AI not only "lacks electricity," but also "lacks water"!
```
While all eyes on Wall Street are fixed on Nvidia’s GPU shipments and the gigawatt loads of the power grid, a more hidden, yet equally deadly, “gray rhino” is approaching: water.
According to Chasewind Trading Desk, on November 26 Morgan Stanley’s latest research report pointed out that AI is not only a power guzzler, but also a super “water sucker.”
While energy bottlenecks have long been a cliché, investors are severely underestimating the disruptive power of water resource constraints on AI expansion. Data centers are not just about computing power; it’s about how to cool these burning-hot semiconductors. On this front, the global water crisis is evolving into a local project killer.
100 Billion Liters at Stake: The Terrifying Data Uptick
Data centers are entities highly reliant on physical resources. Morgan Stanley’s model shows that the “thirstiness” of AI data centers is growing at a staggering rate.
Baseline forecast: By 2028, the direct cooling and power generation water consumption of AI data centers alone will reach 106.8 billion liters.Bull case risk: If AI demand explodes as optimists predict, that number will further balloon to 148.5 billion liters. Even in a pessimistic scenario, consumption will reach 63.7 billion liters.
This means that to keep large models running, we might have to consume the equivalent of several mid-sized cities’ annual water use.
The Real Bottleneck: Not Who Drinks the Most, But Who Can’t Get Water
There is a huge expectation gap here. In terms of total volume, power generation uses the most water (thermal and nuclear power plants require massive cooling), which typically accounts for the largest portion of AI’s water footprint.
However, the real bottleneck is not here.
Investor research shows that onsite cooling and chip manufacturing are the fatal weaknesses. The reason is simple: water resource risk is highly localized.
Electricity can be transmitted across regions via high-voltage lines, but you can’t instantaneously ship water from the Colorado River to the Arizona deserts. If there’s no available water locally, a data center can’t be built, or will be forced to shut down. This is no longer just a theoretical deduction, but a reality unfolding now:
Amazon’s defeat: In Tucson, Arizona, Amazon’s “Project Blue” was outright rejected by local authorities because of massive water and power demands.Britain’s blockade: Anglian Water publicly opposed plans to build an AI data center in North Lincolnshire, on grounds the water supply in the area is already overstretched.
As long as local municipalities say “No”, no matter how many H100 chips you have, the project comes to a halt.
Tech Self-Rescue: From “Drinking Tap Water” to “Drinking Seawater”
To survive in this increasingly parched world, tech giants are being forced to upgrade technology. Investors are now eyeing emerging technologies that can significantly reduce WUE (Water Use Efficiency):
Microchannel cold plates: By having higher thermal performance and less coolant flow, this directly reduces water consumption.Free natural cold sources: Google’s Finnish case is considered a benchmark—using an undersea tunnel to pull in cold seawater for cooling via heat exchange modules, minimizing drinking water use to the greatest extent.
The Regulatory Iron Fist
If you think this is just for ESG reports, you’re gravely mistaken. Regulators are sharpening their axes.
Asia-Pacific Region: Singapore and Malaysia have already drawn red lines, aiming to cap the WUE of data centers at 2.0 m³/MWh or below within ten years.
Europe: The EU is expected to introduce mandatory minimum water performance standards for data centers by the end of 2026.
Who Benefits?
In this “water panic,” the direct beneficiaries are not the water sellers, but those who “manage water.”
The market logic is clear: as hyperscale data center operators scramble to achieve “net water positive” targets by 2030, demand for desalination and water recycling solutions will see structural growth.
This is not about who has the best algorithm; it’s about who can keep your servers from overheating without being cut off from the local water supply. In the next stage of AI, companies holding core water treatment technologies—such as Ecolab, Toray Industries, Veolia, and DuPont—may prove even more reliable than some pure tech stocks.
~~~~~~~~~~~~~~~~~~~~~~~~
The above wonderful content comes from Chasewind Trading Desk.
For more detailed interpretations, including real-time commentary and frontline research, please join [Chasewind Trading Desk Annual Member]
Risk Warning and DisclaimerThe market has risks, be cautious with investments. This article does not constitute personal investment advice, nor does it consider the special investment goals, financial situation or needs of individual users. Users should consider whether any opinions, viewpoints, or conclusions in this article fit their specific circumstances. Investment according to this is at your own risk. ```