AI development necessitates substantial computing power and energy, contributing significantly to U.S. electricity consumption. While the focus on AI's water usage has begun, it primarily addresses data center cooling. However, this overlooks the much larger water footprint associated with electricity generation and chip manufacturing. These upstream processes can double or triple a data center's total water impact. Thermoelectric power plants use vast amounts of water for cooling, and producing advanced processors requires thousands of gallons of ultrapure water. Consequently, the visible water consumption of a data center is only a fraction of its true impact. By 2030, U.S. data centers could demand nearly as much additional water capacity daily as New York City uses. This highlights a broader issue of communities bearing the resource costs of AI growth. The physical infrastructure supporting AI, including water systems, is increasingly becoming a constraint. Water availability may dictate where data centers can be built, especially in drought-prone regions. Existing U.S. water infrastructure is often ill-equipped for the demands of hyperscale data centers, population growth, and climate change. Companies must begin treating water as a strategic resource, similar to energy, focusing on management, reuse, and optimization. Proactive water strategies, like closed-loop systems, will become crucial for long-term resilience and growth. Understanding these infrastructure realities, particularly water, is essential for the future of AI development.