15.09.20265 min

The water strain behind AI’s rise

Cyril Suter - CFA - Active Advisor

Behind the billions being invested in artificial intelligence (AI) lies a far more tangible resource: water. From data centres to semiconductor plants, the digital revolution depends on ever-larger volumes of freshwater. As a result, water is at risk of becoming one of the key constraints on the further development of AI.

Summary

The rapid expansion of AI and digital infrastructure is making water an increasingly critical constraint on technological growth.

Data centres and semiconductor manufacturing plants are placing increasing pressure on already stressed water systems, while climate change and ageing infrastructure are further weakening supply.

In response, operators are accelerating efforts to reduce freshwater dependence, creating a long-term thematic opportunity across the whole water supply chain.

Why water will become the new constraint for technological expansion

As AI accelerates across industries, it creates extraordinary opportunities but simultaneously adds new pressures on essential infrastructure and resources. One of the resources central to this transformation is water.

AI and digital infrastructure are adding a new, fast-growing layer of water demand to an already stressed system. Data centres need water for cooling, while semiconductor manufacturing plants rely on ultra-pure water in the manufacturing process. At the same time, the supply side is becoming more fragile due to climate change and an outdated water infrastructure system.

Water is no longer only an environmental and social issue: it is becoming a real constraint on industrial capacity, site selection and the pace of technological expansion.

Sources: Bloomberg, Indosuez Wealth Management.

A planetary resource under structural pressure

Freshwater is a finite and increasingly constrained resource which is already under pressure worldwide.

Only around 3% of the Earth’s water is fresh, and a significant proportion of that is locked away in ice caps and glaciers. At the same time, global water demand is projected to exceed supply by 40% by 2030.

Agriculture remains by far the largest consumer of water, accounting for approximately 70% of total water use. However, the fastest-growing demand is expected to come from digital infrastructure, driven by the rise of AI. Water consumption linked to data centres and semiconductors could increase by more than 450% by 2050.

A supply side that is becoming increasingly fragile

At the same time, supply-side constraints continue to build. Climate change is increasing the frequency and intensity of droughts, while ageing infrastructure is causing significant losses. On average, close to 30% of water carried through networks is lost before reaching its destination.

By 2040, large parts of the United States, Spain and Eastern China could face high levels of water stress. It is in several of these regions that the most ambitious technological investments are being concentrated.

AI and semiconductor plants: New giants of water consumption

In this already strained context, AI is adding a new layer of pressure on the resource.

In data centres, water is primarily required for cooling. Evaporative systems are highly efficient from an energy perspective, but consume significant volumes of water, as evaporation requires continuous replenishment.

Alphabet, Google’s parent company, is estimated to have consumed nearly 40 million cubic metres of water in 2025, roughly equivalent to the annual consumption of a country the size of Luxembourg.

Semiconductor manufacturing is even more demanding: plants use ultra-pure water to clean wafer, remove contaminants, cool certain equipment and ensure the quality of industrial processes.

Water quality is also critical in this industry and even minor disruptions to supply can affect production. TSMC (Taiwan Semiconductor Manufacturing Plant) illustrates the scale of the issue: a giga factory can consume around 38 million litres of water per day, equivalent to the daily consumption of roughly 300,000 households.

In other words, behind every advance in AI lies industrial infrastructure that is highly dependent on water.

*E=estimates Source: Global Market Insight, Indosuez Wealth Management

Limits are already visible

Tensions are already visible on the ground. In Chile, Google’s second data centre project in Santiago remains suspended following concerns about the impact of its water consumption. In the US, AWS’s (Amazon Web Services) Project Blue near Tucson, Arizona, a USD 3.6 billion development, generated significant local opposition in a drought-prone region.

The semiconductor industry is no exception. During the drought that hit Taiwan in 2020 and 2021, TSMC had to cope with a 15% reduction in water supply and rely on emergency deliveries to maintain production.

These examples illustrate an important evolution. Water is no longer solely a sustainability issue. It is becoming an operational risk that can have a direct impact on the growth of the technology sector.

How the industry is adapting

Faced with these constraints, major digital players are looking to reduce their dependence on freshwater by:

improving cooling system efficiency;

monitoring consumption in real time and detecting leaks;

reducing the use of evaporative cooling in regions exposed to high water stress;

increasing the recycling and reuse of treated water;

making greater use of recycled water or rainwater for non-critical uses;

strengthening network control, measurement and monitoring capabilities.

The goal is no longer simply to reduce costs or environmental impact, but also to secure access to a resource that has become strategic.

Blue Gold: a long-term investment theme

Long viewed as primarily an environmental issue, water is now emerging as an economic, industrial and strategic concern. As demand surges due to the impact of AI, supply is becoming more fragile in many regions worldwide. This reality is beginning to affect the development of digital infrastructure and could create new investment opportunities.

More than just a cyclical theme, water is emerging as a long-term structural trend, opening up opportunities across the entire water value chain: treatment, leak detection, valves, control equipment, network modernisation and recycling technologies.

If AI is to be one of the major growth drivers of the coming decades, the ability to manage water resources efficiently is likely to be one of the essential conditions for its development.