For years, the narrative surrounding the expansion of the cloud and the subsequent race for artificial intelligence has been one of seamless scalability. Amazon, Microsoft and Google have framed the construction of massive data centers as a logical extension of digital progress. However, that narrative is now colliding with the hard limits of physical geography, power grids, and local water tables.
The pressure is no longer coming solely from environmental regulators or local residents protesting land use. Instead, a significant shift is occurring within the boardrooms. A growing group of shareholders is now demanding datos de agua y electricidad—concrete, site-specific data on how much water and power these companies are consuming across their U.S. Data center portfolios.
This movement, led by more than a dozen investors ahead of the spring proxy season, signals a transition from abstract sustainability goals to a focus on operational risk. The core concern is no longer just about whether these companies will be profitable in the long term, but whether the physical resources—specifically electricity and water—will even exist to sustain the infrastructure required for AI growth.
The scale of the resource demand is staggering. According to reports from Reuters, U.S. Data centers consumed nearly 1 trillion liters of water in 2025, a volume comparable to the annual water demand of New York City. This figure, however, only tells part of the story.
The Hidden Cost of Indirect Consumption
One of the most critical points of contention for investors is the distinction between direct and indirect water use. Even as a data center uses water directly for cooling its servers, a massive amount of water is consumed indirectly through the generation of the electricity that powers the facility. A 2024 study estimated this indirect water consumption at approximately 800 billion liters.
As a former software engineer, I’ve seen how the “magic” of the cloud often obscures the heavy machinery beneath it. The transition to Large Language Models (LLMs) has intensified this problem; AI queries require significantly more compute power—and thus more cooling—than a standard Google search or a cloud storage request. This has turned data centers into high-intensity industrial sites that can strain local utilities to the breaking point.
A Fragmented Approach to Transparency
Investors are frustrated not only by the volume of consumption but by the lack of standardized reporting. Currently, the “Big Three” employ wildly different metrics to describe their environmental footprint, making it nearly impossible for shareholders to conduct a side-by-side risk analysis.
| Company | Reporting Scope | Key Limitation |
|---|---|---|
| Owned and leased centers | Excludes third-party facilities | |
| Microsoft | Total aggregate water use | Lacks site-specific data |
| Amazon | Energy-unit metrics | No global total water volume published |
While all three companies have moved toward closed-loop cooling systems to mitigate water waste, these technical fixes do not address the fundamental demand for transparency. Investors wish to know the “local cost”—how much water is being extracted from a specific aquifer in a drought-prone region and how that impacts the political and operational stability of the site.
Operational Risks and Project Cancellations
The consequences of this lack of transparency are already manifesting in the real world. Several multibillion-dollar data center projects have recently been abandoned by Amazon, Microsoft, and Google due to fierce opposition from local communities. The friction is no longer just about “Not In My Backyard” (NIMBY) sentiment; it is about the survival of local infrastructure.
A data center in a region with abundant water behaves differently than one in a water-stressed area. When a facility forces a municipality to reinforce its entire electrical grid or divert water from agriculture, it creates a political liability that can lead to sudden project cancellations. For investors, these cancellations represent wasted capital and a failure in risk management.
The Path Forward: Transparency as a Metric of Stability
The demand for datos de agua y electricidad is ultimately a demand for a more honest accounting of the AI era. By insisting on site-by-site transparency, investors are attempting to quantify the “resource tension” that these companies leave in their wake. If a company cannot prove it has a sustainable water source for the next decade of growth, its projected earnings are essentially built on sand.
For Amazon, Microsoft, and Google, the challenge is now twofold: they must continue to innovate in cooling and energy efficiency while simultaneously opening their books to a level of scrutiny they have historically avoided. The era of generic sustainability promises is ending; the era of the audit has begun.
The next critical checkpoint for these companies will be the upcoming spring shareholder meetings, where the results of these investor pressures will be formally debated. Whether the companies concede to these transparency demands or double down on aggregate reporting will likely determine the level of trust investors place in the long-term scalability of their AI infrastructure.
Do you think tech giants should be required to disclose resource use by specific location, or is aggregate reporting sufficient? Share your thoughts in the comments or join the conversation on our social channels.
