In March 2026, a detailed assessment published by researchers affiliated with the U.S. Geological Survey delivered a clear national picture of local water stress. Drawing on water years 2010-2020, the study estimated that 26.7 million Americans about 8 percent of the contiguous U.S. population live in areas experiencing chronic high or severe water limitation. The highest stress concentrations appear across the central and Southwestern United States, where low renewable supply collides with elevated consumptive use, particularly for irrigation.

The findings, led by Edward G. Stets and colleagues, underscore a fundamental geographic reality: while water supply far exceeds demand when averaged across the entire contiguous United States, local and regional imbalances are already significant. “Although ws greatly exceeds wc at the CONUS scale, water is limited locally or regionally due to spatial and temporal patterns in climate and wc,” the authors wrote. The central and Southwestern regions recorded the highest values on their supply-and-use index, driven by the combination of arid conditions and intensive agricultural withdrawals.

These patterns arrive at a moment when American industrial policy is deliberately concentrating large, water-intensive investments in precisely the same geographies. Semiconductor fabrication plants, hyperscale data centers, and advanced manufacturing facilities require reliable volumes of high-quality water for process rinsing, ultrapure water production, and cooling. The question now confronting corporate planners and public officials is whether locally reliable water rather than electricity prices, available land, or tax incentives will increasingly dictate where major projects can actually be built and sustained.

The Geography of Scarcity Meets the Geography of Growth

The Stets assessment maps chronic limitation most heavily across the Southern High Plains, Central High Plains, Texas, the Mississippi Embayment, and the Southwest Desert. These regions already support substantial agricultural economies that depend on groundwater and surface flows that are under pressure. At the same time, the same corridors have become magnets for the next wave of industrial expansion.

Arizona offers the clearest illustration. The state has long hosted semiconductor operations and has attracted major new commitments under the CHIPS and Science Act. Intel’s facilities in the Chandler area and TSMC’s expanding Phoenix campus represent tens of billions of dollars in capital investment and thousands of high-wage jobs. Yet each advanced fab can require millions of gallons of water daily. TSMC’s first Arizona fab has been reported to use roughly 4.75 million gallons per day; when multiple fabs reach full operation, total site demand rises substantially even after aggressive recycling. Intel’s Arizona operations have similarly drawn several billion gallons annually in recent reporting years.

Both companies have invested heavily in on-site reclamation. Intel operates advanced treatment systems, including a long-standing public-private partnership for brine reduction, and has pursued net-positive water goals through conservation and watershed restoration projects. TSMC has broken ground on a large industrial reclamation plant intended to push recycling rates toward 90 percent. These engineering responses reduce net consumption, but they do not eliminate dependence on a constrained regional water portfolio that includes Colorado River deliveries, local surface supplies, and limited groundwater.

Colorado River allocations themselves face tightening constraints. Federal decisions in recent years have already reduced deliveries to lower-basin states, with Arizona among the most exposed. In a system where agricultural, municipal, and industrial users compete for the same finite resource, each new large industrial load intensifies the zero-sum character of local water budgets.

Data Centers Amplify the Pressure

Parallel growth in data centers adds another layer of demand. Cooling systems especially evaporative designs preferred in hotter climates can withdraw hundreds of thousands to several million gallons per day at individual hyperscale facilities. Multiple analyses have shown that a substantial share of new U.S. data center capacity since 2022 has been sited in high or extremely high water-stress areas, including portions of Arizona, Texas, Nevada, and New Mexico. Indirect water use embedded in electricity generation further magnifies the footprint, particularly where grids still rely on thermoelectric plants.

In the Phoenix metropolitan area, projections of data-center cooling demand have pointed to multi-fold increases over the coming decade. Similar concerns have surfaced around clusters in Texas and parts of the intermountain West. While data centers typically represent a smaller share of total regional water use than agriculture or municipal supply, their rapid growth and high peak demands during hot periods can strain local systems already operating near capacity.

Site-selection practices are beginning to reflect this reality. Industry observers and consultants report that water availability, long-term supply security, and permitting risk are moving from secondary diligence items to first-tier decision criteria alongside power capacity and labor markets. In some jurisdictions, new groundwater certificates for large industrial users have become difficult or impossible to obtain. Elsewhere, utilities and regulators are imposing conditions that require non-potable or reclaimed sources, or that limit net additional withdrawals.

Incentives Versus Hydrologic Reality

Federal and state incentives under the CHIPS Act and related programs have successfully steered semiconductor investment toward Arizona, Texas, Ohio, New York, and Idaho. These locations offer different combinations of existing industrial ecosystems, workforce potential, and policy support. Ohio and parts of the Northeast and Midwest generally enjoy more abundant surface-water resources and less chronic supply-use imbalance than the Southwest. That hydrologic advantage is increasingly visible in corporate risk assessments.

Yet the pull of established clusters remains strong. Arizona’s dry climate reduces certain humidity-related process challenges in chip fabrication, and the state has developed specialized infrastructure and permitting pathways. The result is a tension between path-dependent industrial geography and the physical limits documented by the 2026 assessment. Companies can mitigate through recycling, wastewater reuse, and off-site restoration credits, but these strategies carry capital costs and still leave residual dependence on regional water systems that are already stressed.

Advanced manufacturing beyond semiconductors battery plants, chemical processing, and certain food and beverage facilities faces analogous constraints. Water quality requirements and continuous process needs make intermittent or drought-vulnerable supplies operationally risky. In regions where the Stets study identified chronic high or severe limitation, the margin for new large industrial loads is correspondingly thinner.

Implications for Economic Geography

The emerging pattern suggests that water reliability is shifting from a background assumption to an active determinant of industrial location. Regions with surplus renewable supply and lower competition among users may capture a larger share of future water-intensive investment, even if they offer less aggressive tax packages or less mature supplier ecosystems. Conversely, areas of chronic limitation may continue to attract projects that can demonstrate near-closed-loop water systems or that secure senior water rights and long-term contracts, but the bar for approval and community acceptance is rising.

Public policy is adapting unevenly. Some Western states have strengthened long-term water planning, invested in infrastructure, and required industrial users to demonstrate sustainable sourcing. Others continue to treat water as secondary to job creation and tax-base expansion. The Stets assessment provides a quantitative baseline against which these choices can be evaluated: when 26.7 million people already live under chronic high or severe limitation, additional large industrial demands cannot be absorbed without trade-offs.

Corporate behavior is also evolving. Leading semiconductor and data-center operators now routinely commission watershed-scale modeling, climate-scenario analysis, and multi-decade supply forecasts before finalizing sites. Some have publicly committed to net-positive water goals or to sourcing a majority of process water from reclaimed sources. These practices reduce local impact but do not erase the underlying geographic mismatch between where water is chronically limited and where industrial demand is concentrating.

The 2026 USGS-linked research does not claim that the United States faces a national water shortage. Aggregate supply remains large relative to aggregate demand. The critical insight is spatial: local reliability is already constrained for millions of residents and for the ecosystems that depend on the same flows. As semiconductor manufacturing, data-center capacity, and advanced industry continue to expand under explicit national-security and economic-growth mandates, the availability of locally reliable water is moving closer to the center of the industrial location decision. In that emerging calculus, hydrology is no longer merely an environmental constraint it is becoming a first-order economic one.