Introduction: The Resource Constraint Nobody Saw Coming

The next major constraint on AI infrastructure may not be electricity. It may be water. While hyperscale operators race to secure gigawatts of power for AI campuses, an equally important resource is becoming harder to secure: freshwater.

Hyperscale operators, colocation providers, and AI infrastructure developers have spent billions securing electricity supplies, investing in renewable energy projects, and competing for increasingly scarce grid connections.

In major data center markets, access to power has become the defining factor determining where facilities can be built and how quickly they can be deployed.

However, while the industry has become exceptionally sophisticated in evaluating power availability, another resource constraint is emerging that could prove equally important over the coming decades. That resource is water.

Water has traditionally been treated as an operational requirement rather than a strategic consideration. Developers evaluate local utility availability, estimate cooling requirements, and assume that municipal infrastructure will continue supplying the necessary volumes throughout the life of the facility. Yet this assumption is becoming growingly risky.

Climate change, population growth, urbanization, industrial expansion, and the explosive rise of artificial intelligence are fundamentally changing the relationship between digital infrastructure and freshwater resources.

As data centers become larger and more computationally intensive, cooling requirements are increasing dramatically. At the same time, many regions around the world are experiencing worsening drought conditions, declining groundwater reserves, and growing competition for water among municipalities, agriculture, manufacturing, and critical infrastructure sectors.

The result is a growing disconnect between how operators evaluate sites and the environmental realities they will face over the next thirty years.

The data center industry learned an important lesson about electricity during the past decade: power availability cannot be taken for granted. A similar realization is now emerging around water.

For many future developments, access to sustainable water resources may become just as important as access to sustainable energy.

The AI Revolution Is Transforming Cooling Requirements

Artificial intelligence has fundamentally altered the economics of digital infrastructure. Traditional cloud workloads certainly generated heat, but AI workloads operate at a completely different scale. Training large language models requires tens of thousands of GPUs running continuously for extended periods.

Inference workloads are also becoming increasingly resource-intensive as AI applications move into enterprise software, healthcare, finance, manufacturing, and consumer services.

Every watt consumed by computing equipment ultimately becomes heat. That heat must be removed continuously to maintain operational reliability.

As rack densities rise from traditional levels of 5-15 kW to 50 kW, 100 kW, or even higher in AI environments, cooling systems are becoming one of the most critical components of data center infrastructure. The scale of future cooling demand is staggering.

Global cooling-related greenhouse gas emissions currently stand at approximately 4.1 billion tonnes of CO₂ equivalent annually. Without intervention, these emissions could increase to 7.2 billion tonnes by 2050 as cooling demand expands worldwide.

Installed cooling capacity is projected to increase from approximately 22 terawatts in 2022 to nearly 68 terawatts by 2050 under business-as-usual scenarios. This represents more than a threefold increase in cooling infrastructure requirements within less than three decades.

These figures are not solely driven by data centers. However, the rapid expansion of AI infrastructure is becoming one of the most significant contributors to future cooling demand.

Depending on the cooling architecture and climate conditions, a large hyperscale data center facility can consume up to 5 million gallons of water annually for cooling operations.

Consequently, the growth of AI is indirectly increasing pressure on freshwater resources across multiple regions.

Why Water Matters More Than Most Operators Realize

When discussing infrastructure constraints, power shortages typically dominate headlines. Water shortages rarely receive the same level of attention.

This difference exists because electricity disruptions tend to be immediate and visible, while water stress develops gradually over time. Aquifers decline slowly. Reservoirs shrink season by season. Drought conditions intensify over years rather than days.

Yet for long-lived infrastructure assets such as data centers, gradual risks can be even more dangerous than sudden ones.

Most hyperscale facilities are designed to operate for twenty to thirty years or more. Site selection decisions made today will influence operational performance well into the 2050s. This creates a significant challenge.

Many site selection processes rely heavily on historical water availability data. Operators assess current municipal capacity, existing water rights, and historical weather patterns. However, future climate conditions may differ substantially from historical trends.

A region that appears water-secure today may experience severe water stress in the future.

The data center industry is expanding and performs sophisticated power forecasting, energy market analysis, and grid resilience assessments. Comparable rigor is often absent from water risk evaluations.

This imbalance creates a strategic blind spot that could have major implications for future infrastructure development.

Climate Change Is Rewriting the Rules of Site Selection

Climate change is no longer a future concern. It is already influencing water availability across major economic regions.

Rising temperatures increase evaporation rates from reservoirs, rivers, and soil. Changes in precipitation patterns alter groundwater recharge rates. Snowpack reductions affect seasonal water supplies in many regions. More frequent and severe droughts create additional pressure on already stressed water systems.

These trends are fundamentally changing the assumptions that have historically guided infrastructure development.

For instance, a location that received consistent rainfall over the past fifty years may experience significant variability over the next thirty years. Groundwater resources that once appeared abundant may become more and more constrained due to agricultural demand, population growth, or reduced recharge rates.

For data centers, this is particularly important because infrastructure investments are inherently long-term.

Unlike commercial real estate projects that may be repurposed after a decade, data centers are designed for continuous operation over multiple decades.

