Big Tech’s Hidden Water Crisis: How Silicon Valley’s Data Centers Are Draining the Desert While Preaching Climate Virtue

ByEduardo Bacci

November 7, 2023
Big Tech Water Crisis Data CentersBig Tech Water Crisis Data Centers — TIJ News Investigation. Photo: Wikimedia Commons

Municipal water utility records reveal that major tech companies are consuming billions of gallons of water annually to cool their data centers — in some of the most water-stressed regions of the American Southwest. An OSINT investigation into the gap between corporate climate messaging and on-the-ground resource consumption.

Image directive: Create a bar chart comparing annual water consumption: Meta Goodyear AZ facility (56M gallons), average US household (109,500 gallons), with projected 2028 industry totals. Search Unsplash for “data center servers” or “desert landscape Arizona.”

The Thirst Beneath the Cloud

Every time you upload a photo, run a search query, or ask an AI chatbot a question, water flows. Not metaphorically — physically. The servers that power the digital economy generate enormous amounts of heat, and the most common cooling method is evaporative cooling: pushing water through systems that absorb thermal energy and release it as vapor. In the arid Southwest, that water comes from already-stressed municipal supplies.

A Lawrence Berkeley National Laboratory report found that U.S. data centers consumed 17 billion gallons of water directly through cooling systems in 2023 — a figure projected to double or quadruple by 2028. When indirect water consumption from electricity generation is included, the total reaches 211 billion gallons.

The Arizona Numbers

Meta’s data center facility in Goodyear, Arizona consumes approximately 56 million gallons of potable water annually — roughly 153,000 gallons per day. This is in a state where the Colorado River compact is under severe stress, where residential water restrictions are increasingly common, and where every gallon directed to a server farm is a gallon not available for agriculture, households, or ecosystem maintenance.

Google’s 2023 environmental report disclosed that the company’s global data center operations consumed over 6 billion gallons to cool facilities worldwide. The company has been expanding its footprint in the Southwest, drawn by cheap land, abundant solar energy, and — critically — business-friendly water policies that don’t always require full public disclosure of consumption volumes.

In Nevada, a proposed data center facility projected consumption of 650,000 gallons of water daily — in a state where Las Vegas has been ripping out ornamental lawns to conserve municipal water supply.

The Transparency Gap

Arizona’s regulatory framework does not consistently require data center operators to publicly report water withdrawal amounts. This opacity allows tech companies to consume massive quantities of a scarce public resource without the kind of scrutiny that would apply to, say, a new housing development or agricultural operation requesting the same allocation.

Phoenix-area data centers now account for 7.4% of the state’s total power consumption — and with each megawatt of power comes associated water demand for cooling. As AI workloads drive explosive growth in computing demand, the water equation is getting worse, not better.

The Corporate Climate Paradox

The irony is acute. The same companies draining desert aquifers publish elaborate sustainability reports touting their commitment to water stewardship, carbon neutrality, and environmental responsibility. Google pledges to be “water positive” by 2030. Meta touts its renewable energy purchases. Microsoft publishes detailed environmental impact metrics.

But the local reality doesn’t match the global narrative. A data center in Goodyear, Arizona consuming 56 million gallons of water annually is not “water positive” from the perspective of the community whose water supply it draws from. The renewable energy powering the servers doesn’t reduce the evaporative water loss from cooling towers.

The AI Accelerant

The explosion of artificial intelligence workloads is accelerating the problem dramatically. AI training and inference require significantly more computing power — and therefore more cooling — than traditional cloud services. As every major tech company races to build AI infrastructure, data center construction is booming, and water demand is scaling with it.

The Lawrence Berkeley projections are stark: from 17 billion gallons in 2023 to potentially 68 billion gallons by 2028, just for direct cooling. In a region where water scarcity is already reshaping urban planning, agricultural policy, and interstate compacts, the tech industry’s water appetite represents a fundamental conflict between digital growth and physical sustainability.

The companies building these facilities have the engineering talent and financial resources to develop water-free cooling technologies. The question is whether they’ll invest in those alternatives before the communities hosting their data centers run dry.

Eduardo Bacci is an investigative journalist at The Investigative Journal. Data sources include Lawrence Berkeley National Laboratory reports, municipal water utility records, Google Environmental Report 2023, and Meta facility disclosures.

ByEduardo Bacci

Investigative journalist and founder of The Investigative Journal. Specializing in OSINT-driven reporting on corporate malfeasance, government accountability, and institutional corruption.