Take a glass of tap water. Put it on the table. Look at it.
This glass was drawn from a river or an aquifer somewhere. Treated. Carried for kilometres through pipes that will lose 20% of it on the way. It will reach you. You will drink it. And a few hours later it will leave again through a sewer network, be treated a second time at great expense, then discharged into a river, with whatever the treatment has not removed.
This glass of water will have passed through three industrial systems to complete a journey that nature has been making in a closed loop for four billion years.
I am not describing an anomaly. I am describing the normal workings of hydraulic France in 2026.
Every day, this country's public network withdraws 15.6 million cubic metres. Before reaching a tap, 3 million vanish in leaks, in pipes whose average age is over fifty years. What remains supplies 150 litres per person per day. At the end of the chain, cities alone discharge 11.5 million cubic metres of waste water a day. Add the businesses, public administrations, hospitals and food-processing industries that go through the same collective sewerage networks: the total volume of waste water produced in France exceeds 20 billion cubic metres a year. That is 55 million cubic metres a day. 385 million a week. 1.6 billion a month.
Every day we produce the equivalent of 22,000 Olympic swimming pools of collected waste water. Every day. Without interruption. For decades.
We call it waste.
The misreading of the figures
The statistics department of the French Ministry for Ecological Transition has just published the final figures for 2023. France withdraws 29.4 billion cubic metres of fresh water a year. Reading the details is a statistical slap in the face.
The largest user is not agriculture. It is the cooling of power stations, at 45%. Next come navigation canals, at 20%. Drinking water follows at 18%. Agriculture brings up the rear at 10%, industry at 7%.
These two giants, energy and navigation, return almost all the water they withdraw. A few hours later, warmed, it flows back into the river.
Agriculture, on the other hand, returns nothing. It cannot. Plants eat water. They turn it into living matter, into leaf, fruit and root. They evapotranspire it to the sky. That is the metabolism of living things, not an agronomic fault.
And that is precisely why, with only 10% of gross withdrawals, agriculture accounts for 61% of the water actually consumed in France, the water that does not return to the environment. In basins such as the Adour-Garonne, the proportion rises to 83%.
This 61% figure, confirmed by the French Biodiversity Office (OFB) and taken up by WWF in July 2026, is the key to the whole French water crisis. A farming sector that irrigates only 7% of the country's utilised agricultural area mobilises two thirds of the water consumed in France. Not through excess. Through necessity.
Now let us look at what this agriculture actually draws from our aquifers and rivers every day: 8 million cubic metres. That is the gross figure, for the whole sector, all crops together. The volume farmers go and fetch from deep in the ground, the volume for which storage reservoirs are built, the volume that crystallises conflicts of use and splits regions apart.
Agriculture draws 8 million cubic metres a day from our aquifers. Our sewerage networks collect 20 million that could be treated for reuse. Two and a half times as much. Which we let dissipate into rivers.
To these 20 million cubic metres of collected urban waste water must be added the direct industrial discharges into the natural environment, outside the sewerage network: effluents from paper mills, foundries and the food industry, loaded with heavy metals, solvents and PFAS that conventional treatment plants do not remove. The total volume discharged every day into our rivers and aquifers exceeds 55 million cubic metres. For every cubic metre a farmer pumps from an aquifer, our cities and industries discharge seven into a river.
The debate about storage reservoirs is not a debate about water. It is a debate about our inability to connect what we waste with what we need.
What the aquifers record
The water table is the slow memory of water. It does not react to yesterday's storm. It integrates the seasons, the years, the decades. It reads the land on a different time scale from ours.
In May 2026, the French geological survey (BRGM) published its first seasonal warning bulletin. Despite a winter recharge judged satisfactory, a single dry month of April had been enough to tip 79% of monitoring points into a falling trend. The soil had closed up too quickly. Rainwater had run off without soaking in.
In July, the same BRGM found that 93% of points were now falling. 54% of aquifers had dropped below seasonal norms, against 42% on the same date the previous year. The large, high-inertia aquifers of the Paris Basin and the Rhône corridor, which form the country's deep strategic reserve, continue a slow decline that summer storms do not correct. These aquifers recharge in winter. Not in July. Storm rain falls too fast, too hard, on soils that no longer recognise it.
At the end of July 2026, 94 of the 96 departments of mainland France were under restriction orders. A broad diagonal (Maine-et-Loire, Vienne, Puy-de-Dôme, Creuse, Loir-et-Cher) had moved to Crisis level. Agricultural withdrawals there were partly suspended. Water is being rationed to grow food.