This means operators must increasingly evaluate future climate scenarios rather than relying solely on historical conditions.

The sites that appear optimal today may not remain optimal throughout the life of the asset. These risks are no longer theoretical. Several high-profile projects have already demonstrated how water availability can influence infrastructure planning decisions.

When Water Becomes a Development Constraint: Google's Chile Experience

One of the clearest examples of water becoming a strategic infrastructure issue emerged in Chile. Google faced significant scrutiny over plans for a data center project near Santiago, a region that has experienced prolonged drought conditions and increasing water stress.

Community concerns and environmental reviews focused heavily on the facility's proposed water consumption and its potential impact on local resources.

The project ultimately highlighted a broader industry lesson: even global technology companies with significant financial resources cannot assume that access to water will be socially or politically acceptable.

As a result, Google revised aspects of its cooling strategy and progressively emphasized alternative cooling technologies that reduce freshwater dependence.

The Chile experience demonstrated that water availability alone is no longer sufficient. Operators must also secure public trust and demonstrate long-term water stewardship if they expect projects to move forward smoothly.

Water Risk Is More Than Just Water Consumption

Many discussions about data centers and water focus exclusively on how much water facilities consume. This perspective is incomplete. Water risk encompasses a much broader range of issues.

These include:

• Drought exposure

• Groundwater depletion

• Municipal allocation restrictions

• Regulatory intervention

• Rising water prices

• Community opposition

• Operational interruptions

• Climate-related supply variability

• Environmental compliance requirements

• Investor scrutiny

A facility located in a water-stressed region may face multiple overlapping risks even if its direct water consumption is relatively modest.

For instance, local governments may impose restrictions during drought periods. Communities may oppose expansion projects. Environmental permitting processes may become more complex. Investors may require additional disclosures regarding climate resilience.

The cumulative effect of these factors can significantly influence the long-term viability of a project.

Water risk therefore extends far beyond operational efficiency. It is becoming a strategic business risk.

Communities Are Beginning to Push Back

Historically, data centers enjoyed strong public support. They created jobs, generated tax revenue, and were generally perceived as clean economic development projects.

That perception is changing. As awareness of water scarcity grows, communities are beginning to scrutinize the resource requirements associated with digital infrastructure.

This trend is especially evident in regions experiencing recurring drought conditions.

Residents facing water restrictions often question whether industrial facilities should continue consuming significant quantities of freshwater resources. The rapid expansion of AI infrastructure has intensified these concerns.

To many citizens, the trade-off appears straightforward: should scarce water resources support household needs, agricultural production, ecosystem preservation, or the training of artificial intelligence models?

Whether or not this framing accurately reflects actual water consumption patterns, it is increasingly shaping public perception.

As a result, water is becoming a social-license issue. Future projects may require not only engineering approval and regulatory permits but also community trust.

Developers that underestimate public concerns around water use could face delays, legal challenges, and reputational risks.

Arizona's Growing Tension Between Growth and Water Security

Arizona has become one of the world's most attractive markets for digital infrastructure due to its abundant land availability, business-friendly policies, and strong connectivity ecosystem. However, it is also one of the most water-stressed regions in North America.

As major technology companies expanded their presence in the state, local communities and policymakers increasingly questioned how future growth would affect long-term water availability. Several municipalities began examining whether continued data center expansion could place additional pressure on already constrained water resources.

The debate illustrates a challenge likely to emerge in many global markets: the locations most attractive for large-scale infrastructure investment are not always the locations best positioned to sustain increasing water demand over the coming decades.

For operators evaluating future sites, Arizona demonstrates that future site selection decisions will depend not only on access to land and power, but also on confidence in long-term water availability.

The Economics of Water Are Changing

One reason water has historically received limited attention is that it has been relatively inexpensive compared with electricity.

For many operators, water costs represent only a small fraction of total operating expenses. However, focusing exclusively on water prices misses the larger economic picture.

The greatest cost associated with water is not purchasing water. The greatest cost is lacking access to water.

A facility facing operational restrictions due to water shortages may experience reduced efficiency, higher cooling costs, increased capital expenditures, or constraints on future expansion.

In extreme scenarios, water scarcity could influence whether a site remains economically viable. This dynamic mirrors the industry's experience with electricity. A decade ago, operators primarily focused on electricity prices.

Today, availability matters far more than price. The same transition is beginning to occur with water. Access is becoming more important than cost.

Why Investors Are Watching Water More Closely

Institutional investors increasingly recognize that climate-related risks can directly affect infrastructure performance. Water availability is emerging as one of the most important climate resilience indicators.

Large infrastructure funds, pension funds, sovereign wealth funds, and private equity firms are paying closer attention to environmental constraints that could affect long-term returns.

Data centers represent multibillion-dollar investments designed to generate revenue over several decades.

Investors therefore need confidence that critical resources will remain available throughout the asset's lifespan.

Water stress introduces uncertainty into this equation. Facilities located in regions facing worsening water scarcity may encounter higher operating costs, additional regulatory requirements, or limitations on future expansion.

Consequently, water resilience is gradually becoming part of investment due diligence and infrastructure valuation frameworks.

