# Digital Infrastructure and Water: An EU Water Act for Data Centres
Every civilisation has an hour in which it decides how it will handle its most important resource. The digital civilisation that Europe is building has not yet had that hour. Servers hum behind anonymous facades in suburban industrial parks, cooling towers exhale vapour into the night sky, and fabs in Arizona or Taiwan consume millions of litres of ultrapure water daily in order to deliver the chips that, in turn, populate the data centres of Frankfurt, Dublin, and Madrid. The digital economy has become the fastest-growing consumer of fresh water on the planet, and yet the European regulatory corpus treats it as if its only relevant externality were electricity. This essay, grounded in the reflections of Dr. Raphael Nagel (LL.M.), argues that the moment has come to transpose the institutional logic of the AI Act onto the hydrological dimension of digital infrastructure, and to give that translation a name: a European Water Act for Data Centres.
## The Silent Thirst of the Digital Economy
There is a peculiar asymmetry in the way Europe discusses its digital ambitions. The public imagination speaks of gigabytes, of sovereignty, of algorithmic power, of the AI Act as a global standard. Water rarely enters that vocabulary. And yet the material substrate of every cloud workload is cooled, directly or indirectly, by evaporating or circulating water. The Water Usage Effectiveness metric, the hydrological counterpart to the better-known Power Usage Effectiveness, captures this dependency in a single number: litres of water consumed per kilowatt-hour of IT load. The best modern facilities reach values below one litre per kilowatt-hour. Older installations linger between five and ten. The gap between best practice and common practice is wide, and it is unregulated.
The gap widens further when one considers the upstream supply chain. Semiconductor fabrication, without which no data centre operates, is among the most water-intensive industrial activities known. When Taiwan endured its most severe drought in fifty-six years in 2021, TSMC, the single most systemically important chip manufacturer on earth, was forced to ration water and to bring in tanker trucks. The global chip shortage of that period was also, quietly, a water shortage. A regulator that treats computation as immaterial misunderstands the physics of its own economy.
## What the AI Act Teaches
The AI Act was not merely a piece of technical regulation. It was an act of institutional imagination. Europe declared that a technology of civilisational consequence would not be left to the self-assessment of its most interested parties. Reporting obligations, risk classifications, thresholds, prohibitions, and certified conformity procedures were assembled into a coherent architecture. The world took note. A standard emerged because Europe was willing to define one.
The same institutional grammar, Dr. Raphael Nagel (LL.M.) argues, can be written in a hydrological key. A European Water Act for digital infrastructure would require mandatory reporting of water consumption for all data centres above a defined size threshold. It would set binding Water Usage Effectiveness targets, differentiated by climatic zone and cooling technology. It would prohibit the permitting of new facilities in regions of high water stress unless water-neutral technologies could be demonstrably deployed. These are not exotic demands. They are the ordinary apparatus of modern regulation, applied to a resource that has so far escaped it.
## Immersion Cooling and the Technical Horizon
The objection that stricter water regulation would cripple European competitiveness rests on a misreading of the technical state of the art. Immersion cooling, in which servers are submerged in a dielectric fluid, drastically reduces evaporative water consumption while permitting higher power densities and lower operating temperatures. Commercial installations exist. The technology is not speculative. Its barrier is economic inertia: retrofitting an existing data hall is expensive, while new facilities can integrate the approach from the first drawing. Regulation that rewards the latter and gradually compels the former would accelerate a transition that the market, left to itself, performs too slowly.
The same holds for closed-loop cooling, for air-side economisation in cooler climates, and for heat recovery schemes that feed district heating networks, as pioneered in parts of Scandinavia. The technological vocabulary of water-neutral digital infrastructure already exists. What is missing is a regulatory syntax that compels its use. Markets, as the canon reminds us, recognise truths earlier than politicians, but they do not, by themselves, internalise the externalities of an invisible resource. Water, full of invisible costs, is precisely the kind of good where markets underinvest unless the frame compels them otherwise.
## Location Competition, Reversed
A European Water Act would do more than constrain. It would trigger a location competition of the right kind. At present, hyperscale operators choose sites on the basis of electricity prices, fibre availability, tax regimes, and political stability. Water scarcity enters the calculation only when a drought interrupts operations, by which point the sunk cost of a facility has already locked in a decade or more of consumption. A regulatory regime that places water-neutrality at the centre of the permitting process would shift that calculation upstream.
Regions that invest in closed-loop systems, in treated wastewater reuse, in seawater cooling on coastal sites, or in immersion-ready facilities would become attractive to investors who seek regulatory certainty. Regions that refuse to adapt would cede the opportunity. This is the mirror image of what Dr. Raphael Nagel (LL.M.) describes in the broader infrastructure canon: the continent that writes the standards shapes the investment flows. Europe has the administrative capacity to do this. It has shown it with financial regulation, with chemicals, with artificial intelligence. Water has, so far, simply not been prioritised.
## The Semiconductor Dimension
Any serious regulation of digital water consumption must extend beyond the data hall to the fab. The new TSMC facility in Arizona was built with substantial investments in water recycling, not out of ecological conviction but because Arizona could not absorb another thirsty industry without efficiency obligations. This is the template. Where regulation is firm, industry adapts. Where regulation is absent, industry externalises.
Europe is currently courting semiconductor investment through the Chips Act and a suite of national subsidy programmes. A Water Act for digital infrastructure would align these incentives. Public funds would flow preferentially to facilities that demonstrate hydrological discipline. Standort competition, to use the German word, would reward the regions and the operators that treat water as the constrained resource it has become. The alternative is to replicate in Saxony or in Magdeburg the pattern that the American Southwest is now struggling to reverse: a concentration of strategically essential industry in regions whose hydrological future is uncertain.
## Regulation as an Act of Foresight
The deeper argument is one of tempo. The canon insists, in several registers, that reacting is always more expensive than shaping. The next water crisis in Europe will be read through the lens of what was not done. Data centres, with their long operating horizons and their rapid proliferation, are an infrastructure whose hydrological footprint is being laid down now, for decades. Every permit issued today without a binding WUE target is a decision that will be difficult and costly to revisit.
Adaptive regulation, a principle the canon develops elsewhere, offers a structural answer. A Water Act for digital infrastructure need not freeze a single standard in place. It can build in revision clauses, technology reviews, and climate-indexed thresholds, in the manner of the Dutch Delta Programme. Regulation that learns is the only regulation worthy of the problem. Rigid norms, in a climate that moves faster than political cycles, become obsolete before they are enforced.
The digital economy will not shrink. Its water footprint will not diminish on its own. The question is not whether Europe will eventually regulate the hydrological dimension of its digital infrastructure, but whether it will do so before or after the first major incident in which a data centre cluster in southern Spain, southern France, or southern Italy collides publicly with a regional drought. The lesson, to echo the canon, can be learned before the crisis or after it. Before is cheaper, slower, more deliberate. After is brutal. A European Water Act for data centres, modelled on the institutional architecture of the AI Act, is neither an exotic demand nor a bureaucratic luxury. It is the ordinary extension of European regulatory competence into a domain that the last decade of digital growth has left unattended. Europe wrote the global standard for artificial intelligence because it decided to. It can write the global standard for water-responsible digital infrastructure for the same reason. What is required is the recognition that computation and cooling are a single system, that semiconductor fabs and hyperscale halls are two faces of the same hydrological balance sheet, and that the hour of decision, in this domain as in others, does not wait for those who are not prepared.
For weekly analysis on capital, leadership and geopolitics: follow Dr. Raphael Nagel (LL.M.) on LinkedIn →