France's nuclear fleet takes the heat as a record July scorches the grid
Three reactors offline and eight more throttled as a July heatwave stretches EDF's cooling capacity, exposing the engineering ceiling of a fleet that supplies most of France's power.
France cut power at three nuclear reactors on 13 July 2026 and began reducing output at eight more, after record-breaking river temperatures forced the country's operator to throttle cooling systems that cannot legally discharge water warm enough to harm fish. The shutdowns and deratings, announced by Electricité de France (EDF) and surfaced via the Polymarket wire at 15:03 UTC, are the clearest signal yet that the climate ceiling on the world's most nuclear-dense grid is no longer theoretical.
France's 56 commercial reactors normally deliver around 70 percent of the country's electricity. Their rivers are the heat sink. When those rivers warm, the reactors must run quieter or stop. Each summer that constraint is binding for longer.
The cooling problem, in plain engineering
Nuclear reactors do not consume water the way a coal plant does; they expel the heat they generate into whatever large body of water sits next to them. French law, like the European Union's broader water framework, sets a temperature ceiling on what a plant can return to the river. Cross it and aquatic life suffers. EDF has to derate, or in the worst cases, drop offline entirely.
The reactor cooling constraint is not a story about reactor design failing. It is a story about a grid built around a resource that climate change is quietly rewriting. The 2022 and 2023 heatwaves forced similar deratings along the Rhône and Loire. French nuclear output has never quite recovered to its mid-2010s baseline, even after the corrosion crisis at the end of last decade was supposed to be in the rear-view mirror.
The operative question is no longer whether nuclear plants can run during a heatwave. They can, at reduced load. The question is how often, and for how long, the throttling erodes the baseload advantage that made nuclear so attractive in the first place.
How much the fleet actually lost
EDF has not yet published a single unified figure for the 13 July reductions, and the Polymarket-flagged advisory did not specify megawatts curtailed. Independent French press routinely tracks the country's daily nuclear output through RTE, the grid operator, where the disponible and produit graphs move in near-real time. The pattern, on a hot mid-July day, is a curve sloping downward through the afternoon as Rhône, Garonne and Moselle readings hit their limits.
That opacity matters. A country that depends on a single technology for most of its low-carbon power does not have a great fallback when that technology's output is constrained. Hydropower, in a dry summer, is itself depleted. Gas peakers are available, but they import the price shock that nuclear baseload is supposed to suppress. Solar peaks precisely when demand also peaks for cooling, and the country has spent a decade building far less solar capacity than either Spain or Germany in per-capita terms.
What the grid actually does next
RTE has spent years drawing up the contractual and demand-response architecture for a French summer that might run short. Industrial users can be paid to switch off. Imbalances can be settled against the European coupled market. That machinery works in kilovolt-hours. It does not work as a substitute for the missing terawatt-hours that a derated nuclear fleet simply does not produce.
The longer-run answer is one France's energy planners have been quietly discussing since the 2022 stress. New reactors, of the EPR2 design, are queued. They are not yet pouring concrete at the pace assumed. Small modular reactors are still a sales deck. The fleet that has to carry the next decade is, by and large, the fleet France already has. It is also the fleet most exposed to a warming river.
Stakes beyond France
Western Europe's deepest low-carbon baseload sits on a riverbank. If that baseload oscillates more than markets expect, the price signal travels: Italian and German imports from France look less reliable, German gas peakers run more hours, and the continent's combined nuclear output, which jumped when Czech and British plants extended lifespans, slips against the rising demand from electrification and cooling. None of that is a crisis today. All of it is the slow background music against which Europe's climate and industrial policy has to be composed.
The honest reading is that the 2026 event is not a freak. It is the operating envelope. France does not have to abandon nuclear to face a structural problem; it has to price, build and legislate around a fleet whose output will be more variable than the textbooks promised. That work has begun. It has a lot of summers left.
Desk note: the Polymarket flag at 15:03 UTC is treated as a single-wire trigger, not as a primary source for the underlying EDF dispatch notice. Independent French press and RTE's published production indexes will be the citation layer once the day's specific deratings are confirmed. Where that confirmation is not yet in hand, this piece states the constraint, not the number.
Wire provenance
This editorial synthesis draws on the following public wire/social posts:
- https://x.com/polymarket/status/...
- https://en.wikipedia.org/wiki/Nuclear_power_in_France
- https://en.wikipedia.org/wiki/%C3%89lectricit%C3%A9_de_France
- https://en.wikipedia.org/wiki/RTE_(electricity_transmission_system_operator)