The intelligence is in recognizing that these facilities are the physical backbone of the digital economy and the current AI boom—not optional luxuries. Without large-scale, high-density data centers, the services people actually use and demand (search, cloud storage, streaming, real-time AI models, financial systems, logistics, scientific computing, etc.) either don’t exist or become far more expensive and less capable.
What the numbers actually show
• Power: Data-center electricity demand is rising sharply because AI training and inference are compute-hungry. Hyperscale facilities are being built because that is the efficient way to deliver the required FLOPs at scale. Distributing the same compute across millions of smaller machines would usually waste more energy on overhead, networking, and under-utilization.
• Water: Many centers still use evaporative cooling because it is thermodynamically efficient (lower energy use than pure air conditioning in most climates). A large facility can use hundreds of thousands to over a million gallons per day. U.S. data centers used roughly 17 billion gallons directly in 2023; projections show significant growth. Indirect water use at power plants is often larger still.
That consumption is real and locally painful in water-stressed areas.
It is also smaller than many people assume when put in context: agriculture, power generation, and even some recreational uses (golf courses, landscaping) dwarf it nationally. The industry knows this and is moving—closed-loop systems, direct-to-chip liquid cooling, immersion cooling, free-air cooling in cooler climates, higher operating temperatures, and recycled/non-potable water. Microsoft, Google, Amazon and others have already cut water intensity dramatically (Microsoft’s WUE dropped ~90% over two decades to ~0.27 L/kWh by 2025) and are targeting further reductions and water-positive operations.
There. Leave me alone.
