As drought strains water supplies across much of the United States, Virginia Tech researchers have developed an artificial intelligence (AI) model designed to help policymakers manage growing competition between agriculture and semiconductor manufacturing.
Feras Batarseh , associate professor in the Department of Biological Systems Engineering and the Commonwealth Cyber Initiative and the project's lead researcher, said the recent explosion of AI queries via language learning models and the increased need for chip production is increasing pressure on already-strained water systems in many parts of the country.
To build the model, members of the A3 lab team, Ph.D. student Lauren Pincus and Dan Sobien , research associate, analyzed semiconductor facilities, irrigation patterns, and water stress indicators across all 50 states. The project is one of the first national‑scale studies to examine where agriculture and semiconductor manufacturing compete most directly for water resources.
New AI models built for a tangled water system
Unlike traditional predictive models, the team's causal AI model identifies cause‑and‑effect relationships between water availability, crop needs, semiconductor expansion, and regional basins. It can show, for example, how adding a fabrication plant in Arizona might affect irrigation capacity in neighboring states or how improving irrigation efficiency in the Midwest could free up water for industrial growth.
This approach, Batarseh said, is essential because water decisions in the U.S. rarely operate in isolation.
"You have state water, you have basins, and then you have economic regions and the geographic scope of federal jurisdiction, all of which impact each other," Batarseh said.
By integrating data from agriculture, hydrology, climate, and industrial operations, the model generates optimized recommendations for each state. It is designed to support decisions at multiple levels — from state water managers to federal agencies shaping national semiconductor manufacturing strategies.
Competition for the same water
Semiconductor fabrication requires enormous volumes of ultra‑purified water to clean and cool silicon wafers. Many facilities rely on municipal water systems, which ultimately draw from the same surface and groundwater sources that support farms and communities.
"While manufacturing at home is a national priority, most of the semiconductor manufacturing facilities are placed in states like Arizona, California, and Texas, which are very dry," Batarseh said.
Agriculture, meanwhile, remains the nation's largest water user — consuming about 70 percent of freshwater withdrawals. Crops such as corn, cotton, rice, and soybeans require substantial irrigation, especially in drought‑prone regions.
"While technologies such as drip and smart irrigation could be used to minimize water waste in agriculture, the water need of a crop still varies based on multiple factors and is still generally high," Batarseh said.
In shared basins like the Colorado River, increased semiconductor activity could directly reduce irrigation capacity. Conversely, improvements in irrigation efficiency could create room for industrial growth without increasing water stress.
AI as both a stressor and a solution
Beyond drought and industrial expansion, U.S. water systems face additional pressures, including aging infrastructure and cybersecurity vulnerabilities that can disrupt water treatment and distribution. Batarseh said AI also offers powerful tools to improve water efficiency — especially in agriculture.
"If smart irrigation techniques are applied and you optimize by 10 or 20 percent, then that gives room for other industries like semiconductor manufacturing," he said.
The causal AI model is designed to support exactly that kind of optimization. It can simulate scenarios, test policy options, and identify strategies that reduce water stress without sacrificing economic growth.
As pressure on U.S. water systems grows, Batarseh said, "the goal is not choosing between farms and fabrication plants but finding smarter ways to support both."
He said tools like causal AI can help policymakers make more informed decisions before water shortages become crises. By showing how choices in one region ripple across farms, factories, and communities nationwide, the model offers a clearer path toward balancing economic growth with long-term water sustainability.
