Investor Kevin O’Leary argues AI data centers use significantly less water than golf courses in the U.S., a claim that holds true today but faces scrutiny as data center water consumption is set to rise sharply, particularly in drought-prone regions like Utah.
- Golf courses use 4.6 times more water than U.S. data centers today.
- Data center water usage expected to surpass golf by 2026 or 2027.
- Local opposition and regulatory actions intensify around large projects.
What happened
Kevin O’Leary compared the water usage of AI data centers to that of golf courses in the United States, highlighting that golf courses consume significantly more water daily. According to estimates, golf courses use about 2.08 billion gallons per day for irrigation, while data centers use roughly 449 million gallons daily for cooling purposes. O’Leary’s comments came amid controversy surrounding his 40,000-acre Stratos data center project in Utah, which sparked protests and prompted a gubernatorial executive order due to concerns over water demands near the environmentally sensitive Great Salt Lake.
In response to opposition, O’Leary scaled back the project by 75% to 10,000 acres and claimed the data center would utilize a closed-loop chilling system designed to minimize water usage. However, experts have indicated there is insufficient data to fully verify these claims. A second water rights application associated with the project was withdrawn, and O’Leary faced a defamation lawsuit after alleging that Chinese-funded protesters orchestrated opposition without evidence.
Why it matters
While it is currently true that golf courses use substantially more water than AI data centers, projections indicate that data center water consumption will increase rapidly as AI workloads expand. Estimates suggest that by 2026 or 2027, total water use by data centers could surpass that of golf courses, rising to an estimated 590 billion gallons by 2028 compared to a steady 425 billion gallons for golf courses. This trajectory raises significant concerns about local water resource management, especially in drought-affected regions like Utah.
The issue extends beyond mere water volume comparisons. Golf course irrigation is generally stabilizing or declining due to adopting drought-resistant grasses and recycled water, whereas data center water demand is accelerating exponentially to meet the cooling needs of energy-intensive AI operations. This situational complexity challenges policymakers, environmentalists, and communities to weigh the ecological trade-offs involved in approving large-scale data centers in sensitive watersheds.
What to watch next
Given the increasing water demand projected for AI data centers and their potential environmental impacts, regulatory bodies and local governments are expected to scrutinize new and existing projects more closely, particularly in water-stressed areas. Utah’s executive order introducing new development standards in response to the Stratos project exemplifies this emerging trend. Community resistance is likely to continue, with groups advocating for prioritizing ecological preservation over industrial water use, especially near fragile ecosystems like the Great Salt Lake.
Industry stakeholders and data center developers will need to provide transparent disclosures about water usage and implement verified water-saving technologies such as closed-loop cooling. Monitoring how advancements in technology, regulatory policies, and community responses evolve will be critical to understanding the balance between AI infrastructure growth and sustainable resource management.