This is a free preview of a paid episode. To hear more, visit bownereport.substack.com
I have added an original song by myself and my family at the bottom of the article. It seemed appropriate.
On July 23, 2025, President Trump signed three executive orders aimed at solidifying U.S. leadership in artificial intelligence, with one specifically targeting the acceleration of data center construction by streamlining environmental regulations and permitting processes. This move, part of a broader “AI Action Plan,” revokes the Biden era executive order that imposed stricter DEI and climate requirements on AI infrastructure development.
The new policy promotes the use of federal lands, including Brownfield and Superfund sites, for data center projects and leverages the FAST-41 framework to expedite permits for qualifying projects. Brownfield and Superfund sites are basically former industrial sites generally in urban areas that are complicated by the potential presence of hazardous substances, pollutants, or contaminants. The administration’s goal is to facilitate rapid infrastructure buildout to support AI’s computational demands, positioning the U.S. to outpace global competitors, particularly China, in technological innovation. Aligning with Trump’s previous "Stargate" initiative announcement on January 21, 2025, a $500 billion private sector partnership with companies like OpenAI, Oracle, and SoftBank to construct up to 20 AI data centers across the U.S.
Meanwhile, Meta has unveiled ambitious plans to build massive AI data centers, including the “Hyperion” project in Louisiana, projected to scale up to 5 gigawatts of computational power over several years, and the “Prometheus” cluster in Ohio, set to come online in 2026 with 1 gigawatt and 500,000 GPUs. Additionally, Meta is developing multiple “Titan clusters,” with each facility’s footprint rivaling significant portions of Manhattan. Meta plans to deploy between 2.75 and 3 million GPUs across its clusters. Enough energy to power millions of homes. Additionally, data centers use water for cooling servers and other IT equipment to prevent overheating. A hyperscale center can guzzle 3 to 5 million gallons per day. In 2021, data centers collectively consumed 449 million gallons per day.
The average total public water supply withdrawals in 2015 were approximately 39,000 million gallons per day, equivalent to 39 billion gallons per day or 142.35 trillion gallons annually. This represents about 14% of total U.S. water withdrawals, including irrigation, thermoelectric power, etc.
Of this, 61% came from rivers and lakes, and 39% came from groundwater sources, such as aquifers.
Public supply served 283 million people, or 87% of the U.S. population for household use, with additional deliveries for commercial, industrial, and public services. You know, firefighting, parks, water treatment plants and the like.
But as small as these numbers are today, the AI data center juggernaut is inevitably barreling headlong into a water starved future, and the battleground is smack dab in America’s driest regions. Out in the Southwest, Arizona’s a ticking time bomb, Phoenix and Mesa are drowning in data centers while the Colorado River chokes.
Google is already gulping millions of gallons a day, elbowing out farmers and folks just trying to keep their taps running. Nevada’s no better, its parched deserts are now hosting a data center gold rush while groundwater dwindles to nothing. California’s Central Valley and southern sprawl are getting hammered too, with new server farms piling onto a state already choked by drought and Newsom’s Big Ag water grab.
Then there’s Texas, where cheap power and loose rules have opened the floodgates for data centers. Abilene’s staring down the barrel of OpenAI’s Stargate behemoth, a project set to suck up water in a state where droughts are legendary and farmers are fighting for scraps. Over in Georgia, the humid air is irresistible to cooling hungry tech giants, places like Newton County are already paying the price wells running dry, water turning to sludge, and a looming deficit by 2030.
