On 28 and 30 July 2026, Romania did something its nuclear operator had never done before in three decades of commercial operation: it shut down both reactors at the Cernavoda nuclear plant at the same time, on purpose, because the Danube no longer had enough water to safely cool them. Unit 1, in service since 1996, and Unit 2, in service since 2007, went offline within two days of each other, removing about 1,400 MW, close to 10 percent of the country’s electricity production, in the middle of summer peak demand.
For offtakers, traders and investors who model Romanian summer generation on the assumption that hydro and nuclear are the stable, weather-independent part of the stack, this is the event that breaks that assumption. Hydro and nuclear did not fail for two unrelated reasons this summer. They failed for the same reason, at the same time, because both draw on the same river.
What actually happened on the Danube this summer
The Danube’s flow at Baziaș, where the river enters Romania, was already running well below normal by mid-July, and over the weekend of 19 and 20 July it dropped to its lowest level since 1996, roughly three decades. By 28 July, flow near Cernavoda had fallen to around 1,630 to 1,650 cubic meters per second, against a normal July average of about 4,700 to 4,750 cubic meters per second, meaning the river was running at roughly one third of its seasonal norm. Forecasts at the time pointed to a further drop toward 1,500 cubic meters per second by early August. For reference, the lowest flow ever recorded at Romania’s entry point was 1,400 cubic meters per second, in 1985.
Authorities responded on multiple fronts at once: Level III water-use restrictions were imposed on irrigation along the Călărași to Cernavoda section, river ferry and barge traffic was disrupted, and Apele Romane, Romania’s water authority, coordinated with Hidroelectrica to release water from the Olt and Argeș river reservoirs specifically to help maintain the flow Cernavoda needed for cooling. Hydro reservoirs, in other words, were already being drawn down to protect the nuclear plant before the plant itself had to shut down.
Why Cernavoda had to shut down, twice
Cernavoda’s reactors use Danube water to cool three pumps essential to safe operation. When flow at the plant dropped to roughly 1,630 cubic meters per second on 28 July, Nuclearelectrica, the state-owned operator, took Unit 1 offline as a preventive measure, stating plainly that the move was necessary to protect the equipment tied to the cooling pumps. Unit 2 followed within two days, disconnected progressively from the national grid, taking the combined loss to close to 1,400 MW, roughly 706.5 MW from Unit 1 and 704.8 MW from Unit 2. Nuclearelectrica described this as the first time both units had been voluntarily and simultaneously shut down as a precaution since the plant became fully operational.
The immediate system impact was significant. With peak national demand around 7,300 MW and domestic production limited to roughly 4,000 to 4,300 MW during the outage, Romania had to cover the gap through imports, a gap that widens further in the evening once solar output falls away. National battery storage, at just over 600 MW, covers only about 15 minutes of national consumption, nowhere near enough to smooth an outage of this size. Authorities said household and business supply would not be interrupted, but the cost of closing that gap did not disappear, it moved into the wholesale market.
The price the market has already paid
Romania got an earlier preview of what a Cernavoda disruption does to prices during a separate reactor outage in May 2026. Day-ahead prices on OPCOM, Romania’s power exchange, rose from roughly 461 lei per MWh before the outage to a peak of 761 lei per MWh, about 65 percent above the pre-crisis baseline, with the week’s average landing 44 percent above the prior week. Analysts estimated the elevated pricing added on the order of EUR 60 million to the market’s costs over the following month, based on daily spot volumes of roughly 36,000 MWh trading at rates 250 to 350 lei above seasonal norms. During the July shutdown, imported power at peak hours reportedly touched over EUR 1,000 per MWh, a level that turns a generation shortfall into a direct and immediate cost for anyone short on hedged volume.
Otilia Nuțu of Expert Forum summarized the underlying condition bluntly: Romania’s spot prices rank among Europe’s highest largely because there is not enough cheap electricity available to meet demand when the flexible sources falter. A drought that hits hydro and nuclear at once is precisely the scenario that exposes that shortage.
