Rifan Muazin
As the global race for artificial intelligence supremacy accelerates, the debate surrounding data centers has reached a fever pitch. Stakeholders, from Silicon Valley CEOs to grid operators, are fixated on a singular, massive metric: power capacity. How do we provide the gargantuan amounts of electricity required to train the next generation of large language models?
While this is undeniably the defining question of the intersection between economic growth and sustainability, it is missing a critical component. Hidden behind the buzz of megawatts and carbon footprints lies a silent, under-accounted, and potentially catastrophic risk: water.
The Hidden Water Footprint: Beyond the Cooling Tower
The public perception of data center water usage is largely focused on the facilities themselves—specifically, the cooling systems used to prevent high-performance chips from overheating. However, this is a narrow view that misses the forest for the trees. According to an analysis by the nonprofit Ceres and projections from Bluefield Research, the vast majority of a data center’s water footprint is not tied to on-site cooling at all. Instead, it is inextricably linked to the electricity usage of the power plants feeding the grid.
Most of the electricity powering the AI boom comes from thermal power plants—coal, natural gas, or nuclear. These facilities operate by heating water into steam to drive turbines, then utilizing vast quantities of additional water to condense that steam back into a liquid state. Consequently, a data center that prides itself on "water-neutral" on-site operations is still consuming massive volumes of water indirectly.
Data from Janus Henderson, citing research from the Lawrence Berkeley National Laboratory, indicates that this indirect consumption can account for up to 75 percent of a facility’s total water footprint. For context, Meta, one of the few industry giants to provide transparent disclosures, reported that its indirect water consumption in 2024 was more than 20 times greater than the volume of water consumed directly at its own data centers.
Chronology of a Growing Crisis
The intersection of water scarcity and digital infrastructure is not a new concept, but it has hit a critical inflection point due to the generative AI explosion.
- Early 2010s: Data centers were primarily viewed through the lens of energy efficiency (PUE). Water was a secondary concern, largely managed at the facility level to keep cooling costs down.
- 2020–2022: As cloud computing scaled, local communities began to push back against data centers in drought-prone areas, leading to isolated incidents of public outcry regarding municipal water usage.
- 2023: The rapid deployment of high-density AI clusters began to strain regional power grids. The link between grid capacity and water scarcity moved from a theoretical concern to an operational reality.
- 2026 (Present): A surge in private credit financing—now handling between 70 and 90 percent of project-level borrowing for new data centers—has created a disconnect. Lenders are demanding rigorous energy planning but have yet to integrate "water-at-the-source" risk into their loan covenants.
- August 2026: In a landmark move, Texas Governor Greg Abbott ordered a freeze on interconnection approvals for new data centers until developers can prove the sustainability of their power-and-water footprint. This signals a transition from voluntary reporting to regulatory mandates.
Supporting Data: The Disconnect in Diligence
When private credit firms and banks finance a multi-billion-dollar data center project, they conduct exhaustive due diligence. They look at tenant credit, lease terms, power delivery, and queue positions. Yet, water at the generating plant is rarely a line item in the financial model.
This represents a structural failure in risk pricing. Lenders often argue that they don’t need to account for water at the power plant because the responsibility lies with the utility provider. "The exposure sits with a regulated company carrying its own water-planning duties," the logic goes. However, this ignores the changing climate. A utility’s historical water planning is based on decades of stable weather patterns—a baseline that is rapidly becoming obsolete.
Furthermore, there is no standardized framework for reporting indirect water use. While Amazon tracks its indirect consumption, it has not made the data public, citing a lack of industry consensus. Google and Microsoft have remained largely opaque on the subject. Without a "Scope 3" equivalent for water, lenders are essentially underwriting a risk they cannot see, measure, or price.
Official Responses and Regulatory Shifting
The regulatory landscape is shifting, albeit slowly. The Texas moratorium serves as a warning shot to the industry: infrastructure projects can be halted if they threaten the essential resources of the local population.
Industry advocacy groups, such as the World Resources Institute (WRI) and the CEO Water Mandate, are now pushing for a new "scopes-based" water accounting framework. This initiative, which includes participation from SCS Global Services and the WWF, aims to create the same level of accountability for water that the Greenhouse Gas Protocol created for carbon.
When asked about these developments, industry experts note that the resistance to reporting isn’t necessarily malicious—it is practical. "You cannot manage what you do not measure," notes one analyst. "And right now, the industry has no standardized ruler for water."
Implications for the Future: The Cost of Ignoring the Well
The current approach to data center construction contains a fundamental paradox: the pursuit of "water-efficient" cooling. To save municipal water, developers often pivot to dry-cooled systems. While this looks good on a corporate sustainability report, dry-cooled systems are inherently less energy-efficient. They require more electricity to run, which in turn increases the demand on the grid, leading to higher water consumption at the power plant.
The water simply moves from the data center’s backyard to the power plant’s intake pipe.
The Financial Risk
For investors and cloud customers, the implications are severe. If a data center is built in an area where the local power plant faces water curtailment due to drought, the data center faces the risk of "brownouts" or forced shutdowns. This is not just an environmental issue; it is a fiduciary one. A risk that is not priced at the financing stage will eventually manifest as a cost—either through higher electricity rates to pay for water-mitigation infrastructure or through the total loss of operational continuity.
The Strategic Mandate for Enterprises
For companies signing long-term cloud service agreements, the "hidden" water footprint should now be a standard part of vendor due diligence. If a company is basing its own ESG goals on the sustainability of its cloud provider, it must ask: "Was the water availability at the generation source assessed for the duration of this contract?"
A 15-year lease is a long time. In an era of shifting climate patterns, a site that has plenty of water today may be in a state of crisis by year 12.
Conclusion: Bridging the Gap
The AI revolution is bringing unprecedented innovation to the global economy, but it is also stress-testing the physical limits of our environment. The current mismatch—where carbon is meticulously tracked while water is left to chance—is a relic of a pre-AI world.
To sustain the current trajectory of growth, the financial sector must demand the same level of water transparency that it demands of energy. The development of a formal, universal water-accounting standard is no longer a "nice-to-have" for sustainability teams; it is a necessity for the stability of the digital economy. As we look toward the next decade of development, the winners will be the firms that recognize that in the world of high-performance computing, the most valuable commodity isn’t just the data—it’s the water that makes the processing possible.
Sagoh
Asep Darmawan
Pevita Pearce
Iffa Jayyana
