Wyoming AI Facility to Outstrip State's Home Power Use
A new artificial intelligence (AI) facility planned for Wyoming is projected to consume more electricity than all residential households in the state combined, highlighting a growing national trend where AI infrastructure is rapidly outpacing the capacity of existing power and water systems. This development, spearheaded by AI data center developer Crusoe and energy infrastructure provider Tallgrass, will be located near Cheyenne. The initial phase of the campus is set for 1.8 gigawatts (GW) of power use, with the potential to scale up to 10 GW. To put this in perspective, 1 GW can power approximately one million homes.
The immense power demands of AI data centers are becoming a significant strain on utility systems across the United States. Industry estimates suggest that data centers currently account for 4.4% of U.S. electricity consumption, a figure that could reach 12% by 2028. Some utilities are even warning of a potential 50% rise in national electricity demand within the next five years, a level of growth for which no state is currently prepared. The energy intensity of AI queries is substantially higher than traditional internet use; for instance, a typical ChatGPT request uses about 2.9 watt-hours of electricity, nearly ten times more than a standard web search. This escalating demand is already prompting utilities like Pacific Gas & Electric to reverse plans for retiring coal plants, while grid operators in Texas have issued emergency alerts.
Wyoming has emerged as an attractive location for such infrastructure due to its geographic stability, connectivity to the national internet backbone via Interstate 80, and its significant energy surplus. The state produces about 12 times more energy than it consumes, ranking as the third-biggest net energy supplier in the nation. This energy abundance has already drawn companies like Microsoft and Meta, with Meta’s 900-acre hyperscale data center near Cheyenne nearing completion. The new Crusoe and Tallgrass facility intends to leverage multiple energy sources, including natural gas and future renewable energy developments, and will operate with autonomous energy rather than relying solely on the state grid, due to the sheer scale of its projected consumption. At its full 10 GW capacity, the facility’s annual energy consumption could be 87.6 terawatt-hours, double the state’s current total energy output.
Beyond electricity, AI data centers also require substantial water resources, primarily for cooling systems. Many facilities utilize evaporative cooling, with estimates suggesting each large data center could use 2 million liters of water per day, equivalent to the daily consumption of 6,500 U.S. households. This poses a particular challenge for water-stressed Western states. For example, data centers in Central Texas consumed 463 million gallons of water in 2023 and 2024, and projections indicate that Texas data centers could consume nearly 400 billion gallons by 2030, representing about 7% of the state’s total projected water use. Major tech companies like Google, Microsoft, and Meta have used billions of gallons of water for their data centers and are facing scrutiny over their water footprints, particularly in drought-prone areas.
The rapid expansion of AI infrastructure carries significant economic and environmental implications. Meeting the necessary infrastructure requirements could cost up to $2 trillion by 2030, with an estimated $5.7 to $9.2 billion annually in public health-related costs due to increased emissions. These costs are anticipated to be passed on to consumers through higher utility bills. For example, customers in the PJM Interconnection region, which includes a high concentration of data centers, have seen energy bill increases of up to 20% this summer, with an independent monitor attributing three-quarters of these increases to data center demand.
Addressing these challenges requires a multifaceted approach focused on sustainability. Innovations in energy efficiency for data centers include advanced cooling technologies like liquid cooling, which can be significantly more efficient than traditional air cooling. AI itself can play a role in optimizing energy management within data centers through predictive analytics and intelligent workload distribution, potentially reducing power consumption and improving operational performance. Furthermore, integrating data centers with smart grids could allow them to store excess renewable energy and feed it back into the grid during peak demand, enhancing overall grid stability and sustainability. The development of low-power computing and the use of renewable energy sources are also crucial steps toward mitigating the environmental impact of AI.