Imagine three giant new industries showing up at the same town's water department at the same moment. The first is the AI hyperscalers, who use 3–5 million gallons per day per large datacenter (mostly for evaporative cooling). The second is the new semiconductor fabs — TSMC Arizona, Intel Ohio, Samsung Texas, Rapidus Hokkaido — each of which needs roughly 10 million gallons of ultrapure water per day (the World Economic Forum compares this to "as much water as 33,000 US households use every day"). The third is the federal government, which has told every drinking-water utility in America to remove PFAS "forever chemicals" down to 4 parts per trillion by 2029.
Three trillion-dollar buyers, one creaky vendor list. The EPA's own assessments put US water-infrastructure needs at over $1.25 trillion over 20 years. The Middle East is massively scaling desalination. Cities like Phoenix and Frankfurt are starting to refuse new datacenter water permits. Microsoft has committed to zero-water cooling, with first pilots in 2026.
This is the most undervalued chokepoint thesis in the entire document. Water utilities trade like sleepy regulated companies, and the equipment-makers trade like cyclicals, but the demand curve is structural and exponential. Three names sit on near-monopoly positions: Energy Recovery (~90% of seawater desal pressure exchangers), Kurita (Japanese semiconductor ultrapure water), and American Water Works (largest US regulated water utility — a geographic monopoly).