There is no version of the energy transition that does not run through the ground. Grids, drivetrains and storage are built from a short list of minerals — lithium first among them — and in the International Energy Agency’s scenarios demand for them climbs for decades. Some projections put the rise in lithium demand at roughly sixfold by 2030, the great majority of it for electric vehicles. A country that means to decarbonise while keeping an industrial base needs secure, diversified access to these materials. That is not a slogan; it is arithmetic.
Yet almost the entire public conversation about new extraction has narrowed to a single register — local risk — while the question of what the material is for arrives late, if at all. Nowhere is that imbalance clearer than in the argument over water.
The water objection, in proportion
The objection is not frivolous. Most of the world’s lithium reserves sit in continental brines, and the largest producing deposit — the Salar de Atacama in northern Chile — lies in one of the driest places on earth, a closed basin whose only real outlet is evaporation. There, brine is pumped from beneath the crust into vast solar ponds, where some eighteen months of sun raise its lithium concentration from a fraction of a per cent to around six. The process is water-visible in the most literal way: widely cited figures put the brine evaporated at roughly half a million litres for every tonne of lithium carbonate, and a 2025 study reported the salt flat’s groundwater down more than ten metres over fifteen years, with the surface itself subsiding a centimetre or two a year.
Those numbers are real, and they deserve to be taken seriously. But they are not the whole picture, and this is where careful accounting matters. Basin-scale hydrological work on the same salar finds that the surplus evaporation attributable to mining comes to at most a few per cent of the basin’s total — on one recent assessment, no more than about four. A water table falls as the sum of many things at once: the region’s own aridity, long climate cycles, and every other user drawing on the system. Pinning the entire movement on the single most visible operator is rhetorically easy and hydrologically lazy.
Beneath all this lies a measurement dispute that serious researchers openly acknowledge: whether hypersaline brine — thick, lithium-bearing, undrinkable, useless for irrigation — should be counted as “water” at all. The international water-footprint standard, ISO 14046, gives no single answer, and studies reach opposite conclusions depending on the choice they make. A headline figure of half a million litres reads very differently once one asks whether the fluid in question was ever available to a farm or a village.
A water table falls for many reasons at once; naming only the most visible user is rhetoric, not measurement.
The methods are changing, too. Direct lithium extraction, now moving from pilot projects toward deployment, recovers lithium from brine without the great evaporation ponds, cutting precisely the losses that draw the most attention. A serious debate would track that trajectory rather than freeze the picture at its most dramatic frame.
The rule that should govern
None of this is a case for waving projects through. It is a case for symmetry. Where independent, reproducible hydrological data exist, they should decide. Where the only figures on offer come from a party with an interest in stopping a project, they deserve exactly the scrutiny we would rightly apply to numbers supplied by the operator. A recent Institut Kandrave review of extraction-impact claims put the point more bluntly: the credibility of a measurement cannot be separated from the incentives of whoever produced it. That principle cuts in every direction — which is precisely why it is worth defending.
Supply security, industrial resilience and the pace of decarbonisation are public interests too, and when they fail they fail on the same vulnerable populations that opponents invoke. The task for anyone serious about resources is to hold both columns of the ledger open at once.