2 April 2026
An AI data centre complex in an arid region cooling towers releasing steam beside a shrinking reservoir, cracked earth stretching to the horizon. The infrastructure demands of the generative AI boom are placing unprecedented pressure on water systems already under climate stress.
For decades, the language of corporate sustainability has been written in carbon. Emissions targets, net-zero pledges, Scope 1, 2, and 3 disclosures, carbon dioxide gave sustainability a grammar that boards, investors, and regulators all came to speak. But while boardrooms focused on what rose invisibly into the atmosphere, a second and, in some ways, more immediate crisis was building beneath the surface. Water, the resource that underpins every human activity from farming to semiconductor fabrication, has arrived on the corporate agenda, and it is arriving with urgency. This is not simply an environmental story. It is a financial materiality story. In 2026, the two are becoming impossible to separate.
| $265B Annual financial exposure | Estimated annual cost of water stress to S&P 1200 companies by 2050 (S&P Global) | Water vs Carbon ~15 years behind on standardised disclosure frameworks |
Water is no longer a sustainability metric sitting quietly in an appendix. It is a direct operational and financial threat. Across energy, technology, chemicals, agriculture, and manufacturing, companies are confronting a stark and uncomfortable reality: in many regions, water systems simply cannot keep up with demand. Bloomberg Intelligence’s 2026 Global Outlook on Water Risk estimates that roughly one-third of global power generation capacity and over 40% of data centre operations are already exposed to high water stress by the end of this decade. The term that is entering corporate risk registers in 2026 is “water bankruptcy”, the point at which regional water systems structurally fail to meet the combined demands of industrial, agricultural, and municipal users at the same time. This is no longer theoretical. It is already happening in the American West, in parts of southern Europe, across the GCC, and in pockets of northeast Asia. For companies with facilities, suppliers, or critical infrastructure in these regions, water bankruptcy is an existential scenario, not just a planning assumption.
Robeco’s 2026 analysis of the global chemical industry, responsible for an estimated 5 to 10% of global freshwater withdrawals, found that companies face serious revenue and valuation risks under even a modest 10% water supply shock modelled for 2030. More strikingly, water-efficient companies are already beginning to outperform less-efficient peers in sectors like chemicals. This suggests that capital markets are starting to price water resilience into valuations, much like they once began to price carbon risk.
Water is roughly fifteen years behind carbon in standardised measurement and disclosure. The frameworks being built now will define how this data is captured for a generation.”
If any single force has turbocharged the urgency of corporate water risk in this decade, it is artificial intelligence. The generative AI boom that reshaped the technology industry from 2023 onward has brought with it an enormous and largely underreported infrastructure demand: water. A single large AI-oriented data centre can require approximately 300,000 gallons of water per day for cooling. Water absorbs heat roughly 3,000 times more effectively than air, which is why liquid cooling has become the dominant infrastructure choice for high-density AI compute in 2026. It is no longer optional, but essential.
The numbers that follow are stark. Annual data centre water consumption in the United States alone is projected to double or even quadruple by 2028, reaching between 150 and 280 billion litres per year compared to 2023 levels. More than 160 new AI-related data centres have been built in the past three years, many of them in water-stressed regions such as the American West, northeast Asia, Ireland, Chile, and the Netherlands. These locations attract developers due to cooler climates and available land, yet overstressed aquifers are creating growing tension.
The case of Microsoft illustrates the challenge clearly. In 2020, Microsoft President Brad Smith pledged that the company would become “water positive” by 2030, replenishing more water than it consumes. Then came the generative AI expansion. Internal estimates now indicate that the company’s water consumption will increase significantly by 2030 due to accelerating data centre construction, making that earlier commitment increasingly difficult to achieve without major technological shifts.
Google has taken a more structured approach. The company’s Water Stewardship Project Portfolio, published for World Water Day 2026, outlines commitments to replenish more freshwater than it consumes across its global offices and data centres by 2030. In 2025 alone, Google replenished more than 7 billion gallons across 156 projects spanning 97 watersheds, with a target of over 11 billion gallons annually once all projects reach full implementation. The contrast between these two trajectories, one struggling under the weight of AI growth and the other embedding stewardship into its infrastructure strategy, highlights a critical turning point for the technology sector.
Agriculture, Industry and Compound Stress
Data centres do not draw from water systems in isolation. They add pressure to systems already under significant strain. Agriculture accounts for roughly 70% of global freshwater withdrawals. Heavy industry, including chemicals, semiconductor manufacturing, textiles, and food processing, accounts for much of the rest. The regions most attractive to data centre developers are often the same regions already supporting intensive agricultural and industrial water demand.
Semiconductor fabrication plants, the facilities that manufacture the very AI chips driving data centre growth, are themselves highly water-intensive operations. Many are concentrated in the western United States and northeast Asia, precisely the basins experiencing the most severe long-term stress. The result is a cascade of competing demands drawing from the same shrinking aquifers, with no single actor holding enough leverage to resolve the imbalance alone.
