Public fear about data centers draining local water supplies is largely misplaced. New facilities using closed-loop cooling technology have minimal water consumption. For example, the massive Stargate campus in Abilene is projected to use less water in a year than a McDonald's restaurant.

Public outrage over data center water usage is fueled by large, decontextualized numbers. In reality, total U.S. data center water use is a fraction of that used for golf courses, almond farming, or even water lost annually to leaky pipes, revealing a major perception vs. reality gap.

The narrative that AI data centers deplete water and raise electricity prices is largely false. They often use less water than a golf course and, by building their own power, can fund grid upgrades and sell excess energy back, lowering local electricity costs and boosting tax revenues.

Contrary to the negative public narrative, the newest generation of data centers are not just resource drains. Built by tech companies, not real estate firms, they are designed to be efficient, with some even contributing power back to the grid and using minimal water, while also preparing for future chip technologies.

Opponents frame data center energy consumption in alarming but misleading terms, like equating thermal output to '23 atom bombs' daily. This tactic ignores relative scale (e.g., Texas's energy use would be '230 atom bombs') and stifles productive conversation by replacing nuanced analysis with fear-mongering.

To counter concerns about water consumption, Meta proactively chose a more expensive, highly efficient cooling system for its Louisiana data center. This decision resulted in the facility using less water than the agricultural operations on the same land previously, neutralizing a common environmental criticism.

Data scientist Hannah Ritchie clarifies that while AI uses a small fraction of global electricity (0.5%), its high concentration in a few data center locations creates intense, localized strain on power grids. This geographical bottleneck, not overall consumption, is the primary energy challenge for AI's growth.

The energy demands of modern AI are difficult to contextualize. A one-gigawatt data center uses as much power as a city of nearly one million US households. A five-gigawatt facility requires a 5,000-acre building footprint, excluding any power infrastructure.

Contrary to public perception, modern liquid cooling does not waste water. It uses a sealed, closed-loop glycol system that rejects heat through giant external radiators, much like a car. A massive data center's water usage for this system is minimal, comparable to that of a single family home.