Why hydro and nuclear are not two separate risks, they are one
Standard generation-risk models tend to treat nuclear as weather-independent baseload and hydro as a separate, weather-sensitive but largely uncorrelated source. The 2026 Danube drought shows why that separation does not hold in Romania. Cernavoda’s cooling requirement ties its availability directly to the same river flow that determines hydro output on the Olt, Argeș and Danube systems. When the river runs low enough to threaten hydro generation, it is also running low enough to threaten nuclear cooling, at the same time, in the same season. A portfolio or a hedge book that prices these as independent risks is understating the true tail risk of a dry Romanian summer.
This is not a one-off. Hidroelectrica’s 2025 was already its driest year on record, with output expected just above 11 TWh, arguably worse than 2012, the year the company came close to insolvency, though the business stayed profitable, around EUR 590 million, precisely because scarcity pushed prices higher even as volumes fell. Reservoir levels heading into the 2026 summer were already at their lowest for the period since 2018. The 2026 drought landed on top of an already-stressed hydro base, not a well-rested one, and forecasters were already flagging below-normal August precipitation, which threatens wind output as well, widening the correlated-weather-risk problem to a third technology.
What this means for summer hedges and PPA exposure
For offtakers and traders with Romanian exposure, the practical implications are specific. First, hydrological indices such as Danube flow at Baziaș and reservoir fill percentage deserve a place as explicit, correlated risk factors in hedge and portfolio models, not an afterthought footnoted under force majeure. Second, deemed-generation and volume clauses in summer PPAs that assume hydro and nuclear output move independently should be stress-tested against a scenario where both drop together, since 2026 shows that scenario is not a theoretical tail case, it is happening in real time. Third, import price exposure needs a realistic tail assumption, not the seasonal average, given that spot prices have already touched quadruple-digit euros per MWh this year during comparable outages. Fourth, the near-total absence of grid-scale storage, just 15 minutes of national demand, means batteries are not yet a meaningful buffer against this kind of correlated shortfall, which is as much an investment gap as it is a risk factor.
There is also a longer horizon to price in. Cernavoda’s Unit 1 is separately scheduled for a major refurbishment, a EUR 2.9 billion project that Nuclearelectrica has confirmed will take the reactor offline for an extended period starting around the end of 2027, independent of any future drought. Hedging models built only around this summer’s shock will miss that structural reduction in nuclear capacity arriving on a fixed schedule less than eighteen months from now.
Does this make Romania a riskier market, or a more interesting one
The honest answer is both, and the two are related. This summer exposed a real, previously underpriced correlation between Romania’s two largest baseload technologies, and any investor or offtaker ignoring that correlation is carrying more risk than their models show. At the same time, the same event demonstrates that Romania’s market design lets scarcity flow directly into price, which is exactly why Hidroelectrica stayed profitable through its driest year on record, and it is why capital positioned for volatility, flexible generation, storage, import capacity, diversified supply portfolios, has a clearer, better-priced opportunity in Romania than in markets where regulators suppress scarcity pricing instead of letting it signal where investment is needed. The crisis has already reopened debate on long-delayed infrastructure, including the proposed Iron Gates III hydropower project, which suggests the policy response is at least starting to catch up with the risk.
Momentum Energy’s View
We think this summer should permanently change how Romanian generation risk gets modeled. Treating hydro and nuclear as two independent, largely weather-proof pillars of the summer stack was always an approximation, and the Danube just showed how much that approximation can cost when it breaks: a historic dual-reactor shutdown, a real market-wide price shock, and imports at levels that punish anyone under-hedged. The businesses that come out ahead from here will be the ones that explicitly price hydrological correlation into their Romanian hedge books, and the ones building the flexible generation, storage and import capacity this market is now visibly short of. We are watching the Danube’s flow at Baziaș and Hidroelectrica’s reservoir levels as closely as we watch the forward curve, because right now, they are the same signal.
If you hold Romanian summer exposure, how is your book currently pricing the correlation between hydro and nuclear availability? We would be interested to hear how others are stress-testing this after July.
