My boss is always telling me to think about demand and supply (cc'ing @santiagoroel). So I did that today, and thought about an interesting trade. I'm probably wrong, but I think it makes sense? THE WATER BOTTLENECK 🧊 TL;DR: The hyperscalers are building $200B+/year of AI infrastructure on top of a physically finite resource that's governed by laws written in 1902. The market cap ratio between water infrastructure and AI chips is 100:1. When voters realize data centers consume more water than their entire town, the political hammer falls. The trade: Long PHO/PHO/ PHO/AWK/$XYL. Catalyst: Colorado River negotiations conclude Dec-26 I. THE SETUP: Everyone's Betting on the Wrong Bottleneck. The consensus AI infrastructure trade has been painfully obvious: semiconductors (already ran 200%+), power grid buildout (already consensus), uranium (up 69% YTD). Everyone's solving for electricity.Nobody's solving for what the market is missing: you cannot train GPT-5 without water. Not metaphorically. Literally. The H100 clusters generating all this "intelligence" produce so much heat that evaporative cooling is the only economically viable solution at scale. And evaporative cooling has a dirty secret—the water evaporates. It's gone. Into the atmosphere. Forever. The market is pricing AI infrastructure as if we have infinite water. We don't. Western US water supply is capped by snowpack in the Rockies and aquifer recharge rates that haven't changed since the Pleistocene. Meanwhile, demand is growing parabolically. This is the trade. II. THE MATH MATHING: Direct + Indirect Consumption Most analysis stops at "data centers use X gallons for cooling." That's not the full picture. The actual formula is: Total Water Impact = Direct Cooling (On-site) + Indirect Generation (Off-site) The Direct (On-site): Modern AI facilities: 1.8-12 liters per kWh for evaporative cooling "Typical" hyperscale cluster: 15 million gallons/day (5.4B gallons/year) Water Equivalence: Same as 50,000 new homes The Indirect (Off-site) — This Is What Everyone Misses: The thermoelectric power plants (coal, gas, nuclear) generating the 24/7 baseload power for these facilities use ~10x more water than the cooling systems themselves for steam generation and condenser cooling. When you run the full stack: 2030 US AI Infrastructure Projection: Total water consumption: 1.1-1.7 TRILLION gallons annually Context: Equal to every household in California combined Let me repeat that. By 2030, AI infrastructure will consume the same amount of water as the entire state of California's residential use. Nvidia can print more chips. Utilities can build more generation capacity. But you cannot VC-fund more rainfall. The Colorado River flow is determined by Rocky Mountain snowpack, and last time I checked, there's no Series B for the hydrological cycle. III. THE PRINCIPAL/AGENT PROBLEM: Why This Gets Approved (For Now) The reason this trade exists is classic misaligned incentives. The people making decisions today won't be around to face the consequences tomorrow. The Agents (Signing the Deals): City Council/Mayor (Mesa, AZ): Incentive: Immediate tax revenue + "Tech Hub" branding for next election Horizon: 2-4 years until they're termed out Cost: They won't be in office in 10 years when the aquifer is depleted Tech Capex Managers: Incentive: Deploy 50,000 H100s by Q4, bonus tied to speed Calculation: Water is a rounding error ($2/1,000 gallons vs. $40k per GPU) Problem: Their KPI isn't "20-year water security" State Governors: Incentive: Steal jobs from California, get credit for "Business-Friendly Environment" Conflict: Must eventually answer to Agriculture Lobby, which owns the actual senior water rights The Principals (Bearing the Cost): Voters/Residents: See new Google facility that uses more water than their entire town Currently under Stage 3 restrictions (water lawn once/week) Will eventually ask: "Why am I limiting showers while ChatGPT gets unlimited water to hallucinate faster?" Agriculture: Holds senior water rights under Prior Appropriation Doctrine Currently being told to accept cuts to allocations Will not accept permanent junior position to Big Tech This incentive structure flips the moment voters connect the dots. And they will. IV. EVIDENCE IT'S ALREADY STARTING: Phase 2 is NOW We're not speculating about a future problem. The backlash is already happening: Blocked/Delayed Projects (Past 18 Months): $64 billion in data center projects blocked or delayed due to local opposition Tucson, AZ: City blocked Amazon "Project Blue" (would've used hundreds of millions of gallons annually), passed emergency ordinance requiring water conservation plans for any user >7.4M gallons/month Chandler, AZ: 422,000 sq ft data center blocked despite Kyrsten Sinema personally lobbying Virginia: Governor Youngkin vetoed water disclosure bill (industry lobbied hard) Santa Clara, CA: Planning Commission initially denied GI Partners' data center over water concerns Texas: Despite $1B/year in tax subsidies to data centers, passed grid regulation (SB 6) but ZERO water regulation Legislative Activity: Utah, Oregon, Georgia, Indiana, Illinois, New York: Bills introduced requiring water usage disclosure for data centers Industry response: Data Center Coalition strongly opposes any reporting requirements Arizona: "30 or more" data centers currently eyeing Pima County (per city officials) Grassroots Organizing: "No Desert Data Center" movement in Arizona "Save Bren Mar" in Virginia Over 500 residents packed Warrenton town hall to oppose Amazon facility Local opposition increasing despite massive tech lobbying budgets What does this tell us? We're in Phase 2 (Friction) of a three-phase cycle: Phase 1 (2020-2024): Honeymoon - data centers welcomed as "clean industry" Phase 2 (2024-2025): Friction - organized local opposition, first permit denials Phase 3 (2026+): Regulatory reckoning - mandatory rationing, impact fees, moratoriums V. THE CATALYST: December 2026 is The Cliff There's a specific date when the current regulatory framework expires: December 2026: Colorado River Drought Contingency Plan Expires The Colorado River supplies water to 40 million people across seven states. Current negotiations signal that non-human consumption (industrial/tech) will be first category to face mandatory curtailments during shortages. What This Means: If you're a data center pulling 5 million gallons/day from Colorado River basin water: No water = No cooling No cooling = Servers overheat Servers overheat = $5B facility becomes paperweight You cannot negotiate with thermodynamics. The Prior Appropriation Doctrine Trap: Western water law operates on "First in time, first in right." A cattle rancher from 1890 has senior rights over a $10B data center from 2025. Tech companies built their empires on "move fast and break things." But you can't move fast when you're subject to water law written in 1902. The legal framework treats water like property rights, not a utility. And tech companies don't own the rights. Farmers do. VI. THE TRADE: Water Infrastructure as AI Infrastructure If you believe AI continues to scale, you are implicitly long water infrastructure. There is no way around it. The Thesis: Water will transition from "utility expense" (priced at cost) to "strategic asset" (priced at scarcity value) as: Tech companies realize water is THE bottleneck Regulatory environment shifts from permissive to restrictive Companies are forced to acquire senior water rights and fund massive infrastructure The Vehicles: $AWK (American Water Works) $XYL (Xylem) $PHO (Invesco Water Resources ETF) DISCLAIMER: Not investment advice. Do your own due diligence. I'm positioned according to this thesis but could be completely wrong. Physics doesn't care about your P&L.