Projects that demonstrate strong long-term water security may enjoy a competitive advantage when seeking financing.

Technology Is Part of the Solution

The good news is that the industry is not standing still. Significant innovation is occurring across data center cooling technologies.

Direct-to-chip cooling, liquid cooling, immersion cooling, advanced heat recovery systems, and closed-loop architectures are improving efficiency while reducing water intensity.

Many operators are also investing in reclaimed water systems, wastewater recycling infrastructure, and alternative cooling strategies that minimize freshwater withdrawals.
These technologies represent important progress. However, efficiency improvements alone are unlikely to fully offset the scale of future demand growth.

Global digital infrastructure expansion continues to accelerate. AI adoption remains in its early stages. Edge computing deployments are expanding. Cloud adoption continues worldwide.

As a result, total demand for cooling resources is expected to rise even as individual facilities become more efficient. Technology can reduce water intensity. It cannot eliminate the need for strategic water planning.

Water Will Influence Future Data Center Geography

One of the most significant implications of increasing water risk is its potential impact on where future data centers are built.

Historically, developers prioritized locations offering inexpensive land, strong connectivity, favorable tax incentives, and reliable power. Water availability is increasingly joining that list.

Regions with abundant renewable energy and sustainable water resources may attract disproportionate shares of future investment.

Conversely, locations experiencing chronic water stress may encounter growing development challenges. This does not mean that data centers will disappear from water-constrained regions.

Rather, operators will need to invest more heavily in efficiency technologies, alternative water sources, and resilience measures. The competitive landscape may therefore shift toward regions capable of providing both energy security and water security.

Countries with low carbon intensity may not necessarily offer favorable water conditions, while regions with abundant water resources may have a higher environmental footprint from electricity generation.

As a result, site selection decisions must balance carbon emissions, water consumption, and land-use impacts simultaneously rather than optimizing for a single metric.

Environmental Footprint Intensities of Top DC Hubs

Note: The figures below represent the environmental footprint associated with national electricity generation systems rather than data center operations themselves.

Future site selection models will likely incorporate:

• Long-term climate projections

• Watershed resilience assessments

• Groundwater sustainability analysis

• Water pricing forecasts

• Community impact evaluations

• Regulatory risk assessments

• Water replenishment opportunities

In many cases, these factors could become just as important as power availability.

From Water Consumption to Water Stewardship

Perhaps the most important shift occurring within the industry is conceptual. Leading operators are moving beyond simple water consumption metrics. Instead, they are embracing water stewardship.

One indication of how seriously hyperscale operators are treating water risk is the growing focus on Water Usage Effectiveness (WUE), a metric that measures the amount of water consumed per kilowatt-hour of IT energy.

Over the past several years, leading cloud providers have significantly improved water efficiency across their data center portfolios.

Water Usage Effectiveness (WUE) of Major Hyperscale Operators (Liters per kWh)

Water stewardship recognizes that long-term business success depends on the health of the broader watershed.

This approach focuses on protecting local water resources, improving efficiency, supporting community resilience, and investing in replenishment initiatives.

Examples include:

• Watershed restoration projects

• Groundwater recharge programs

• Wetland conservation

• Rainwater harvesting systems

• Recycled water infrastructure

• Water-positive commitments

• Community water partnerships

These initiatives help strengthen regional water resilience while reducing operational risk. They also support stronger relationships with regulators, investors, and local communities.

As water becomes a more strategic issue, stewardship is likely to become a defining characteristic of industry leaders.

The AI Boom Meets Water Scarcity

The rapid expansion of artificial intelligence infrastructure is progressively colliding with water scarcity concerns. Across several drought-prone regions in the United States and Europe, communities have begun questioning whether massive AI training facilities should be located in areas already experiencing pressure on water resources.

Recent investigations have highlighted how hyperscale AI campuses can require significant cooling infrastructure, particularly during periods of extreme heat when water supplies are already strained. The issue becomes even more sensitive during drought conditions, when residents may face restrictions while industrial facilities continue operating.

This tension represents a preview of the broader challenges facing the industry. As AI workloads continue to grow, operators will need to balance computational expansion with responsible water management.

Future success will depend not only on technological innovation but also on the ability to align infrastructure growth with regional environmental realities.

Water Is Becoming the New Power

The data center industry is entering a new era. For years, the primary question driving site selection was simple: Where can we secure enough electricity?

That question remains important. But it is no longer sufficient.

The convergence of artificial intelligence, climate change, cooling demand growth, and increasing freshwater scarcity is elevating water into a strategic infrastructure resource.

Operators that continue treating water as a secondary operational concern may find themselves exposed to growing environmental, financial, regulatory, and reputational risks.

Those that recognize water as a core site selection criterion will be better positioned to build resilient infrastructure capable of supporting the next generation of digital services.

Power fueled the first phase of the digital economy. Water may determine where the next phase can grow.

The industry's next competitive advantage will not simply come from securing more megawatts. It will come from securing sustainable access to one of the world's most valuable and increasingly constrained resources.

In the age of artificial intelligence, water is no longer just a utility. It is infrastructure. And it is becoming the next power.