And what we drink reflects the state of what we discharge. In 2024, 29% of the French population was exposed to pesticides or PFAS in their tap water at levels above the European quality limits. 28.5% of the water distributed failed to comply on pesticides alone. A joint inspection report by IGEDD, IGAS and CGAAER, commissioned by four ministries and published in May 2026, puts the cost of curative depollution at between 6.15 and 11.3 billion euros a year. We pay to treat what we ourselves have introduced into the water cycle.
This is not a climate anomaly. It is the telltale sign of a model that can no longer carry the load.
What we destroyed without seeing it
There is a third player in this equation, a silent one, which we spent a century eliminating without understanding what it did.
France has lost 67% of its wetlands since 1960. Marshes drained, floodplain meadows filled in, riverside woodland torn out to gain a few metres of farmland. What looked like progress was in fact the methodical destruction of the only system able to regulate the water cycle naturally at the scale of a catchment.
A wetland does not only store surface water. It recharges aquifers slowly, through continuous infiltration, where urban sealing and agricultural runoff only rush water downstream. It filters. It buffers floods. It sustains low flows in summer by releasing what it absorbed in winter.
And it stores carbon. Between 500 and 1,500 tonnes of CO₂ per hectare, against 100 to 300 tonnes for a mature forest. When a wetland dries out, it flips: from carbon sink, it becomes a source. It emits CO₂ and methane. It worsens the warming that worsens the drought that worsens the drying out of aquifers.
We have set off a loop. And we keep feeding it.
The equation laid bare
On one side, an agriculture structurally bound to consume water to feed the country, drawing on aquifers that are emptying faster and faster. On the other, cities, businesses, public administrations and offices that every day pour out 55 million cubic metres of discharged waste water, 20 million of it collected and treatable, seven times the agricultural withdrawal, dissipated into rivers after partial treatment. And beneath all this, the natural safety net, wetlands, cut by two thirds in sixty years, now unable to play its buffering role or to lock up the carbon that warming releases.
The World Bank has put a figure on the cost of this disconnection: minus 6% of GDP around the Mediterranean by 2050. The European Commission estimates the Union's climate losses without adaptation at 170 billion euros a year. Food inflation linked to water stress will reach between 0.92 and 3.23 additional points a year by 2035, according to models published in Nature in March 2024.
Yet the total volume withdrawn has fallen by 16% since 2008. Industry and energy have done their part. It is now the urban cycle, the agricultural cycle and the restoration of wetlands that form the three critical fronts, and no large-scale policy treats them as facets of the same problem.
Solutions have of course been imagined. Separating grey water from black water at source, showers on one side, toilets on the other. The idea is sound. It runs up against a housing stock of thirty million homes built without this distinction, where completely renewing the internal pipework would cost more than the homes are worth. Collecting rainwater from every roof. Possible at the margin, insufficient at scale. These paths are not wrong. They are prisoners of a paradigm that treats water as a flow to be managed rather than a resource to be kept circulating.
Some will object that the city produces its water all year round while agriculture only needs it in summer. That forgets that restored wetlands do exactly the work of winter storage that reservoirs clumsily try to imitate, without concrete, without conflict, and returning the water through slow infiltration rather than evaporating it under a plastic liner. Some will object that cities are far from the fields. True, and the infrastructure needed costs no more than the sum of the reservoirs built or planned across the country. Some will object, finally, that capturing treatment plant discharges will dry up rivers. That ignores the fact that water reused to recharge a wetland returns to the river underground, sustaining low flows naturally rather than putting them on an artificial drip. The Landes are the clearest illustration: a pine forest planted on a former marsh, which burns and dries out, and in which restored islands of wetland would change both the water balance of the basin and the summer fire risk.
The resource exists. It is already circulating. It passes under our feet every day, through our drainpipes, at the rate of 22,000 Olympic swimming pools.
What is missing is not water.
It is the change of paradigm needed to make it work twice.
The next article will explore why this shift is now possible, and what makes it inevitable.
Sources: SDES, final 2023 data (July 2026) · BRGM, hydrological bulletins May and July 2026 · OFB / WWF, net consumption by sector (July 2026) · World Bank, High and Dry, 2016 · Kotz et al., Communications Earth & Environment (Nature), March 2024 · JRC PESETA IV, European Commission, 2020 · Choisy G., Collin E., Debrieu-Levrat C., Sandeau-Gruber A.-C., Crossnier G., Buchter J., « Dépollution de l’eau potable : stratégies et pistes de financement », joint inspection report IGEDD no. 016440-01 / IGAS no. 2025-081R / CGAAER no. 25106, May 2026.