection Magyar drew between the Paks crisis and the wind expansion is not subtle. “The Hungarian economy cannot work without the plant for months,” he said last week. The emergency has made viscerally clear what energy security documents have said for years: a grid anchored to a single river-cooled nuclear facility, with no meaningful drought resilience built in, is a grid that can be brought to its knees by precipitation patterns.
What the River Has to Say About It
The Danube is not recovering. It rose 19 centimeters (7.5 inches) last week — a temporary reprieve attributed to rainfall in upstream Austrian catchments — but the gains are expected to be reversed within days. The river is operating in a fundamentally different hydrological regime than the one its bordering infrastructure was designed for.
Cernavodă’s Unit 2 in Romania remains under threat. Serbia’s Djerdap hydroelectric facility on the same river is operating at roughly 20% of its capacity. The drought that emptied the Danube is regional, climate-driven, and, according to a World Weather Attribution study released July 23, has become approximately 40 times more likely in eastern Europe than in the pre-industrial climate — driven not primarily by rainfall deficit but by the accelerated atmospheric evaporation that a warming climate produces.
Hungary is answering that systemic challenge with 150,000 cubic meters (5.3 million cubic feet) of rock, two barges that may be scuttled by the weekend, and a wind tender that has to be published before the end of the month. Whether the weir works, and whether the larger diversification it represents comes fast enough, is a question the Danube has given Hungary no choice but to answer.
Frequently Asked QuestionsWhy couldn’t the Paks plant’s pumps draw water even when the Danube still had water in it?
The plant’s cooling water intake system draws river water through suction inlets positioned at a fixed depth below the normal river surface. When the Danube fell far enough below its normal level, the surface of the river dropped below those inlet points. Even though water remained in the Danube, the pumps could not reach it — the river surface was physically below the intake mechanism. This is the specific engineering constraint that triggered the August 2 shutdown, and it is why the interim solution (building a submerged weir to raise the water surface back above the inlets) is being pursued simultaneously with the longer-term fix of lowering the inlet pipes to match current drought conditions. For technical details, see Hungary Today’s coverage and Argus Media’s reporting on the suction inlet situation.
Why hasn’t Europe retrofitted its river-cooled nuclear plants to remove this drought vulnerability?
The engineering answer exists: dry cooling towers that use air rather than river water can eliminate the once-through cooling constraint. The obstacles are cost and efficiency. Dry cooling towers are expensive to retrofit and reduce a plant’s operating efficiency because air is a less effective heat-transfer medium than river water. No European government has mandated a retrofit program despite documented climate-linked curtailments at French, Hungarian, and Romanian plants in 2003, 2019, 2022, 2023, and multiple times in 2026. The political calculus — paying certain costs now to avoid probabilistic future shutdowns — has consistently not been made. Researchers at McMaster University and the World Nuclear Association have both documented this challenge in detail. The 2026 drought season, with simultaneous shutdowns in two countries and a third requiring military explosives to redirect river flow, may change that calculus.
Was there any warning before 2026 that this could happen at Paks specifically?
Yes. Attila Aszódi, the former government commissioner who oversaw Hungary’s Paks expansion planning, reported warning the government in 2018 that future Danube water levels and temperatures could threaten reliable plant operation. MP Tamas Cseh of Paks has confirmed subsequent governments ignored the specific engineering fix — lowering the pump suction inlets — that would have mitigated the problem. The Paks operator has now confirmed that modification is underway, with a completion timeline of a few years. Meanwhile, Hungary is spending approximately $19 million to put 150,000 cubic meters of rock in the Danube to buy time for the fix that was recommended eight years ago.
What does Hungary’s wind power announcement have to do with the Paks crisis?
The connection is direct. A grid that relies on a single river-cooled nuclear plant for roughly half its electricity has a structural fragility that a drought can expose in a matter of weeks. Hungary’s wind tender — due by August 31, with a target of 4 gigawatts by 2030 — is being framed by the government as an energy security measure, not just a climate one. After a decade in which wind development was effectively frozen by restrictive siting rules, the Paks crisis has made visible what diversification is for: backup capacity that does not depend on a river staying wet.