WATER WATER
THE OVERLOOKED CONSTRAINT
IN THE AI DATA CENTER BOOM
Water decisions shape data center site selection, permitting,
community trust, and long-term operating costs and should be
addressed early in project planning.
As gigawatt-scale projects surge
across the US, “doing water right” can
determine whether communities see
opportunity—or risk.
Scrolling through LinkedIn, it feels
like a new gigawatt-scale data center
project is announced somewhere in
rural America every week. Texas, Ohio,
Oregon, Virginia, Oklahoma—the list
goes on and on. Billions of dollars are
being invested in power, semiconductor
manufacturing, and real estate, all in the
name of artificial intelligence.
Between 2025 and 2030, capital
spending to enable AI infrastructure,
including data centers, semiconductor
manufacturing, power plants, water
infrastructure, and their supply chains
is expected to reach roughly $5 trillion,
or around 3 percent of gross domestic
product. The last time the country
invested in infrastructure at that scale
was during the railroad boom of the
1800s. Then, railroads reshaped
commerce, accelerated industrialization,
and permanently altered how the nation
functioned. Today’s AI-driven buildout
may not look like steel rails stretching
west—but its economic and societal
impact could be just as transformative.
A Surge of Investment–and
Rising Community Concern
In the communities where data centers
are proposed, enthusiasm is often mixed
with skepticism. In 2025, an estimated
$65 billion in data center investments
were delayed or abandoned due to local
opposition. Among the most common
concerns are water use, water and air
quality, and power demand, including
potential impacts on local infrastructure,
user rates, and the environment.
Because water is integral to both cooling
systems and power generation, these
concerns are often interconnected.
For some rural communities, the
skepticism is deeply rooted. Many
have lived downstream of heavy
industry for decades and are wary of
new development that could stress
already constrained infrastructure or
introduce additional environmental
risk. Others question whether local
ratepayers should subsidize water,
wastewater, and power system
upgrades that primarily serve private
facilities. In response, some states
and municipalities are considering
moratoriums or outright bans on new
data center construction.
These concerns cannot be dismissed.
They are rational, experience-based,
and increasingly influential in
project outcomes.
JOHN RYDZEWSKI, PE (jrydzewski@carollo.com)
JACK MALACE, PE
CURRENTS / PROJECT UPDATE
8
Cooling Technology is
Reshaping Water Demand
As computing density increases,
equipment becomes more complex,
and facilities expand in square
footage, traditional air cooling is
no longer sufficient or economical.
Older data centers commonly relied
on water-intensive evaporative cooling
towers. Newer facilities increasingly use
closed-loop fluid coolers, which function
much like automobile radiators, moving
large volumes of air over large surface
areas to dissipate heat.
The Three-Legged Stool:
Water, Power, and Land
Data centers ultimately rely on
three foundational resources:
water, power, and land. Public
discourse tends to focus on
land acquisition and grid
capacity. Water, by contrast,
is often treated as a secondary
issue until late in the project
lifecycle, when it becomes
an expensive and
highly-visible constraint.
WATER WATER