The corporate response has begun, but it remains uneven. Apple has committed to certifying all its owned data centres under the Alliance for Water Stewardship standard and to supporting its suppliers in achieving an average 50% water reuse rate by 2030. PepsiCo is pursuing collective action projects to replenish more than 100% of the water it uses at high water-risk sites by the same deadline. These are meaningful commitments. However, the 2025 Corporate Water Stewardship Benchmark found that only 17% of companies with water quality targets connect them to collaborative, watershed-level efforts, revealing a clear gap between individual pledges and the collective action needed to sustain water systems.
“A company can hit every internal reduction target and still watch its region go water-bankrupt if its competitors, suppliers, and municipalities are not aligned.”
The regulatory environment surrounding water risk is beginning to crystallise, particularly in Europe. The European Sustainability Reporting Standards (ESRS) include ESRS E3, dedicated to water and marine resources, which requires companies subject to the Corporate Sustainability Reporting Directive (CSRD) to disclose water withdrawal volumes, consumption levels, and basin-level stress exposure as part of their double materiality assessments. In 2026, ESRS E3 was updated to sharpen its focus on withdrawal and discharge metrics, reflecting growing regulatory recognition that water data is decision-relevant, not merely aspirational.
It is worth noting the current complexity of the CSRD landscape. The EU’s Omnibus Simplification Package, which entered into force in early 2026, significantly raised the thresholds for mandatory CSRD compliance, removing an estimated 85 to 90% of previously in-scope companies from mandatory requirements. The reporting timeline for the next wave of companies has been pushed to 2028. However, as one analysis noted, regulatory relief does not eliminate business drivers. Investors, lenders, customers, and insurers continue to ask water-related questions regardless of legal obligation, and companies without credible answers will find themselves at a disadvantage in capital markets that are, slowly but unmistakably, pricing water resilience.
Outside Europe, momentum is building through different channels. Nearly 40 global jurisdictions have adopted or are planning to adopt climate disclosure frameworks aligned with the International Sustainability Standards Board’s standards. Australia and Spain have sustainability disclosure laws coming into effect this year. California’s SB 253, requiring large companies to disclose assured Scope 1, 2, and 3 emissions, remains active despite legal challenges. Water, while not yet a standalone mandatory category in most non-EU jurisdictions, is increasingly appearing in climate risk disclosures, physical risk assessments, and supply chain due diligence requirements.
The Financial Materiality Argument
The most consequential shift happening in 2026 is conceptual. Water risk is transitioning from a corporate social responsibility footnote to a balance sheet consideration. Treatment and reuse infrastructure may require trillions in global investment by 2040. Water constraints are already influencing operating costs, asset siting decisions, and capital allocation across power generation, AI infrastructure, agriculture, and heavy manufacturing. When Robeco frames water scarcity as a threat to chemical company liquidity and valuations, rather than simply an environmental compliance issue, the conversation clearly moves from sustainability into finance.
For ESG reporting professionals and sustainability teams, this creates immediate pressure. CSRD-aligned disclosures in Europe are increasingly expected to include water withdrawal, consumption, and stress exposure alongside carbon metrics, even as regulatory thresholds evolve. The comparison with carbon accounting is both instructive and humbling. Carbon had roughly a twenty-year head start in standardised measurement, voluntary frameworks, mandatory disclosure, and financial integration. Water is approximately fifteen years behind on that same path. The platforms, methodologies, and tools being developed today will shape how this data is captured, verified, and valued for years to come.
What Leading Companies Are Doing Differently
The organisations navigating water risk most effectively in 2026 share three capabilities that distinguish them from many of their peers.
The first is granular exposure mapping. Leading companies do not assess water risk at a broad corporate level. They focus at the basin level. Which specific sites draw from water-stressed watersheds? Which tier-one and tier-two suppliers are located in vulnerable river basins? Where could a disruption cascade into production loss? This kind of detailed, location-specific analysis is the foundation of credible water risk management, yet it remains less common than it should be.
The second is measurable, time-bound reduction targets that are aligned with local basin conditions. General commitments to reduce water intensity are no longer sufficient. High-performing companies are setting site-specific withdrawal limits based on the actual capacity of the basin. These targets remain meaningful even if every company in the watershed meets its own reduction goals, because they are grounded in the basin’s natural replenishment rate.
The third, and most challenging, is active participation in collective watershed stewardship. This involves working with regulators, municipalities, agricultural users, and industry peers to manage shared water resources collaboratively. It is not just about reducing individual usage, but about contributing to the long-term sustainability of the entire system. This marks the shift from basic water stewardship to a more integrated and strategic approach to water management.
The Bottom Line
Carbon got a twenty-year head start in corporate sustainability. Water does not have that luxury. The convergence of AI infrastructure expansion, climate-driven drought, intensifying agricultural demand, and tightening regulation is compressing the timeline for corporate action into something far shorter than what carbon was given. The companies that treat water with the same rigour they now apply to carbon, measuring it precisely, pricing it into risk models, reporting it transparently, and managing it at the basin level rather than the boardroom level, will be the ones that avoid operational disruption, maintain investor confidence, and retain their social licence to operate in an increasingly water-constrained world.
Water was always the most fundamental resource. It just took a crisis to make business treat it that way.
